# Thoughtworks Insights

A global technology consultancy that integrates strategy, design and engineering to drive digital innovation

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## Why agentic AI pilots don't scale

DevFeed: [Why agentic AI pilots don't scale](<https://devfeed.tech/articles/why-agentic-ai-pilots-don-t-scale-139733.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/machine-learning-and-ai/why-agentic-ai-pilots-dont-scale>)

Author: Published November

Published: 2026-11-17T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [Agentic AI](<https://devfeed.tech/topics/what-is-agentic-ai.md>), [AI Integration Strategies](<https://devfeed.tech/topics/ai-integration-strategies.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [ai-experiments](<https://devfeed.tech/topics/ai-experiments.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [elt](<https://devfeed.tech/topics/elt.md>)

Tags: [abap](<https://devfeed.tech/tags/abap.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-ai](<https://devfeed.tech/tags/agentic-ai.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [api](<https://devfeed.tech/tags/api.md>), [apis](<https://devfeed.tech/tags/apis.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [article](<https://devfeed.tech/tags/article.md>), [automatically](<https://devfeed.tech/tags/automatically.md>), [autonomous](<https://devfeed.tech/tags/autonomous.md>), [blog](<https://devfeed.tech/tags/blog.md>), [bottleneck](<https://devfeed.tech/tags/bottleneck.md>), [business](<https://devfeed.tech/tags/business.md>), [business-logic](<https://devfeed.tech/tags/business-logic.md>), [capabilities](<https://devfeed.tech/tags/capabilities.md>), [catalog](<https://devfeed.tech/tags/catalog.md>)

### AI overview

The article argues that agentic AI pilots stall chiefly when legacy systems and unprepared data prevent enterprise integration and scale. It proposes a three-stage path: use AI-first reverse engineering to document legacy code, design an evolutionary target architecture, and replace monolith functions gradually with the strangler fig pattern; automate governed data product creation and migrate legacy data platforms; then let agents use decoupled, API-enabled capabilities and curated data through MCP. It claims simple data products can be built in hours and complex ones in up to two weeks, while its migration accelerator can be up to 60 times faster and cut end-to-end lead time by up to 90%; these are vendor-stated results, with a Unieuro example. The text presents modernization as risk mitigation and a growth enabler, while describing governance and operating-model fragmentation as additional scaling challenges.

### Source excerpt

Why agentic AI pilots don't scale

## Attackers are targeting the AI tools that produce our software

DevFeed: [Attackers are targeting the AI tools that produce our software](<https://devfeed.tech/articles/attackers-are-targeting-the-ai-tools-that-produce-our-software-133611.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/generative-ai/attackers-are-targeting-the-ai-tools-that-produce-our-software>)

Author: Jim Gumbley, Gautam Koul

Published: 2026-11-13T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [malware protection](<https://devfeed.tech/topics/malware-protection.md>), [slsa levels](<https://devfeed.tech/topics/slsa-levels.md>), [AI Agent](<https://devfeed.tech/topics/ai-agent.md>), [Cybercrime](<https://devfeed.tech/topics/cybercrime.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [observability pipelines](<https://devfeed.tech/topics/observability-pipelines.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-coding-tools](<https://devfeed.tech/tags/ai-coding-tools.md>), [ai-tools](<https://devfeed.tech/tags/ai-tools.md>), [backdoor](<https://devfeed.tech/tags/backdoor.md>), [blast-radius](<https://devfeed.tech/tags/blast-radius.md>), [breach](<https://devfeed.tech/tags/breach.md>), [build](<https://devfeed.tech/tags/build.md>), [ci-cd](<https://devfeed.tech/tags/ci-cd.md>), [claude-code](<https://devfeed.tech/tags/claude-code.md>), [cloud](<https://devfeed.tech/tags/cloud.md>), [code](<https://devfeed.tech/tags/code.md>), [coding](<https://devfeed.tech/tags/coding.md>), [coding-assistants](<https://devfeed.tech/tags/coding-assistants.md>), [coding-tools](<https://devfeed.tech/tags/coding-tools.md>), [commands](<https://devfeed.tech/tags/commands.md>), [compromise](<https://devfeed.tech/tags/compromise.md>), [confidence](<https://devfeed.tech/tags/confidence.md>), [credentials](<https://devfeed.tech/tags/credentials.md>), [dependencies](<https://devfeed.tech/tags/dependencies.md>), [software](<https://devfeed.tech/tags/software.md>)

### AI overview

AI coding tools and extensions sit close to source code, credentials, and build or deployment systems, so compromising one can expose broader environments. The article cites malicious Claude Code extension versions available through Cursor's marketplace for about 11 hours on October 29, 2026, and TeamPCP attacks on Trivy and LiteLLM as examples. It recommends six measures: assess dependencies using evidence and ongoing checks; make delivery teams responsible for adopted tools; control and verify updates; limit privileges and isolate tools with sandboxes and scoped credentials; prepare incident response, including workstation wipes and token revocation; and monitor developer machines while correlating endpoint, identity, repository, and cloud activity. The article presents these as risk-reduction steps, not a guarantee against compromise.

### Source excerpt

Attackers are targeting the AI tools that produce our software

## We built an AI control plane to reduce LLM expenses and prevent data leaks

DevFeed: [We built an AI control plane to reduce LLM expenses and prevent data leaks](<https://devfeed.tech/articles/we-built-an-ai-control-plane-to-reduce-llm-expenses-and-prevent-data-leaks-129864.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/generative-ai/we-built-ai-control-plane-reduce-LLM-expenses-prevent-data-leaks>)

Author: Rahit Kar, Camila Oliveira, Anshul Kumar

Published: 2026-11-10T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [control-plane](<https://devfeed.tech/topics/control-plane.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [LLM security](<https://devfeed.tech/topics/llm-security.md>), [LLM observability](<https://devfeed.tech/topics/llm-observability.md>), [LLM optimization](<https://devfeed.tech/topics/llm-optimization.md>), [Self Hosted AI Tools](<https://devfeed.tech/topics/self-hosted-ai-tools.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-governance](<https://devfeed.tech/tags/ai-governance.md>), [cache](<https://devfeed.tech/tags/cache.md>), [centralized-routing](<https://devfeed.tech/tags/centralized-routing.md>), [control-plane](<https://devfeed.tech/tags/control-plane.md>), [cost](<https://devfeed.tech/tags/cost.md>), [data-leaks](<https://devfeed.tech/tags/data-leaks.md>), [data-privacy](<https://devfeed.tech/tags/data-privacy.md>), [llm](<https://devfeed.tech/tags/llm.md>), [reduce](<https://devfeed.tech/tags/reduce.md>)

### AI overview

Heimdall is a proxy control plane that routes AI requests through cost and data-protection controls before sending them to LLM providers. Its OpenAI-compatible gateway requires clients to change a base URL; a fail-closed scrubber combines Microsoft Presidio, spaCy NER, and custom patterns to redact PII and credentials. A semantic cache can answer equivalent prompts without a provider call, while complexity routing selects among local Ollama and cloud models. Team, model, and period budgets trigger graceful downgrades from Opus to cheaper models rather than stopping requests. OpenTelemetry provides cost, latency, cache, routing, and filtering telemetry. The text reports about 14 ms of processing overhead, with complexity scoring and scrubbing run in parallel. It describes a four-week pilot with six Thoughtworkers but provides no quantified savings or measured data-leak outcomes.

### Source excerpt

We built an AI control plane to reduce LLM expenses and prevent data leaks

## Closing the context gap is what banking AI needs

DevFeed: [Closing the context gap is what banking AI needs](<https://devfeed.tech/articles/closing-the-context-gap-is-what-banking-ai-needs-121858.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/data-strategy/closing-the-context-gap-is-what-banking-ai-needs>)

Author: Rav Hayer

Published: 2026-11-06T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [context](<https://devfeed.tech/topics/context.md>), [AI-Ready Data](<https://devfeed.tech/topics/ai-ready-data.md>), [Payment Fraud](<https://devfeed.tech/topics/payment-fraud.md>), [Finance](<https://devfeed.tech/topics/finance.md>), [Frontier AI](<https://devfeed.tech/topics/frontier-ai.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [aws](<https://devfeed.tech/tags/aws.md>), [bank](<https://devfeed.tech/tags/bank.md>), [banking](<https://devfeed.tech/tags/banking.md>), [blueprint](<https://devfeed.tech/tags/blueprint.md>), [budget](<https://devfeed.tech/tags/budget.md>), [context](<https://devfeed.tech/tags/context.md>), [credit-card](<https://devfeed.tech/tags/credit-card.md>), [credit-risk](<https://devfeed.tech/tags/credit-risk.md>), [credit-scoring](<https://devfeed.tech/tags/credit-scoring.md>), [customer](<https://devfeed.tech/tags/customer.md>), [customer-experience](<https://devfeed.tech/tags/customer-experience.md>), [cybersecurity](<https://devfeed.tech/tags/cybersecurity.md>), [data](<https://devfeed.tech/tags/data.md>), [databricks](<https://devfeed.tech/tags/databricks.md>), [demo](<https://devfeed.tech/tags/demo.md>), [deployment](<https://devfeed.tech/tags/deployment.md>), [developer](<https://devfeed.tech/tags/developer.md>), [digital](<https://devfeed.tech/tags/digital.md>)

### AI overview

The article attributes the gap between banking AI pilots and production to missing transaction context, not primarily to model quality. General-purpose language models lack native understanding of transactions, while transaction foundation models are trained on transaction events; it cites examples and reported results from Mastercard, Revolut and Stripe. Rebuilding customer and transaction context separately for each pilot adds cost and inconsistency, so the article describes a shared intelligence layer trained on an institution's history for fraud, credit risk, payments and customer experience. It also reports a hybrid strategy: frontier models for complex reasoning and smaller, fine-tuned models for narrow tasks near sensitive data. This requires governance from the outset, including records of prompts and answers, access limits and reconstructable decisions. A financial loss can make the case for moving a use case from proof of concept to deployment.

### Source excerpt

Closing the context gap is what banking AI needs

## Why a language model misses fraud in your transaction data

DevFeed: [Why a language model misses fraud in your transaction data](<https://devfeed.tech/articles/why-a-language-model-misses-fraud-in-your-transaction-data-121859.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/why-a-language-model-misses-fraud-in-your-transaction-data>)

Author: Nathan Hilt

Published: 2026-11-06T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [Payment Fraud](<https://devfeed.tech/topics/payment-fraud.md>), [online-fraud-prevention](<https://devfeed.tech/topics/online-fraud-prevention.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>), [AI Integration Strategies](<https://devfeed.tech/topics/ai-integration-strategies.md>), [Advanced AI workflows](<https://devfeed.tech/topics/advanced-ai-workflows.md>), [audit trail](<https://devfeed.tech/topics/audit-trail.md>)

Tags: [2026](<https://devfeed.tech/tags/2026.md>), [adyen](<https://devfeed.tech/tags/adyen.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-model](<https://devfeed.tech/tags/ai-model.md>), [audit-trail](<https://devfeed.tech/tags/audit-trail.md>), [aws](<https://devfeed.tech/tags/aws.md>), [clusters](<https://devfeed.tech/tags/clusters.md>), [context](<https://devfeed.tech/tags/context.md>), [context-layer](<https://devfeed.tech/tags/context-layer.md>), [credit-card](<https://devfeed.tech/tags/credit-card.md>), [customer](<https://devfeed.tech/tags/customer.md>), [data](<https://devfeed.tech/tags/data.md>), [financial](<https://devfeed.tech/tags/financial.md>), [financial-services](<https://devfeed.tech/tags/financial-services.md>), [fine-tuning](<https://devfeed.tech/tags/fine-tuning.md>), [fraud](<https://devfeed.tech/tags/fraud.md>), [language](<https://devfeed.tech/tags/language.md>), [language-model](<https://devfeed.tech/tags/language-model.md>), [model](<https://devfeed.tech/tags/model.md>), [transaction](<https://devfeed.tech/tags/transaction.md>)

### AI overview

Transaction foundation models (TFMs) and graph neural networks (GNNs) address different fraud patterns: TFMs learn behavioral and temporal patterns in one entity's transaction history, while GNNs use relationships among customers, devices, merchants and cards to detect coordinated activity. The text says they can complement each other in a shared context layer. It describes structuring data with consistent semantic definitions and says AWS typically recommends 25 months of historical data to get started. A cited demonstration using Claude produced vague flags on raw transactions and a structured risk breakdown when grounded in transaction context. For production, the article recommends building an audit trail, access limits and decision reconstruction into the pipeline. Market examples include reported results from Revolut and Adyen; these are presented as company-reported proof points.

### Source excerpt

Why a language model misses fraud in your transaction data

## The commoditization of code: Why empathy is the new core engineering skill

DevFeed: [The commoditization of code: Why empathy is the new core engineering skill](<https://devfeed.tech/articles/the-commoditization-of-code-why-empathy-is-the-new-core-engineering-skill-119841.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/technology-strategy/the-commoditization-of-code-why-empathy-is-the-new-core-engineering-skill>)

Author: Iti Tyagi

Published: 2026-11-05T00:00:00Z

Content type: opinion

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [Software Engineering](<https://devfeed.tech/topics/software-engineering.md>), [engineering-leadership](<https://devfeed.tech/topics/engineering-leadership.md>), [engineering-culture](<https://devfeed.tech/topics/engineering-culture.md>), [Code review](<https://devfeed.tech/topics/code-review.md>), [Code Health](<https://devfeed.tech/topics/code-health.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>)

Tags: [age-of-ai](<https://devfeed.tech/tags/age-of-ai.md>), [ai](<https://devfeed.tech/tags/ai.md>), [algorithms](<https://devfeed.tech/tags/algorithms.md>), [api](<https://devfeed.tech/tags/api.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [binary-search](<https://devfeed.tech/tags/binary-search.md>), [binary-tree](<https://devfeed.tech/tags/binary-tree.md>), [build](<https://devfeed.tech/tags/build.md>), [code](<https://devfeed.tech/tags/code.md>), [code-generation](<https://devfeed.tech/tags/code-generation.md>), [code-review](<https://devfeed.tech/tags/code-review.md>), [code-reviews](<https://devfeed.tech/tags/code-reviews.md>), [collaboration](<https://devfeed.tech/tags/collaboration.md>), [communication](<https://devfeed.tech/tags/communication.md>), [compliance](<https://devfeed.tech/tags/compliance.md>), [conway-s-law](<https://devfeed.tech/tags/conway-s-law.md>), [craftsmanship](<https://devfeed.tech/tags/craftsmanship.md>), [empathy](<https://devfeed.tech/tags/empathy.md>), [engineering](<https://devfeed.tech/tags/engineering.md>)

### AI overview

As AI makes code generation faster, the effort shifts toward understanding requirements, reviewing and integrating changes, and maintaining shared understanding. Generated code still needs human judgment to assess system constraints and consequences; without clear intent and coordination, faster output can create downstream review work, conflicting changes, and code that no one fully understands. The article frames empathy as an engineering practice: write maintainable, observable code; give constructive reviews that preserve psychological safety; and work with security and compliance teams to build requirements into delivery pipelines. It also recommends recognizing mentoring, documentation, reviews, and team outcomes in performance evaluations, and assessing collaboration and communication in hiring and promotion. Technical expertise remains necessary; the argument is to value the work that helps teams deliver reliable software alongside individual code output.

### Source excerpt

The commoditization of code: Why empathy is the new core engineering skill

## Context rot in AI agents: How to recognize and fix it

DevFeed: [Context rot in AI agents: How to recognize and fix it](<https://devfeed.tech/articles/context-rot-in-ai-agents-how-to-recognize-and-fix-it-119840.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/generative-ai/what-is-context-rot>)

Author: Piyush Sonawane

Published: 2026-11-05T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [AI Context Management](<https://devfeed.tech/topics/ai-context-management.md>), [Agent Harness](<https://devfeed.tech/topics/agent-harness.md>), [AI Models](<https://devfeed.tech/topics/ai-models.md>), [context](<https://devfeed.tech/topics/context.md>), [agentic-coding](<https://devfeed.tech/topics/agentic-coding.md>), [AI-assisted coding](<https://devfeed.tech/topics/ai-assisted-coding.md>), [Retrieval Augmented Generation (RAG)](<https://devfeed.tech/topics/retrieval-augmented-generation-rag.md>)

Tags: [agent-harness](<https://devfeed.tech/tags/agent-harness.md>), [agents](<https://devfeed.tech/tags/agents.md>), [agents-md](<https://devfeed.tech/tags/agents-md.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-coding](<https://devfeed.tech/tags/ai-coding.md>), [ai-coding-agent](<https://devfeed.tech/tags/ai-coding-agent.md>), [ai-coding-tools](<https://devfeed.tech/tags/ai-coding-tools.md>), [ai-model](<https://devfeed.tech/tags/ai-model.md>), [anthropic](<https://devfeed.tech/tags/anthropic.md>), [code](<https://devfeed.tech/tags/code.md>), [coding](<https://devfeed.tech/tags/coding.md>), [coding-agent](<https://devfeed.tech/tags/coding-agent.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [coding-tools](<https://devfeed.tech/tags/coding-tools.md>), [compaction](<https://devfeed.tech/tags/compaction.md>), [context](<https://devfeed.tech/tags/context.md>), [context-rot](<https://devfeed.tech/tags/context-rot.md>), [how-to](<https://devfeed.tech/tags/how-to.md>)

### AI overview

Context rot is an agent's declining ability to use information effectively as its working context grows, gets noisy, or becomes outdated; a larger context window alone does not prevent it. Signs include forgotten constraints, repeated failed approaches, outdated plans, distraction, and inconsistent decisions, especially as tasks lengthen. The article recommends filtering tool outputs while retaining enough detail to investigate, summarizing goals, constraints, findings, failures, changes, and next steps, retrieving files or documents on demand, keeping stable project instructions separate from temporary task state, and starting a fresh context with a careful handoff when needed. Each approach can lose useful evidence if applied poorly. An agent harness can coordinate retrieval, output filtering, task-state updates, compaction, and context refresh around model calls, with observability and evaluation to assess results. Context rot concerns effective use of context; context poisoning concerns misleading or harmful information entering it, so reducing clutter alone does not address poisoning.

### Source excerpt

Context rot in AI agents: How to recognize and fix it

## Macro trends in the tech industry | November 2026

DevFeed: [Macro trends in the tech industry | November 2026](<https://devfeed.tech/articles/macro-trends-in-the-tech-industry-november-2026-116275.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/technology-strategy/macro-trends-tech-industry-november-2026>)

Author: Chris Ford

Published: 2026-11-04T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [ai-governance](<https://devfeed.tech/topics/ai-governance.md>), [AI Context Management](<https://devfeed.tech/topics/ai-context-management.md>), [AI software supply chain security](<https://devfeed.tech/topics/ai-software-supply-chain-security.md>), [ai observability](<https://devfeed.tech/topics/ai-observability.md>), [Developer experience](<https://devfeed.tech/topics/developer-experience.md>), [observability](<https://devfeed.tech/topics/observability.md>), [context window](<https://devfeed.tech/topics/context-window.md>), [Tool](<https://devfeed.tech/topics/tool.md>)

Tags: [adoption](<https://devfeed.tech/tags/adoption.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agent-context](<https://devfeed.tech/tags/agent-context.md>), [agent-sandboxing](<https://devfeed.tech/tags/agent-sandboxing.md>), [agent-security](<https://devfeed.tech/tags/agent-security.md>), [agentic-ai](<https://devfeed.tech/tags/agentic-ai.md>), [agentic-engineering](<https://devfeed.tech/tags/agentic-engineering.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-adoption](<https://devfeed.tech/tags/ai-adoption.md>), [antipattern](<https://devfeed.tech/tags/antipattern.md>), [apple](<https://devfeed.tech/tags/apple.md>), [atlas](<https://devfeed.tech/tags/atlas.md>), [bill](<https://devfeed.tech/tags/bill.md>), [blast-radius](<https://devfeed.tech/tags/blast-radius.md>), [industry](<https://devfeed.tech/tags/industry.md>), [november-2026](<https://devfeed.tech/tags/november-2026.md>), [tech](<https://devfeed.tech/tags/tech.md>), [tech-industry](<https://devfeed.tech/tags/tech-industry.md>), [trends](<https://devfeed.tech/tags/trends.md>)

### AI overview

At organizational scale, agentic engineering depends on shared context, cost controls, review capacity, security and an owned platform harness. The article recommends standardizing task definitions and evaluations, reviewing intent at the spec stage, and structuring context so agents receive only what they need. It distinguishes model routing and policy layers such as LiteLLM (Trial) and agentgateway (Assess), and advises tracking cost per completed task alongside quality. To manage review load, it points to session coordination and review tools while warning against throughput metrics and nominal human oversight. It calls for sandboxing, least privilege, supply-chain checks and threat modeling agents and their tools. Finally, it proposes treating the harness as a platform product with an owner, golden path, SLO and support model. These concerns grow with adoption and need coordinated ownership.

### Source excerpt

Macro trends in the tech industry | November 2026

## The harness is a product

DevFeed: [The harness is a product](<https://devfeed.tech/articles/the-harness-is-a-product-116274.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/generative-ai/harness-is-a-product>)

Author: Cecilia Geraldo

Published: 2026-11-04T00:00:00Z

Content type: opinion

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [Agent Harness](<https://devfeed.tech/topics/agent-harness.md>), [Product Management](<https://devfeed.tech/topics/product-management.md>), [product mindset](<https://devfeed.tech/topics/product-mindset.md>), [Maintainability](<https://devfeed.tech/topics/maintainability.md>), [LLM observability](<https://devfeed.tech/topics/llm-observability.md>), [AI software supply chain security](<https://devfeed.tech/topics/ai-software-supply-chain-security.md>), [Tech Debt](<https://devfeed.tech/topics/tech-debt.md>), [AI Context Management](<https://devfeed.tech/topics/ai-context-management.md>)

Tags: [2026](<https://devfeed.tech/tags/2026.md>), [about](<https://devfeed.tech/tags/about.md>), [accelerate](<https://devfeed.tech/tags/accelerate.md>), [accelerated](<https://devfeed.tech/tags/accelerated.md>), [acceptance](<https://devfeed.tech/tags/acceptance.md>), [acceptance-rate](<https://devfeed.tech/tags/acceptance-rate.md>), [access-controls](<https://devfeed.tech/tags/access-controls.md>), [accountability](<https://devfeed.tech/tags/accountability.md>), [adoption](<https://devfeed.tech/tags/adoption.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agent-harness](<https://devfeed.tech/tags/agent-harness.md>), [agent-sandboxing](<https://devfeed.tech/tags/agent-sandboxing.md>), [agent-skills](<https://devfeed.tech/tags/agent-skills.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-engineering](<https://devfeed.tech/tags/agentic-engineering.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [all](<https://devfeed.tech/tags/all.md>), [harness](<https://devfeed.tech/tags/harness.md>), [product](<https://devfeed.tech/tags/product.md>)

### AI overview

An agent harness is the layer between a model and its work: instructions, tools, permissions, context, state, orchestration, evaluation and telemetry. The article argues that as these parts become reusable and depend on versioned plugins, teams should manage the harness as a product, with a named owner, roadmap, dependencies and quality gates. It recommends task-based evaluations drawn from real work, versioned alongside the harness and run on meaningful changes; track cost and latency and use human review where automated scoring falls short. For self-improvement, require evaluation, provenance and rollback, automating proposals while retaining human approval. At enterprise scale, platform teams need to provide auditable access controls, sandboxing, plugin allow-lists, routing and spend policies, version-specific telemetry, deprecation support and clear agent identity and permissions. If the supported harness is slower or less capable than ad hoc alternatives, teams may route around it, creating shadow IT and security risks.

### Source excerpt

The harness is a product

## The token bleed reality: Bridging the chasm from unthrottled AI adoption to sustainable AI transformation

DevFeed: [The token bleed reality: Bridging the chasm from unthrottled AI adoption to sustainable AI transformation](<https://devfeed.tech/articles/the-token-bleed-reality-bridging-the-chasm-from-unthrottled-ai-adoption-to-sustainable-ai-transformation-116272.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/articles/the-token-bleed-reality>)

Author: Rav Hayer

Published: 2026-11-03T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [AI Strategy](<https://devfeed.tech/topics/ai-strategy.md>), [control-plane](<https://devfeed.tech/topics/control-plane.md>), [Language models](<https://devfeed.tech/topics/language-models.md>)

Tags: [agentic-coding](<https://devfeed.tech/tags/agentic-coding.md>), [agentic-engineering](<https://devfeed.tech/tags/agentic-engineering.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-factory](<https://devfeed.tech/tags/ai-factory.md>), [ai-transformation](<https://devfeed.tech/tags/ai-transformation.md>), [algorithms](<https://devfeed.tech/tags/algorithms.md>), [article](<https://devfeed.tech/tags/article.md>), [caching](<https://devfeed.tech/tags/caching.md>), [compute](<https://devfeed.tech/tags/compute.md>), [computing](<https://devfeed.tech/tags/computing.md>), [context-compression](<https://devfeed.tech/tags/context-compression.md>), [control-plane](<https://devfeed.tech/tags/control-plane.md>), [cost](<https://devfeed.tech/tags/cost.md>), [crusoe](<https://devfeed.tech/tags/crusoe.md>), [dependency-graph](<https://devfeed.tech/tags/dependency-graph.md>), [token](<https://devfeed.tech/tags/token.md>)

### AI overview

The excerpt argues that falling per-token prices have not prevented enterprise AI budgets from rising, as agentic tools multiply usage and can consume budgets without clear business value. It presents "Agentic Engineering" as an architectural response: set per-agent budgets and ownership, monitor spend in real time, govern tool and model paths before jobs run, cache repeated prompts, and consider self-hosted GPUs for more predictable costs. A financial-institution example reports 40% less token waste with unchanged feature output. The text also calls for hypothesis-driven prioritization and including embedded AI's variable compute in product TCO. It proposes hybrid quantum processing to reduce token-heavy search and context workloads, but provides no evidence or implementation detail substantiating those quantum claims. The excerpt does not specify the model-tiering strategy it introduces.

### Source excerpt

The token bleed reality: Bridging the chasm from unthrottled AI adoption to sustainable AI transformation

## The next chapter: Why quantum is the catalyst AI has been waiting for

DevFeed: [The next chapter: Why quantum is the catalyst AI has been waiting for](<https://devfeed.tech/articles/the-next-chapter-why-quantum-is-the-catalyst-ai-has-been-waiting-for-116273.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/articles/why-quantum-is-the-catalyst-ai>)

Author: Rav Hayer

Published: 2026-11-03T00:00:00Z

Content type: opinion

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [Quantum Computing](<https://devfeed.tech/topics/quantum-computing.md>), [Post-quantum cryptography](<https://devfeed.tech/topics/post-quantum-cryptography.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-and-ml](<https://devfeed.tech/tags/ai-and-ml.md>), [article](<https://devfeed.tech/tags/article.md>), [banking](<https://devfeed.tech/tags/banking.md>), [business](<https://devfeed.tech/tags/business.md>), [change](<https://devfeed.tech/tags/change.md>), [communication](<https://devfeed.tech/tags/communication.md>), [complex](<https://devfeed.tech/tags/complex.md>), [complexity](<https://devfeed.tech/tags/complexity.md>), [compute](<https://devfeed.tech/tags/compute.md>), [computer](<https://devfeed.tech/tags/computer.md>), [computing](<https://devfeed.tech/tags/computing.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cpus](<https://devfeed.tech/tags/cpus.md>), [critical](<https://devfeed.tech/tags/critical.md>), [cryptography](<https://devfeed.tech/tags/cryptography.md>), [cyber](<https://devfeed.tech/tags/cyber.md>), [data](<https://devfeed.tech/tags/data.md>), [decrypt](<https://devfeed.tech/tags/decrypt.md>), [defense](<https://devfeed.tech/tags/defense.md>), [defensive](<https://devfeed.tech/tags/defensive.md>), [quantum](<https://devfeed.tech/tags/quantum.md>)

### AI overview

The text argues that quantum computing could complement classical AI for problems involving extreme optimization, uncertainty and physical complexity, while creating an urgent encryption risk. It recommends moving to post-quantum cryptography now because stolen data could be decrypted later, especially where systems and sensitive data must remain secure for decades. For commercial use, it describes hybrid computing: CPUs handle logic, GPUs AI training, and quantum processors complex optimization. It says evaluation is shifting from qubit counts toward error rates, cost and performance on specific business problems. Potential applications include logistics and financial risk modeling, precise sensing, and simulating materials and molecules. The text presents talent and infrastructure as capabilities to build ahead of broader adoption; its market-size and compute-demand figures are estimates or forecasts.

### Source excerpt

The next chapter: Why quantum is the catalyst AI has been waiting for

## How we built DriftSense to solve AI-era design drift

DevFeed: [How we built DriftSense to solve AI-era design drift](<https://devfeed.tech/articles/how-we-built-driftsense-to-solve-ai-era-design-drift-119839.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/digital-innovation/how-built-driftsense-solve-ai-era-design-drift>)

Author: Shunfa Xu, Jeen Su, Yooboo Park

Published: 2026-11-03T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [Design system](<https://devfeed.tech/topics/design-system.md>), [AI in frontend development](<https://devfeed.tech/topics/ai-in-frontend-development.md>), [AI-enhanced product development](<https://devfeed.tech/topics/ai-enhanced-product-development.md>), [User interface design](<https://devfeed.tech/topics/ui-design.md>), [Software Engineering](<https://devfeed.tech/topics/software-engineering.md>), [engineering-culture](<https://devfeed.tech/topics/engineering-culture.md>), [human centric design](<https://devfeed.tech/topics/human-centric-design.md>), [Figma](<https://devfeed.tech/topics/figma.md>)

Tags: [accessibility](<https://devfeed.tech/tags/accessibility.md>), [accessible](<https://devfeed.tech/tags/accessible.md>), [adversarial](<https://devfeed.tech/tags/adversarial.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-workflows](<https://devfeed.tech/tags/agentic-workflows.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-coding-agents](<https://devfeed.tech/tags/ai-coding-agents.md>), [alignment](<https://devfeed.tech/tags/alignment.md>), [analyst](<https://devfeed.tech/tags/analyst.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [assets](<https://devfeed.tech/tags/assets.md>), [automated](<https://devfeed.tech/tags/automated.md>), [automatically](<https://devfeed.tech/tags/automatically.md>), [autonomous](<https://devfeed.tech/tags/autonomous.md>), [autonomous-coding](<https://devfeed.tech/tags/autonomous-coding.md>), [brand](<https://devfeed.tech/tags/brand.md>), [building-with-ai](<https://devfeed.tech/tags/building-with-ai.md>), [business-analyst](<https://devfeed.tech/tags/business-analyst.md>), [design](<https://devfeed.tech/tags/design.md>)

### AI overview

DriftSense is a hackathon-built prototype for governing design-system consistency as AI speeds up frontend coding. It proposes two loops: checking pull requests and local commits against design tokens and component contracts, with actionable feedback before merge; and continuously scanning repositories to score existing drift and generate one-click remediation pull requests. The article distinguishes this from Figma MCP, which gives AI context about design assets; coding agents, which generate code; and Percy or Chromatic, which detect pixel changes but do not explain semantic violations such as bypassed tokens or accessibility issues. The team reports winning the North America regional semifinal and describes using AI for ideation, mockups, scaffolding, and critique. It presents DriftSense as a prototype and says its evaluation engine, multi-repository analysis, and integrations remain under development.

### Source excerpt

How we built DriftSense to solve AI-era design drift

## AI readiness starts with data ownership

DevFeed: [AI readiness starts with data ownership](<https://devfeed.tech/articles/ai-readiness-starts-with-data-ownership-113313.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/data-strategy/ai-readiness-starts-with-data-ownership>)

Author: Yuvaraj Birari

Published: 2026-11-02T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [data ownership](<https://devfeed.tech/topics/data-ownership.md>), [AI-Ready Data](<https://devfeed.tech/topics/ai-ready-data.md>), [data-engineering](<https://devfeed.tech/topics/data-engineering.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>)

Tags: [about](<https://devfeed.tech/tags/about.md>), [access-controls](<https://devfeed.tech/tags/access-controls.md>), [access-requests](<https://devfeed.tech/tags/access-requests.md>), [accessible](<https://devfeed.tech/tags/accessible.md>), [accountability](<https://devfeed.tech/tags/accountability.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-ai](<https://devfeed.tech/tags/agentic-ai.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-applications](<https://devfeed.tech/tags/ai-applications.md>), [ai-architecture](<https://devfeed.tech/tags/ai-architecture.md>), [ai-platform](<https://devfeed.tech/tags/ai-platform.md>), [ai-readiness](<https://devfeed.tech/tags/ai-readiness.md>), [also](<https://devfeed.tech/tags/also.md>), [data](<https://devfeed.tech/tags/data.md>), [data-ownership](<https://devfeed.tech/tags/data-ownership.md>), [ownership](<https://devfeed.tech/tags/ownership.md>), [with](<https://devfeed.tech/tags/with.md>)

### AI overview

The article's central finding is that AI readiness depends on trusted, understandable data, which requires clear ownership. It describes shifting accountability for data quality, context, availability and usability to the business domains that know the data, while central technology teams provide infrastructure. Standardized data products--with documented metadata, interfaces, quality expectations, access controls and operational responsibilities--make data easier for people and agents to discover and use. A self-service platform can provide reusable capabilities and guided publishing, while federated governance and policy-as-code enforce shared standards without relying on manual approvals. Suggested readiness measures include discoverable and reused products, compliance with quality and service objectives, onboarding time and the speed of supplying trusted data for agentic use cases. The account presents a proposed approach and architecture for one client, not evidence of measured outcomes.

### Source excerpt

AI readiness starts with data ownership

## The 14 GB cold-start problem: Rebuilding self-hosted GitHub Actions runners on Azure Kubernetes Service

DevFeed: [The 14 GB cold-start problem: Rebuilding self-hosted GitHub Actions runners on Azure Kubernetes Service](<https://devfeed.tech/articles/the-14-gb-cold-start-problem-rebuilding-self-hosted-github-actions-runners-on-azure-kubernetes-service-105698.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/articles/rebuilding-self-hosted-github-actions-runners-on-aks>)

Author: Renicius Pagotto Fostaini

Published: 2026-10-28T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [Azure Kubernetes Service](<https://devfeed.tech/topics/azure-kubernetes-service.md>), [self-hosted runner](<https://devfeed.tech/topics/self-hosted-runner.md>), [GitHub Actions](<https://devfeed.tech/topics/github-actions.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [internal developer platform automation](<https://devfeed.tech/topics/internal-developer-platform-automation.md>), [platform engineering automation](<https://devfeed.tech/topics/platform-engineering-automation.md>), [Azure](<https://devfeed.tech/topics/azure.md>)

Tags: [actions](<https://devfeed.tech/tags/actions.md>), [aks](<https://devfeed.tech/tags/aks.md>), [arc](<https://devfeed.tech/tags/arc.md>), [autoscaling](<https://devfeed.tech/tags/autoscaling.md>), [azure](<https://devfeed.tech/tags/azure.md>), [azure-container-apps](<https://devfeed.tech/tags/azure-container-apps.md>), [azure-kubernetes-service](<https://devfeed.tech/tags/azure-kubernetes-service.md>), [azure-virtual-machines](<https://devfeed.tech/tags/azure-virtual-machines.md>), [build](<https://devfeed.tech/tags/build.md>), [build-times](<https://devfeed.tech/tags/build-times.md>), [build-tools](<https://devfeed.tech/tags/build-tools.md>), [cache](<https://devfeed.tech/tags/cache.md>), [caching](<https://devfeed.tech/tags/caching.md>), [capacity](<https://devfeed.tech/tags/capacity.md>), [cold-start-problem](<https://devfeed.tech/tags/cold-start-problem.md>), [github](<https://devfeed.tech/tags/github.md>), [github-actions](<https://devfeed.tech/tags/github-actions.md>), [kubernetes](<https://devfeed.tech/tags/kubernetes.md>), [net-framework](<https://devfeed.tech/tags/net-framework.md>), [self-hosted](<https://devfeed.tech/tags/self-hosted.md>)

### AI overview

For this team, combining ephemeral Windows runners on Azure Kubernetes Service (AKS), GitHub Actions Runner Controller (ARC), cached container images and prebuilt toolchains reduced reported pipeline duration from about 40 minutes to 10-15 minutes. Serverless containers first enabled more parallel work and brought runs to about 30 minutes, but cold starts spent 5-10 minutes pulling a 14 GB image; on AKS, a cached pull reportedly took 10-20 seconds. The team separated lightweight runner coordination from scalable worker nodes, used off-hours scale-down, ephemeral OS disks and image garbage collection, and packaged reusable workflows for application teams. These results are specific to the implementation: scaling workers to zero can lose the cache and restore cold-start delays, while keeping warm capacity trades lower latency for idle cost. The excerpt also reports an approximately 66% compute-cost reduction (about $7,800 annualized), contingent on comparable measurement scope. AKS adds operational responsibilities; the article recommends measuring queue and end-to-end time, cold versus warm starts, costs, and behavior at configured concurrency limits.

### Source excerpt

The 14 GB cold-start problem: Rebuilding self-hosted GitHub Actions runners on Azure Kubernetes Service

## Ethical and commercial risk regarding product development, GDR and AI

DevFeed: [Ethical and commercial risk regarding product development, GDR and AI](<https://devfeed.tech/articles/ethical-and-commercial-risk-regarding-product-development-gdr-and-ai-100078.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/product-innovation/ethical-commercial-risk-regarding-product-development-gdr-ai>)

Author: Mike Tiffany

Published: 2026-10-26T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [ai regulation](<https://devfeed.tech/topics/ai-regulation.md>), [ai-governance](<https://devfeed.tech/topics/ai-governance.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Product Management](<https://devfeed.tech/topics/product-management.md>), [ai-ethics](<https://devfeed.tech/topics/ai-ethics.md>)

Tags: [2](<https://devfeed.tech/tags/2.md>), [3](<https://devfeed.tech/tags/3.md>), [acceptance](<https://devfeed.tech/tags/acceptance.md>), [actions](<https://devfeed.tech/tags/actions.md>), [adoption](<https://devfeed.tech/tags/adoption.md>), [advice](<https://devfeed.tech/tags/advice.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agent-identity](<https://devfeed.tech/tags/agent-identity.md>), [agent-memory](<https://devfeed.tech/tags/agent-memory.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-adoption](<https://devfeed.tech/tags/ai-adoption.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [ai-assistant](<https://devfeed.tech/tags/ai-assistant.md>), [ai-ethics](<https://devfeed.tech/tags/ai-ethics.md>), [ai-risk](<https://devfeed.tech/tags/ai-risk.md>), [development](<https://devfeed.tech/tags/development.md>), [product](<https://devfeed.tech/tags/product.md>), [product-development](<https://devfeed.tech/tags/product-development.md>), [risk](<https://devfeed.tech/tags/risk.md>)

### AI overview

The article argues that AI accountability must be built into product decisions early, since a team's role, feature risk and agent autonomy shape its regulatory and commercial exposure. It distinguishes GDPR controller/processor responsibilities from the EU AI Act's provider/deployer roles: using a vendor tool internally is typically deployment, while shipping AI under a company's name generally makes it a provider, even when a vendor supplies the model. It recommends scoped, short-lived agent credentials, human approval for high-impact actions, explicit authorization logs, tenant isolation, data sanitization and lifecycle tracking for derived data such as embeddings and agent memory. Product managers should turn these concerns into engineering requirements and vendor checks before architecture and procurement are settled. In the article's example, replacing cross-customer model fine-tuning with separate customer RAG indexes lowered the organization's internal risk rating from high to medium and enabled routine review. The piece presents opinion, not legal advice.

### Source excerpt

Ethical and commercial risk regarding product development, GDR and AI

## The agentic control tower: From supply chain visibility to action

DevFeed: [The agentic control tower: From supply chain visibility to action](<https://devfeed.tech/articles/the-agentic-control-tower-from-supply-chain-visibility-to-action-98310.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/articles/the-agentic-control-tower-supply-chain>)

Author: Saumya Chandra, Thiyagu Palanisamy

Published: 2026-10-23T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [Agentic AI](<https://devfeed.tech/topics/what-is-agentic-ai.md>), [LLM observability](<https://devfeed.tech/topics/llm-observability.md>), [Data Management](<https://devfeed.tech/topics/data-management.md>), [streaming data analytics](<https://devfeed.tech/topics/streaming-data-analytics.md>), [Automation](<https://devfeed.tech/topics/automation.md>)

Tags: [2026](<https://devfeed.tech/tags/2026.md>), [3](<https://devfeed.tech/tags/3.md>), [about](<https://devfeed.tech/tags/about.md>), [accelerators](<https://devfeed.tech/tags/accelerators.md>), [actions](<https://devfeed.tech/tags/actions.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-ai](<https://devfeed.tech/tags/agentic-ai.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [alert](<https://devfeed.tech/tags/alert.md>), [all](<https://devfeed.tech/tags/all.md>), [also](<https://devfeed.tech/tags/also.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [approach](<https://devfeed.tech/tags/approach.md>), [architectures](<https://devfeed.tech/tags/architectures.md>), [control](<https://devfeed.tech/tags/control.md>), [supply-chain](<https://devfeed.tech/tags/supply-chain.md>), [tower](<https://devfeed.tech/tags/tower.md>), [visibility](<https://devfeed.tech/tags/visibility.md>)

### AI overview

The article argues that traditional supply chain control towers show fragmented signals but leave teams to interpret disruptions and coordinate responses, which can delay action and produce conflicting, siloed decisions. Thoughtworks' agentic control tower organizes work into Sense, Simulate and Solve: specialized agents connect internal and external signals, model response options and trade-offs across the value chain, then coordinate agreed actions with downstream systems and teams. Humans assess options and make decisions. The article says implementation depends on timely, agent-ready data; business context; coordinated agents with guardrails and checkpoints; and traceable, observable recommendations. Thoughtworks describes a semantic intelligence layer, agent orchestration with Agent/works™ governance, and scenario-based decision experiences, with a 3/3/3 cadence for validating, prototyping and moving toward production. The approach can start with sensing scenarios and expand over time; the article does not present agents as replacing human decision-makers.

### Source excerpt

The agentic control tower: From supply chain visibility to action

## How Sensible Harness can help teams create an AI-first delivery environment

DevFeed: [How Sensible Harness can help teams create an AI-first delivery environment](<https://devfeed.tech/articles/how-sensible-harness-can-help-teams-create-an-ai-first-delivery-environment-95432.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/generative-ai/how-sensible-harness-help-teams-create-ai-first-delivery-environment>)

Author: Vitória Regina Ragazzon Toebe, Amy Pellegrini, María José Lalama

Published: 2026-10-22T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [Agent Harness](<https://devfeed.tech/topics/agent-harness.md>), [engineering-culture](<https://devfeed.tech/topics/engineering-culture.md>), [Code review](<https://devfeed.tech/topics/code-review.md>), [Test-driven development](<https://devfeed.tech/topics/tdd.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agent-costs](<https://devfeed.tech/tags/agent-costs.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-ai](<https://devfeed.tech/tags/agentic-ai.md>), [agentic-workflow](<https://devfeed.tech/tags/agentic-workflow.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-adoption](<https://devfeed.tech/tags/ai-adoption.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [ai-agent-costs](<https://devfeed.tech/tags/ai-agent-costs.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-assisted](<https://devfeed.tech/tags/ai-assisted.md>), [ai-code](<https://devfeed.tech/tags/ai-code.md>), [ai-code-generation](<https://devfeed.tech/tags/ai-code-generation.md>), [ai-cost](<https://devfeed.tech/tags/ai-cost.md>), [ai-governance](<https://devfeed.tech/tags/ai-governance.md>), [delivery](<https://devfeed.tech/tags/delivery.md>), [environment](<https://devfeed.tech/tags/environment.md>), [harness](<https://devfeed.tech/tags/harness.md>), [teams](<https://devfeed.tech/tags/teams.md>)

### AI overview

Sensible Harness is a CLI that scaffolds repositories for AI-assisted delivery with governed workflows, reusable skills, shared state, and a code audit. Its audit grades a repository from A to F across 10 dimensions against its documented standard, with Markdown reports and interactive HTML diagnostics. Three agents handle requirements, implementation, and quality checks: they use human approval checkpoints, and mark unexecutable test scenarios as blocked. The developer workflow includes acceptance-criteria checks, TDD cycles, commits, pull requests, and CI gates. Skills are versioned from templates and checked for drift in CI. The harness assigns models by task and reports estimated model costs to each ticket. The text describes the tool and internal demonstrations; it does not provide independent comparative results against other products.

### Source excerpt

How Sensible Harness can help teams create an AI-first delivery environment

## Don't ask AI to write the user story yet

DevFeed: [Don't ask AI to write the user story yet](<https://devfeed.tech/articles/don-t-ask-ai-to-write-the-user-story-yet-95319.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/generative-ai/dont-ask-ai-to-write-the-user-story-yet>)

Author: Himanshu Sharma

Published: 2026-10-22T00:00:00Z

Content type: opinion

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [User Stories](<https://devfeed.tech/topics/user-stories.md>), [Business, Industry, Usability, User Experience, User Research, Web Strategy](<https://devfeed.tech/topics/business-industry-usability-user-experience-user-research-web-strategy.md>), [AI-enhanced product development](<https://devfeed.tech/topics/ai-enhanced-product-development.md>)

Tags: [acceptance](<https://devfeed.tech/tags/acceptance.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-assistant](<https://devfeed.tech/tags/ai-assistant.md>), [business](<https://devfeed.tech/tags/business.md>), [csv](<https://devfeed.tech/tags/csv.md>), [customer](<https://devfeed.tech/tags/customer.md>), [customer-experience](<https://devfeed.tech/tags/customer-experience.md>), [customers](<https://devfeed.tech/tags/customers.md>), [dashboard](<https://devfeed.tech/tags/dashboard.md>), [data](<https://devfeed.tech/tags/data.md>), [development](<https://devfeed.tech/tags/development.md>), [discussion](<https://devfeed.tech/tags/discussion.md>), [download](<https://devfeed.tech/tags/download.md>), [draft](<https://devfeed.tech/tags/draft.md>), [example](<https://devfeed.tech/tags/example.md>), [export](<https://devfeed.tech/tags/export.md>), [feature](<https://devfeed.tech/tags/feature.md>), [file](<https://devfeed.tech/tags/file.md>), [story](<https://devfeed.tech/tags/story.md>), [user](<https://devfeed.tech/tags/user.md>)

### AI overview

The article recommends using AI to uncover unresolved product decisions before asking it to draft a user story. Give it the task, desired outcome, confirmed rules and source notes; have it separate confirmed facts, assumptions and unknowns, quote evidence for confirmed items, and suggest stakeholder questions without filling gaps. Prioritize questions whose answers could change scope, permissions or the meaning of data or outcomes, and keep optional enhancements separate. Bring useful questions and concrete examples into refinement, Three Amigos or Example Mapping, where the team must agree on behavior. Record each decision, owner, source and remaining uncertainty before drafting acceptance criteria from confirmed decisions. The excerpt stresses that analyst judgment remains necessary and notes that date boundaries and access rules may still need explicit definitions.

### Source excerpt

Don't ask AI to write the user story yet

## Cheaper code won't kill software development, but it's already reshaping it

DevFeed: [Cheaper code won't kill software development, but it's already reshaping it](<https://devfeed.tech/articles/cheaper-code-won-t-kill-software-development-but-it-s-already-reshaping-it-90161.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/articles/cheaper-code-wont-kill-software-development>)

Author: Chris Ford

Published: 2026-10-19T00:00:00Z

Content type: opinion

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [rachels-ramblings](<https://devfeed.tech/topics/rachels-ramblings.md>), [software-development](<https://devfeed.tech/topics/software-development.md>), [Vibe coding](<https://devfeed.tech/topics/vibe-coding.md>), [Product Management](<https://devfeed.tech/topics/product-management.md>), [Developer experience](<https://devfeed.tech/topics/developer-experience.md>), [Code Modularity](<https://devfeed.tech/topics/code-modularity.md>), [google-adk](<https://devfeed.tech/topics/google-adk.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [career](<https://devfeed.tech/tags/career.md>), [coding](<https://devfeed.tech/tags/coding.md>), [coding-agent](<https://devfeed.tech/tags/coding-agent.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [data](<https://devfeed.tech/tags/data.md>), [developer](<https://devfeed.tech/tags/developer.md>), [developers](<https://devfeed.tech/tags/developers.md>), [it-s](<https://devfeed.tech/tags/it-s.md>)

### AI overview

The article argues that cheaper code generation from coding agents is likely to expand software development, while shifting constraints toward validation, quality, governance, and organizational choices. It surveys possible effects on SaaS, staffing, junior developer training, software ownership, and engineering careers, while marking several predictions as uncertain.

### Source excerpt

Cheaper code won't kill software development, but it's already reshaping it

## Why executive judgment decides who wins with AI

DevFeed: [Why executive judgment decides who wins with AI](<https://devfeed.tech/articles/why-executive-judgment-decides-who-wins-with-ai-88660.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/technology-strategy/why-executive-judgment-decides-who-wins-with-ai>)

Author: Tom Holbourn, Alexander Moore

Published: 2026-10-16T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [agentic workflows](<https://devfeed.tech/topics/agentic-workflows.md>), [anthropic](<https://devfeed.tech/topics/anthropic.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [breach](<https://devfeed.tech/tags/breach.md>), [build](<https://devfeed.tech/tags/build.md>), [cio](<https://devfeed.tech/tags/cio.md>), [cios](<https://devfeed.tech/tags/cios.md>), [claude](<https://devfeed.tech/tags/claude.md>), [claude-opus](<https://devfeed.tech/tags/claude-opus.md>), [cost](<https://devfeed.tech/tags/cost.md>), [defensive](<https://devfeed.tech/tags/defensive.md>), [delivery](<https://devfeed.tech/tags/delivery.md>), [detection](<https://devfeed.tech/tags/detection.md>), [executive](<https://devfeed.tech/tags/executive.md>), [technology-strategy](<https://devfeed.tech/tags/technology-strategy.md>)

### AI overview

At a technology leadership forum, Thoughtworks' Alex Moore argues that human judgment, clear accountability and measurable workflows are key to getting value from AI systems. The article recommends defining what agents may do, keeping experts able to override them, choosing deliberately what AI capabilities to own or rent, and using verifiable outcomes and checks before outputs reach systems of record. It reports that organizations applying checked workflows in software delivery have seen coding price-performance improve by 50% to 80% and delivery speed rise by as much as 82%.

### Source excerpt

Why executive judgment decides who wins with AI

## The future of SaaS in the agentic world: Beyond the disruption narrative

DevFeed: [The future of SaaS in the agentic world: Beyond the disruption narrative](<https://devfeed.tech/articles/the-future-of-saas-in-the-agentic-world-beyond-the-disruption-narrative-87893.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/technology-strategy/the-future-of-saas-in-the-agentic-world--beyond-the-disruption-n>)

Author: Tom Holbourn

Published: 2026-10-15T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [SaaS Management](<https://devfeed.tech/topics/saas-management.md>), [agentic-coding](<https://devfeed.tech/topics/agentic-coding.md>), [Loop Engineering](<https://devfeed.tech/topics/loop-engineering.md>), [rachels-ramblings](<https://devfeed.tech/topics/rachels-ramblings.md>)

Tags: [agentic](<https://devfeed.tech/tags/agentic.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-value](<https://devfeed.tech/tags/ai-value.md>), [build-vs-buy](<https://devfeed.tech/tags/build-vs-buy.md>), [business](<https://devfeed.tech/tags/business.md>), [data](<https://devfeed.tech/tags/data.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [saas](<https://devfeed.tech/tags/saas.md>)

### AI overview

A panel of enterprise technology leaders discussed how AI agents are reshaping SaaS build-versus-buy decisions, pricing, user interfaces and system architecture. They described a move toward hybrid pricing tied partly to usage or outcomes, applications serving as systems of context, and agent-executed workflows, while noting that frontline workers may still need directive interfaces.

### Source excerpt

The future of SaaS in the agentic world: Beyond the disruption narrative

## The shift to the AI-first enterprise: Insights from technology leaders forum

DevFeed: [The shift to the AI-first enterprise: Insights from technology leaders forum](<https://devfeed.tech/articles/the-shift-to-the-ai-first-enterprise-insights-from-technology-leaders-forum-87894.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/technology-strategy/the-shift-to-the-ai-first-enterprise-technology-forum-2026-insights>)

Author: Tom Holbourn

Published: 2026-10-15T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [ai security](<https://devfeed.tech/topics/ai-security.md>), [LLM observability](<https://devfeed.tech/topics/llm-observability.md>), [API Monetization](<https://devfeed.tech/topics/api-monetization.md>), [observability](<https://devfeed.tech/topics/observability.md>), [session affinity](<https://devfeed.tech/topics/session-affinity.md>)

Tags: [agentic](<https://devfeed.tech/tags/agentic.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-integration](<https://devfeed.tech/tags/ai-integration.md>), [api](<https://devfeed.tech/tags/api.md>), [build-versus-buy](<https://devfeed.tech/tags/build-versus-buy.md>), [coaching](<https://devfeed.tech/tags/coaching.md>)

### AI overview

At Thoughtworks' Technology Leaders Forum, executives and practitioners discussed how organizations can move from basic chatbots toward AI systems that carry out multi-step work. The article covers autonomy boundaries, human oversight, governance and security, changing SaaS interfaces and pricing, and the data and architecture needed to support agents. It also describes how cross-functional teams, build-versus-buy choices, and feedback loops shape enterprise AI strategy.

### Source excerpt

The shift to the AI-first enterprise: Insights from technology leaders forum

## Renting the Frontier, Owning the Core IP: How We Cut AI Forecast Costs from $40M to $11M

DevFeed: [Renting the Frontier, Owning the Core IP: How We Cut AI Forecast Costs from $40M to $11M](<https://devfeed.tech/articles/renting-the-frontier-owning-the-core-ip-how-we-cut-ai-forecast-costs-from-40m-to-11m-90545.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/generative-ai/renting-the-frontier--owning-the-core-ip--how-we-cut-ai-forecast>)

Author: Andy Nolan

Published: 2026-10-15T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [Frontier AI](<https://devfeed.tech/topics/frontier-ai.md>), [Claude](<https://devfeed.tech/topics/claude.md>), [LLM security](<https://devfeed.tech/topics/llm-security.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [api](<https://devfeed.tech/tags/api.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [control-plane](<https://devfeed.tech/tags/control-plane.md>), [costs](<https://devfeed.tech/tags/costs.md>)

### AI overview

Thoughtworks describes a token-sourcing strategy that routes workloads among frontier model APIs, open-weight models, managed inference, and rented GPUs. It bases those choices on measured traffic, workload needs, cost, IP sensitivity, and regional constraints. The company reports that its forecast puts 2028 spending at $6.8 million to $11 million, compared with a do-nothing forecast of $16 million and a worst case of $40 million.

### Source excerpt

Renting the Frontier, Owning the Core IP: How We Cut AI Forecast Costs from $40M to $11M

## Evaluating decision models for document categorization

DevFeed: [Evaluating decision models for document categorization](<https://devfeed.tech/articles/evaluating-decision-models-for-document-categorization-87892.md>)

Original publisher: [Read original article](<https://www.thoughtworks.com/insights/blog/generative-ai/evaluating-decision-models-for-document-categorization>)

Author: Pallavi Sinha

Published: 2026-10-15T00:00:00Z

Content type: article

Language: en

Sources: [Thoughtworks Insights](<https://devfeed.tech/sources/thoughtworks-insights.md>)

Topics: [Factuality scoring in AI](<https://devfeed.tech/topics/factuality-scoring-in-ai.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [gpt-6-astra](<https://devfeed.tech/topics/gpt-6-astra.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [api](<https://devfeed.tech/tags/api.md>), [benchmarking](<https://devfeed.tech/tags/benchmarking.md>), [claude-sonnet](<https://devfeed.tech/tags/claude-sonnet.md>), [context-window](<https://devfeed.tech/tags/context-window.md>), [decision-models](<https://devfeed.tech/tags/decision-models.md>), [human-review](<https://devfeed.tech/tags/human-review.md>)

### AI overview

The article compares Laya, a locally runnable decision model, with Claude Sonnet for constrained document categorization. In a small benchmark on six technical documents, both models classified five correctly; Laya averaged 1.72 seconds and Claude Sonnet 1.45 seconds. Laya also returns probabilities that can signal uncertainty for thresholding or human review. The article discusses criteria, latency, and chunking, noting that sentence-based chunking reduced section count and latency by about 40% without changing classifications in the experiment.

### Source excerpt

Evaluating decision models for document categorization

[Next page](<https://devfeed.tech/sources/thoughtworks-insights.md?cursor=WyIyMDI2LTEwLTE1VDAwOjAwOjAwKzAwOjAwIiwgIjdiNDEzYzRlLWMzNjYtNGRmMC1iZDExLWM1NzNmNmJjMGQxNSJd>)