# 8th Light Published articles for 8th Light. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## From AI Code to Trusted Software: Harness Engineering in Practice DevFeed: [From AI Code to Trusted Software: Harness Engineering in Practice]() Original publisher: [Read original article]() Author: Travis Frisinger Published: 2026-09-04T21:55:00Z Content type: opinion Language: en Sources: [8th Light](), [8th Light Insights]() Topics: [AI-assisted coding](), [Software](), [trust](), [observability](), [Authorization]() Tags: [agent](), [ai](), [ai-and-emerging-tech](), [code](), [engineering](), [engineering-and-devops](), [observability](), [permissions](), [trust](), [verification]() ### AI overview The article presents harness engineering as a repository-centered discipline for making AI-generated software more trustworthy. It argues that prompts and written standards are insufficient, and that permissions, quality gates, evidence, and observability should enforce organizational standards and support verification. ### Source excerpt If the same reasoning path writes the system change and defines the proof of success, you may be setting yourself up for an avoidable failure in the future. Harness engineering is meant to act as an extension of your own organizational guardrails, which were always meant to reduce risk and improve quality. Travis Frisinger, Head of Agentic AI Your team is shipping more AI-written code every quarter. How do you know it is any good? Good means it meets your standards, and you have probably already tried handing your agents the standards: a context file, a style guide, the wiki pasted into the prompt. The agent reads them, agrees, and still breaks them, because instructions to a model are suggestions. What the repository permits is what actually happens. Your people absorb standards through review comments and hallway corrections, and the lessons stick. An agent apologizes and forgets by the next session. The only place its lessons can accumulate is the repository itself. That gap used to be an annoyance. With AI doing real engineering work, the quality gap is the whole game. Harness engineering is the discipline that closes it. It is the process of imbuing a repository with your standards so that the repository itself enforces them: permissions and boundaries that say what any actor may touch, quality gates that fail closed, evidence attached to every change, and observability that spans runs rather than moments. Models supply software delivery capacity. The harness supplies observable accountability: every change carries what was done, which rule allowed it, and what happened as a result, no matter which model, agent, or person did the work. Why now The industry started using the term harness engineering back in February, 2026. Since then, Thoughtworks, LangChain, and others have built serious thought leadership around the same shape. When several firms independently converge on the same word, it usually means they are trying to name the same problem. The real proble ## Waggy, Claude, and Code: Celebrating Drake Fish, Our Newest Beacon DevFeed: [Waggy, Claude, and Code: Celebrating Drake Fish, Our Newest Beacon]() Original publisher: [Read original article]() Author: Juan Santana Published: 2026-08-10T05:00:00Z Content type: opinion Language: en Sources: [8th Light](), [8th Light Insights]() Topics: [Artificial Intelligence](), [Claude](), [anthropic](), [Robotics](), [Code](), [Slack]() Tags: [ai](), [anthropic](), [claude](), [culture-and-news](), [robotics](), [slack](), [work]() ### AI overview An 8th Light profile celebrates Drake Fish as a Beacon Award recipient, highlighting his open sharing of the Waggy autonomous robot project, experimentation with Claude and real-world hardware, and work helping colleagues and clients adopt AI systems. ### Source excerpt Working at 8th Light, we have the daily privilege of collaborating with truly brilliant minds--people who bridge the gap between deep engineering and meaningful human connection. We are also incredibly fortunate to work alongside folks who, no matter how tough the problem or early the hour, are overflowing with honesty, curiosity, and of course, the jokes. That's why it brings us immense joy to announce our latest Beacon Award recipient at 8th Light, Drake Fish! The Beacon Award is a peer-nominated honor celebrating the individuals who bring 8th Light's values to life and elevate everyone around them. Drake was nominated by multiple colleagues last quarter, which tells you everything you need to know about his impact. As a Lead Engineer, Drake has built a reputation at 8th Light that reaches far beyond client teams. The secret sauce? He learns in the open, bringing everyone along for the ride. Waggy is an autonomous robot pooper scooper that Drake started building in his spare time. He stepped into the project without much background in electronics or robotics. Instead of tinkering quietly in private until everything worked, Drake shared the entire journey in Slack (mistakes, wiring mishaps, and all) so the rest of the company could learn, and giggle, alongside him. Along the way, Drake used Waggy to give us a front-row seat to what agentic tools like Claude can actually accomplish when paired with real-world hardware and a little bit of experimentation. As if building an autonomous pooper scooper wasn't enough, Drake took his AI experimentation even further. He became the first of many 8th Light team members to earn the Claude Certified Architect credential, one of Anthropic's benchmarks for engineers building production-grade AI systems. Right after passing the exam, he presented to the company to share tips, break down the material, and demystify the process. If you drop into our internal AI Slack channels or our "AI in the Open" sessions, odds are high you will f ## Teaching the Agent Our Craft: Structured Agentic Development on a Real Codebase DevFeed: [Teaching the Agent Our Craft: Structured Agentic Development on a Real Codebase]() Original publisher: [Read original article]() Author: Alex Haldeman Published: 2026-07-06T16:00:00Z Content type: article Language: en Sources: [8th Light](), [8th Light Insights]() Topics: [Agentic development](), [Claude Code](), [Development](), [AI-assisted coding](), [Test-driven development]() Tags: [agentic](), [agentic-development](), [ai-and-emerging-tech](), [claude-code](), [clean-architecture](), [structured](), [test]() ### AI overview 8th Light describes a structured agentic development workflow built around Claude Code. The approach adapts Research-Plan-Implement by separating research, planning, and implementation, adding review cycles and involving product managers and designers at each transition. It also applies test-driven development, clean architecture, and explicit project conventions to address common agent failure modes. ### Source excerpt The Mission We recently partnered with a startup that had developed a clinically proven approach to alleviating neuroplastic chronic pain. Their program worked: a coach-led model that helped patients ease chronic pain at a lower cost than conventional treatment. The problem was reach. The in-person model could not scale to the demand they were seeing, and there were not enough coaches to close the gap. We were brought in to build a digital platform that could deliver the program to every patient who needed it. Our Development Philosophy and Inspiration At 8th Light, we approach agentic development the same way we approach any software engagement: with discipline around test-driven development, clean architecture, and code that is built to embrace change. But agentic development comes with its own failure modes. An agent produces code that covers the happy path and misses critical behaviors. A context window fills with stale reasoning from earlier attempts, and the agent starts working against itself. Without explicit conventions, the output works but looks like nobody on the team wrote it. Tyler Burleigh's Research-Plan-Implement gave us a useful frame for thinking about this. His core observation: the bottleneck is not code generation, it is ensuring the model understands what to build before it starts building. RPI addresses that by separating research, planning, and implementation into distinct phases, each with a review cycle before the next begins. We adapted that structure into our Claude Code workflow, with the additional goal of keeping product managers and designers genuinely in the loop at each transition, not just developers. What follows is a description of the harness we built from a Claude Code-specific perspective. The HarnessStructure Before walking through the pieces, it helps to see how they fit together. Everything that teaches the agent our craft lives in a handful of files at the project root and inside a single .claude/ directory. None of it is ## Celebrating our Beacon: Kevin Buchanan DevFeed: [Celebrating our Beacon: Kevin Buchanan]() Original publisher: [Read original article]() Author: Juan Santana Published: 2026-06-23T05:00:00Z Content type: article Language: en Sources: [8th Light](), [8th Light Insights]() Topics: [Artificial Intelligence](), [Claude]() Tags: [ai](), [claude](), [culture-and-news](), [productivity](), [work]() ### AI overview 8th Light recognizes Principal Engineer Kevin Buchanan with its Beacon Award. The article highlights his role in helping a client adopt Claude and develop an AI-supported Research-Plan-Implement workflow that improved productivity while reducing costs and time. ### Source excerpt Congratulations to Kevin Buchanan, our latest Beacon Award recipient, a peer-nominated honor that celebrates the people who bring our values to life and light the way for others to follow. Kevin, Principal Engineer, has been an 8th Light for over 12 years. A long-time favorite of our clients and his peers, he has a reputation for raising the bar on whatever he touches, and those who work alongside him come away sharper and stronger for the experience. Kevin's nomination came out of a year-long re-platforming effort for a long-time client. As AI reshaped what software teams could do, the client was questioning what was truly possible given their limited resources. Collaborating with past Beacon honorees Kelsey Roy and Eric Olson, Kevin introduced AI-forward solutions in a practical, grounded way. He helped the client adopt Claude, build patterns for using it well, and shape a Research-Plan-Implement (RPI) workflow they've kept refining. The result was real productivity for the team, but more importantly, cost and time savings for our client. Kevin showed what it looks like to stand on years of craft and still adapt quickly, bringing others along as he goes. Our CEO, Jenny Farver, put it well: "I'm really excited to see somebody like Kevin, who can stand on all that experience he's been building for years at 8th Light, and also rapidly change, adapt, grow, and bring along others with him." Join us in congratulating Kevin Buchanan, our newest Beacon! ## Production is the New Prototype: Notes from LangChain Interrupt 2026 DevFeed: [Production is the New Prototype: Notes from LangChain Interrupt 2026]() Original publisher: [Read original article]() Author: Shawn DeVries Published: 2026-05-17T14:45:00Z Content type: opinion Language: en Sources: [8th Light](), [8th Light Insights]() Topics: [Artificial Intelligence](), [observability](), [Tooling](), [Demo]() Tags: [2026](), [agentic](), [ai](), [ai-and-emerging-tech](), [conference](), [enterprise](), [observability](), [production](), [technical]() ### AI overview An opinion article from 8th Light about discussions at LangChain Interrupt 2026, arguing that agent systems are moving from demonstrations toward production operations. It emphasizes reliability, governance, enterprise constraints, and observability as essential to operating these systems. ### Source excerpt 8th Light's Head of Agentic AI, Travis Frisinger, and I spent two days in San Francisco at LangChain Interrupt 2026 watching what happens when agent demos meet production traffic. The room was full of companies that have crossed that line, and a few that wish they hadn't tried. The conversation a year ago was "can agents work?" The conversation in May was "how do we operate reliable, observable, governable agent systems at scale?" In the discussions we had at the conference, we validated that the shift was real and not just product marketing. What follows is what stood out across the two days, with Travis's read on the technical patterns and what I'm hearing from clients who are trying to ship. Demos are Easy. Operating is the Job. Travis's summary of the two days: agents are moving from demos to real operational systems, with companies like Lyft, Cisco, Toyota, and LATAM already running large-scale production deployments. The keynote framed agents as different from traditional software because of an infinite input space and non-deterministic models. The successful teams ship early and iterate. The unsuccessful ones, according to MongoDB's CJ Desai in conversation with Harrison Chase, built proof-of-concepts for two years and called it strategy. He called it agent washing. 2025 was supposed to be the year of agents at scale. It didn't materialize. He thinks 2026 will, but only for teams who treat the operational work as the actual work. That matches what I hear in client conversations. The question is rarely "can we build an agent?" It is "how do we run one inside an enterprise that has compliance, vendor reviews, change management, and a board that wants to see ROI?" Observability is Not Optional. It's the Foundation. Nearly every speaker named LangSmith or an equivalent as mandatory infrastructure rather than nice-to-have tooling. LATAM Airlines made the point with the most authority. Nico Venegas and Claudio Urbina described a concierge agent on their B2C site se ## Accelerating Delivery: How AI Agents Can Own the Initial Code Draft DevFeed: [Accelerating Delivery: How AI Agents Can Own the Initial Code Draft]() Original publisher: [Read original article]() Author: Stephen Walker Published: 2026-05-05T21:58:00Z Content type: article Language: en Sources: [8th Light](), [8th Light Insights]() Topics: [Artificial Intelligence](), [Loop Engineering](), [Pull Request](), [human review](), [jira](), [coding](), [hooks](), [Script]() Tags: [agents](), [ai](), [ai-agents](), [ai-and-emerging-tech](), [code](), [hooks](), [human-review](), [jira](), [pull-request](), [review](), [sandbox]() ### AI overview The article describes a pragmatic workflow in which coordinated AI agents turn a Jira ticket into a code draft and pull request for human review. It emphasizes safety through branch permissions, scoped access, scripts and hooks, plus deterministic structures around probabilistic language models. The supplied text is truncated before the full reliability hierarchy and later conclusions. ### Source excerpt Why This Matters in the Era of Agentic Everything The gap between a Jira ticket and the first pull request (PR) is often a graveyard of productivity. Context switching, boilerplate setup, and requirement analysis can eat up hours of a senior engineer's day before they write a single line of business logic. Recently, industry leaders have showcased "Harness Engineering" [1][2] -- the practice of building autonomous agent workflows that handle end-to-end coding tasks. While their results are inspiring, the barrier to entry can feel insurmountable. The challenge for most teams is finding a starting point that balances long-term vision with immediate ROI. We recently helped a large-scale, high-traffic consumer services platform bridge this gap. Instead of an all-or-nothing architectural overhaul, we took a pragmatic first step: automating the journey from Jira ticket to PR. The Vision: "Let's work on FEAT-123" Our goal was simple. We wanted to enable a developer to provide a single prompt -- "Let's work on FEAT-123" -- and have a coordinated team of AI agents handle the planning, coding, and verification. This culminates in a PR ready for human review. To make this a reality, we focused on three strategic pillars: safety, determinism, and quality. Safety Through Pragmatic Guardrails One of the primary concerns for any engineering leader is "agent drift" -- an AI making unauthorized or hallucinated changes. While an isolated digital sandbox is an ideal long-term goal, we proved that teams can achieve significant safety and progress using existing infrastructure: Branch Permissions: Restrict agent access to specific feature branches. Scoped Access: Use scripts and hooks to define exactly what an agent can and cannot touch. Human-in-the-Loop: Ensure that while the agent proposes the change, a human must always review, approve, and merge the code. By defining the agent's scope as a specialized contributor rather than a system administrator, we mitigated risk while maximizing ou ## Automated Accessibility Tools Support Audits but Cannot Replace Human Evaluation DevFeed: [Automated Accessibility Tools Support Audits but Cannot Replace Human Evaluation]() Original publisher: [Read original article]() Author: Celeste Aronow Published: 2026-04-15T13:44:00Z Content type: opinion Language: en Sources: [8th Light](), [8th Light Insights]() Topics: [Accessibility](), [Tooling](), [Large Language Model](), [axe-core]() Tags: [accessibility](), [ai-and-emerging-tech](), [axe-core](), [llms](), [tooling]() ### AI overview The article argues that automated accessibility scanners and LLM-assisted auditing can identify many concrete issues and expand teams' coverage, but they cannot fully assess what people with disabilities experience when using a screen reader. It also warns that scan output can include false violations and requires human judgment and accessibility expertise. ### Source excerpt The Gap is Real If your accessibility process ends with a passing scan, your users with disabilities are likely encountering barriers you do not know about. Automated tools are useful. LLM-assisted auditing has raised the ceiling. Neither closes the gap between what a tool can verify and what a person actually experiences when navigating with a screen reader. The gap is real, it is measurable, and it has organizational consequences. I have spent years running every category of automated accessibility tool the industry has produced: rule-based scanners like axe-core and WAVE, visual analyzers like Lighthouse, browser extensions with built-in screen reader simulations. When LLMs entered the picture, the promise was compelling. Tools that could finally understand context, interpret user flows, and reason about accessibility the way a human specialist does. I wanted that to be true. It isn't. What Automation Does Well Let me be fair to the tools. Automated accessibility scanners like axe-core, Lighthouse, and WAVE are genuinely good at what they do. They catch heading hierarchy violations quickly and reliably. They flag color contrast failures with precision. They identify missing alt attributes on images in many cases. These are real wins. If you have run these tools on your product, you have addressed real issues, and the engineers who did that work should feel good about it. Automated tooling is a force multiplier. It speeds up the work and amplifies how much ground a team can cover. The problem is not what these tools do. It is what they cannot do, and what organizations assume they have done. What Automation Invents The same tools I praised above (SilkTide, axe-core, Lighthouse, and WAVE) are genuinely valuable parts of a real accessibility workflow. They speed up discovery, give teams a shared vocabulary, and help non-specialists see problems that would otherwise be invisible. They also generate a lot of output. A single scan on a moderately complex page can retur ## Reframing Healthcare Technology in the AI Era DevFeed: [Reframing Healthcare Technology in the AI Era]() Original publisher: [Read original article]() Author: Jon Wettersten Published: 2026-03-06T22:14:00Z Content type: opinion Language: en Sources: [8th Light](), [8th Light Insights]() Topics: [Artificial Intelligence](), [Generative AI](), [Automation](), [decision-making](), [trust](), [digital]() Tags: [ai](), [automation](), [decision-making](), [generative-ai](), [governance](), [healthcare](), [healthtech](), [human-centered-design](), [safety](), [technology](), [transparency](), [trust]() ### AI overview This article argues that healthcare organizations should approach AI adoption through human-centered design, stronger governance, transparency, and alignment with clinical standards. It emphasizes that clinicians need technology to reduce administrative burden and support patient care without compromising safety, trust, or practical and ethical readiness. ### Source excerpt At 8th Light, we help healthcare organizations design and build AI-enabled systems that put people first, combining product strategy, human-centered design, and technology leadership to create innovative solutions that work for clinicians as well as patients. Why This Matters Although AI continues to dominate healthtech headlines, many frontline clinicians are still battling clunky systems, fragmented workflows, and administrative overload. The reality? Healthcare practitioners are not asking for more technology, they're asking for more time with patients. The tension between innovation and adoption continues to shape healthcare's digital transformation. As AI tools quickly evolve and adapt to revolutionize diagnosis, documentation, and decision-making, many healthcare providers remain cautious. Their questions are not about the potential of technology, but about its practical and ethical readiness for real-world care. For instance: How do we vet these tools? Can we trust the outcomes? Will this actually improve care or just add more work? Will this solution deliver measurable ROI without compromising patient safety? These are not questions of resistance, they're questions of responsibility. As healthcare systems race to deploy generative AI, predictive analytics, and automation, clinicians and administrators are demanding stronger governance, transparency, and alignment with clinical standards. Their goal isn't to slow innovation, but to ensure technology protects both patients and the licensed physicians who care for them. Behind these questions lies a workforce already under significant strain, and technology that fails to ease that burden risks making it worse. The Urgent Need for Human-Centered Healthtech The U.S. is already experiencing workforce strain across many health professions, driven by geographic maldistribution, burnout, and demographic pressures. Federal workforce projections show ongoing gaps across multiple professions, with non-metro communities ## A Primer: MCP Servers and the Model Context Protocol DevFeed: [A Primer: MCP Servers and the Model Context Protocol]() Original publisher: [Read original article]() Author: Priya Patil Published: 2026-02-25T06:00:00Z Content type: tutorial Language: en Sources: [8th Light](), [8th Light Insights]() Topics: [Model Context Protocol](), [Server](), [Security](), [client](), [Authentication]() Tags: [ai](), [ai-and-emerging-tech](), [ai-assistant](), [audit](), [authentication](), [mcp](), [model-context-protocol](), [protocol](), [sandboxing](), [security](), [servers]() ### AI overview This primer explains MCP (Model Context Protocol), an open protocol that gives AI models a standard interface for connecting to external tools and data sources. It also outlines current gaps in authentication, discovery, security, sandboxing, secret management, and audit trails. ### Source excerpt Navigating Change The way we interact with software is shifting. Instead of clicking through web interfaces and learning different UIs for each system, some people are moving toward natural language as the interface - telling an AI what they want done, and it handles the task completion for them. This shift is happening now, but there's been a missing piece: how does the AI actually connect to and control those systems? MCP Defined MCP (Model Context Protocol) is an open protocol that gives AI models a standardised interface to external tools and data sources. Think of it as a USB-C for AI - before USB-C, every device needed its own connector. MCP does the same for AI-system integration. Instead of building custom integrations for every combination of AI model and external service, you implement MCP once and your AI can talk to everything it needs to. Instead of context-switching between Slack, JIRA, GitHub, and dashboards for example, we'll be able to describe what we need and have an AI assistant handle the interactions. Or as Anthropic defines it: "MCP (Model Context Protocol) is an open-source standard for connecting AI applications to external systems." Current State: A Nascent Protocol Yes, MCP has gaps. Here are a few: It's relatively new (launched November 2024) and the ecosystem is still figuring things out. Authentication and authorisation have a standard but implementation is patchy. Discovery is a work in progress, with a real need for global registry mechanics. Security is largely left to implementers - no built-in sandboxing, no secret management standard, no consistent audit trails. The community is actively working on all of these, and thing are moving very fast. But these are solvable problems and a lot of the changes are happening on the client side so building out servers feels a little more stable. These gaps shouldn't prevent you from getting value today - they just mean you'll need to handle some plumbing yourself and take an abundance of cauti