# M. Touheed

Published articles for M. Touheed.

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

## Agentic workloads break assumptions about software testing. Here's how to cope

DevFeed: [Agentic workloads break assumptions about software testing. Here's how to cope](<https://devfeed.tech/articles/agentic-workloads-break-assumptions-about-software-testing-here-s-how-to-cope-84704.md>)

Original publisher: [Read original article](<https://siliconangle.com/2026/10/11/agentic-workloads-break-assumptions-about-software-testing-heres-how-to-cope/>)

Author: M. Touheed

Published: 2026-10-11T13:00:56Z

Content type: article

Language: en

Sources: [SiliconANGLE](<https://devfeed.tech/sources/siliconangle.md>)

Topics: [Software Engineering](<https://devfeed.tech/topics/software-engineering.md>), [telemetry](<https://devfeed.tech/topics/telemetry.md>), [AI-assisted coding](<https://devfeed.tech/topics/ai-assisted-coding.md>), [Agentic AI](<https://devfeed.tech/topics/what-is-agentic-ai.md>)

Tags: [agentic-ai](<https://devfeed.tech/tags/agentic-ai.md>), [agentic-behavior](<https://devfeed.tech/tags/agentic-behavior.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-risk](<https://devfeed.tech/tags/ai-risk.md>), [apis](<https://devfeed.tech/tags/apis.md>), [application](<https://devfeed.tech/tags/application.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [audit](<https://devfeed.tech/tags/audit.md>), [auditability](<https://devfeed.tech/tags/auditability.md>), [automation](<https://devfeed.tech/tags/automation.md>), [autonomous-ai](<https://devfeed.tech/tags/autonomous-ai.md>), [generatinve-ai](<https://devfeed.tech/tags/generatinve-ai.md>), [m-touheed](<https://devfeed.tech/tags/m-touheed.md>), [news](<https://devfeed.tech/tags/news.md>), [software-testing](<https://devfeed.tech/tags/software-testing.md>), [the-latest](<https://devfeed.tech/tags/the-latest.md>), [top-story-6](<https://devfeed.tech/tags/top-story-6.md>)

### AI overview

Agentic workloads challenge assumptions behind conventional software systems: tasks can run for minutes, retries consume metered compute, and results may not be reproducible. The article recommends defining output acceptance criteria, limiting retries, recording inputs and execution details, monitoring quality and cost per completed unit, assigning accountable reviewers, and testing model updates before production deployment.

### Source excerpt

Most traditional enterprise systems were built around three assumptions: Jobs finish quickly, retrying one is free and the same input always produces the same output. Agentic workloads break all three of these expectations, which is why pilots that perform well turn into operational problems once they run unattended. The difficulty is rarely the model; it's [...] The post Agentic workloads break assumptions about software testing. Here's how to cope appeared first on SiliconANGLE.