The West Coast Builders — coding-agent collection
Is Agent Harness relevant in 2026
Mayank Gupta 2 of 7
In this collection Browse 7 summaries 2 of 7
Modem co-founder and former Sentry VP of Engineering Ben Vinegar joins host Mayank Gupta to discuss vertical product agents, agent-friendly engineering, software vendors, harness experimentation, and building expertise over time.
Key Points Covered
- Modem is Vinegar's vertical product-agent thesis: He describes Modem as a cloud agent with institutional memory that can respond to triggers, work through chat surfaces, expose MCP access, and act as a sub-agent. These are first-party product claims, not capabilities independently demonstrated in the interview [00:39:15]-[00:43:34].
- Generated-code share depends on engineering practice: Vinegar reports that 99.9% of Modem's code is generated and argues that navigable code and isolated components help agents work effectively. The percentage concerns code generation, not productivity or quality, and the referenced experiments are not shown [00:47:21]-[00:50:35].
- Tool narratives reflect vendor incentives: He cautions that model vendors and influencers benefit from claims that general agents can replace software, while distinguishing low-stakes generated utilities from important tools his company prefers vendors to maintain [00:51:39]-[00:55:01].
- Finite context leaves room for specialization: Vinegar argues that agents cannot carry unlimited memory and that task-specific memory, skills, and evaluations can support vertical products. He says Modem spent thousands of dollars on evaluations, but supplies no results or comparison [00:56:38]-[00:58:31].
- Harness design remains open to experimentation: He points to continuing changes in coding-agent interfaces as evidence that harnesses are not settled, and characterizes Modem as a cloud harness that borrows coding-agent concepts without being a coding agent [00:59:20]-[01:04:10].
- Expertise requires staying with a problem: Vinegar advises builders to spend enough time in a problem area to develop distinctive knowledge, separating thematically related experiments from repeatedly chasing unrelated quick wins [01:04:59]-[01:06:39].
Full video: https://www.youtube.com/watch?v=wbsDZpVB4Lg(opens in a new tab)