@garrytan: Harnesses from labs have an incentive to burn tokens, which means harnesses from startups have a real utility, like how…
Summary
Garry Tan argues that AI agent harnesses built by startups have real utility because they can convert token-burning agent behavior into deterministic, tested, repeatable code, highlighting Grep's use of agents to automate compliance alert reviews at UK brokerage IG.
View Cached Full Text
Cached at: 10/05/26, 05:30 PM
Harnesses from labs have an incentive to burn tokens, which means harnesses from startups have a real utility, like how Grep can automatically swap token burn into deterministic tested code that is repeatable just by observing agent use
AJ Asver (@_aj): We’re building self-improving agents to automate tedious, repetitive operational work @GrepdotAI. At IG, a UK online brokerage, Grep agents review compliance alerts, helping over 80% of traders onboard without manual intervention. IG is now a FTSE 100 company.
Operational work
Similar Articles
@ycombinator: Harnesses often get dismissed as just scaffolding, just prompt engineering, and not real research. But that couldn't be…
Y Combinator discusses the importance of harnesses in AI, highlighting their role in improving model performance, self-improving agents, and real-world applications such as personal AI and work automation.
@turingbook: The greatest potential value of Harness (Agent) may be going into the real work scenarios of seasoned experts across industries, observing how they work and think, and recording and "distilling" their valuable experience—data that was previously largely tacit.
The tweet highlights the potential of agent harnesses to capture and distill experts' tacit knowledge as new training data, referencing a Berkeley AI Summit talk by Jianfeng Gao on agentic modeling as an emerging AI paradigm.
@HarryStebbings: Why harnesses need to be neutral “You want to be able to switch models. The harness can be very tied to the model, or i…
Harry Stebbings shares a quote arguing that AI agent harnesses should be model-neutral to avoid lock-in, since the harness handles data, APIs, and memory. He also praises investor Miles Grimshaw's early bet on LangChain via Benchmark.
@garrytan: Giving your harnesses access to a real browser with your real credentials is a real unlock - highly recommend @AsideAI
Garry Tan recommends AsideAI for giving AI coding harnesses access to a real browser with real credentials, calling it a significant productivity unlock.
best of the best agentic harnesses do this…
The author shares insights on building effective agent harnesses: the best ones minimize LLM reliance for trivial tasks and reserve LLMs for complex reasoning, distinguishing genuine harnesses from simple wrappers.