Nicheloom visits: 45,709 all-time · 11,048 in the last 7 days

The opportunity tracker for new startups.

Today: 2,453 launches · Niche of the day: browser-based utility tools · See what launched today →

Harness²

Harness the agent. Then harness its self-improvement.

Get picks like this daily. Today's launch numbers, the top launches, AI and tech news, and a daily opportunity pick, straight to your inbox.

Read the full description

Harness² from Google Research is a plug-and-play recursive agent harnessing framework that lets LLM agents automatically evolve their execution harnesses (prompts, tools, memory) for open-world tasks—without ground-truth verifiers. Dual recursion: task-level Propose-Probe-Compose refinement + harnessing-level accumulation of reusable experience. Up to +11.5 points on professional benchmarks like WorkBuddy-Bench.

Visit website ↗ Source listing ↗

PH · launched Oct 20, 2026 · 167 upvotes (top 1%)

Similar launches

See all 660 similar launches →

Could you build this?

No It is described as a Google Research framework for recursive self-improvement of agent harnesses, which is novel research rather than an application pattern.

What it would actually take: Reproducing this means implementing the paper's Propose-Probe-Compose refinement and meta-level harness evolution, then evaluating it on open-ended tasks without ground-truth verifiers. That requires researchers who can design the search objectives and evaluation methodology, plus substantial LLM compute for experiments. Vibe coding can scaffold the code but not the research contribution.

Competitors

Other products that read as similar to this one : 660 launches clear the similarity bar, closest 8 shown.

Attention rank: #7 of 661 (itself plus its competitors, highest first, normalized so YC and Product Hunt are compared fairly).

Launched 356 days after the earliest competitor.

Other launches for this product