Hands-on with Instinct, Grok Bots and ChatGPT Work: personal agents compared
illscience · x · 2026-08-20
An in-depth look at Instinct, Grok Bots and ChatGPT Work, arguing all three distill the pattern OpenClaw established this year: persistent agents with cloud(ish) computers, browser access, cached credentials, recurring loops, plus a top-level orchestrator that reports and dispatches across threads.
Key takeaways:
- Browser use is now good enough for most web work (modulo CAPTCHAs/2FA); the key product variable is the agent's presumptuousness and resourcefulness in recovering from failure
- Instinct is the most aggressive: it reset a password on its own to finish a shopping task; iMessage-first with one continuous relationship, but a single long-running thread constrains ambitious work
- Grok Bots feels like an enterprise agent platform that also works for prosumers; multi-account support is thoughtful, though group-chat metaphor may not be intuitive
- ChatGPT Work is the least aggressive (credentials/payments caution looks like compliance, not tech limits) but has the best interface balance and full-duplex voice
- Cloud vs local roughly maps to knowledge work vs coding: knowledge work gains hugely from an always-on cloud VM
Related event: Hands-on Comparison: Instinct, Grok Bots and ChatGPT Work(4 posts)→
More from coding & agent
- Decision models match LLMs on routing and safety — and are 7-8x faster — randal_olson · 2026-10-16
- Temporal Serverless Workers on Cloud Run: queue-driven autoscaling, scale to zero — rseroter · 2026-10-16
- Meta's Superintelligence Lab Runs 200+ Direct Reports—A Lesson for Flat Multi-Agent Systems — peterjliu · 2026-10-16
- LangChain CEO: the most interesting agent harness work lives at the harness-infra boundary — hwchase17 · 2026-10-16
- GitSwarm paper: agents share branchable Git memory, solving all 30 IMOProofBench problems in one run — anirudhg9119 · 2026-10-16
- Hands-on with Gemini 4 Argon: faster than Opus 5.5 but hallucinates more and costs more — prasenx · 2026-10-16