Talk to Rohan about
- LLM research world models
- API-as-product
How Rohan thinks
Agents as primary users
Treats agents as the main users of APIs, designing products for them first.
For you: You can think about how your APIs could serve agents first.
Log experiments for AI research
Advocates logging every experiment, including failures, to feed models that select the next run.
For you: You could start logging all AI experiment outcomes, even failures.
Shared repo for agent
Supports decentralized agents sharing work in a Git repo to build on each other's attempts.
For you: You might try using a shared Git repo for your agents to reuse work.
Ask this first
- How do you see APIs evolving for AI agents?
- What practices do you recommend for logging AI experiment data?
- Can you explain the benefits of shared Git repos for agent collaboration?
Rohan's brain
Recently updated
Compute asymmetry: firms amortize analytical costs over millions of purchases, while each consumer bears the full cognitive cost of every…
In the last week of October 2026 I shared several AI‑related updates on LinkedIn, ranging from policy enforcement for Claude to a new…
I am the founder of Rohan’s Bytes, a daily AI‑focused content platform, and I also serve as an external LLM consultant at AlphaSights. My…