Technical essays by Omar U. Espejel on AI engineering, crypto, verifiable systems, and production engineering.https://gradiente.ai/https://gradiente.ai/essays/half-of-erc8004-reputation-is-one-agent/https://gradiente.ai/essays/half-of-erc8004-reputation-is-one-agent/I read all 558,094 reputation events ever written to ERC-8004. 300,692 of them (53.9%) are about one agent: a mining coordinator, rated by the 57 wallets that mine for it, score 1 every time. It stopped writing them on 17 August 2026. On Ethereum, 145 agents out of 50,127 have a review from anything that is not a crawler. I wrote the Starknet integration for this standard, so this is a post-mortem of something I shipped on.Tue, 18 Aug 2026 09:00:00 GMThttps://gradiente.ai/essays/the-environment-contract-is-part-of-the-product/https://gradiente.ai/essays/the-environment-contract-is-part-of-the-product/A postmortem on an auth flag regression in an agent-built app, and the deploy contract that stops stale environments from becoming releases.Sun, 10 May 2026 00:00:00 GMThttps://gradiente.ai/essays/tablero-the-boundary-between-proofs/https://gradiente.ai/essays/tablero-the-boundary-between-proofs/A guide to the missing receipt around proof artifacts: how proof bytes, statement meaning, verifier domain, and replay assumptions become one accepted object.Thu, 07 May 2026 09:03:00 GMThttps://gradiente.ai/essays/proof-validity-is-not-statement-validity/https://gradiente.ai/essays/proof-validity-is-not-statement-validity/A proof can verify while the AI claim around it is still relabelable. In a local EZKL-style receipt test, the proof-only path rejected 1 / 7 relabels; the statement-envelope path rejected 7 / 7.Thu, 07 May 2026 09:02:00 GMThttps://gradiente.ai/essays/where-transformer-proof-pressure-appears/https://gradiente.ai/essays/where-transformer-proof-pressure-appears/The useful transformer-proving question is not whether the model contains arithmetic. It is where proof plumbing gets reused or repeated.Thu, 07 May 2026 09:01:00 GMThttps://gradiente.ai/essays/why-transformers-fit-starks/https://gradiente.ai/essays/why-transformers-fit-starks/Transformer decode repeats a state transition over carried context. That makes it natural to express as a STARK trace, even though proving the difficult operations efficiently still requires careful design.Thu, 07 May 2026 09:00:00 GMThttps://gradiente.ai/essays/why-documentation-structure-fits-agent-workloads/https://gradiente.ai/essays/why-documentation-structure-fits-agent-workloads/How curated manifests, canonical metadata, crawler boundaries, and simple health checks make documentation easier for AI agents to retrieve and cite.Tue, 21 Apr 2026 00:00:00 GMT