Blog
Practical writing on productivity, AI, and building software.
The most reliable multi-step agents aren't the smartest — they're the ones handed structured places to stash state, permissions, and checkpoints. Here's how to build those places.
Most RAG hallucinations aren't model failures — they're retrieval failures. Here's a breakdown of chunking, recency, and permission bugs, plus fixes you can defend in code review.
Permissions decide what an agent may touch. Isolation decides what happens when it gets tricked. Here's how to design agent access like a network DMZ.
A practical taxonomy of AI agents delegation built on reversibility, verifiability, and blast radius — five tiers from 'let it run' to 'don't, ever.'
Stop writing clever prompts and start writing employment contracts. A concrete walkthrough of scoping a Tamaton custom agent with permissions, escalation thresholds, and an audit trail.
Identical agents don't cooperate by default — they clash, collude, and coordinate in ways nobody specified. That's an architecture problem, and it has fixes.
Federated training lets models learn across organizations without pooling sensitive data. Here's what that shift means for privacy-preserving AI and knowledge work.
"Humans lead agent success" is a polite way of saying context engineering. Here's why a unified workspace beats a pile of disconnected tools as RAG substrate.
Prompt and model tweaks that pass every smoke test can quietly wreck real outcomes. Here's how to evaluate agents on results instead of vibes.
Public benchmarks won't tell you if a model can triage your inbox or reconcile your spreadsheet. Here's how to build a private, brutally specific eval set in an afternoon.
A chatbot returns a response. An agent decides and executes a sequence. Confusing the two is why so many 'agentic' rollouts quietly stall out.
Picking a single LLM for every task leaves capability and money on the table. Route by task instead: cheap triage, strong drafting, dedicated verification.
Get started
Claim your address before someone else does — free to start, with an AI-native inbox built in.