Home › Blog
Long-form essays on the practice of multi-agent AI: how a debate is structured, how to read what comes out, where the trade-offs live, and what's coming next. New essays roughly every two weeks. All written by Secan98.
A foundational essay on why a single confident answer hides the disagreement that hard questions deserve, and why putting several models in the same room is the smallest fix.
How a multi-agent debate is actually structured: parallel fan-out, round-robin critique, a small judge, threshold exit, and a synthesizer that preserves minority dissent.
A hands-on guide to running three local models in a debate on your own machine. Hardware math, model selection, latency, and when to reach for the cloud.
Why an AI app that stores your model-provider keys server-side is a liability waiting to be discovered. The OS keychain is the only defensible default.
An agent that can only talk is a chatbot. The Model Context Protocol turns 'AI assistant' from a chat metaphor into something that reads files, runs tests, and opens pull requests.
What convergence percentage actually measures, how to spot superficial agreement, and how to find the load-bearing claim every synthesis really rests on.
Three patterns you can use tomorrow at your desk: spec-review, code-synth, pair-debug. Setup, worked examples, honest limits, and back-of-envelope cost.
What has shipped, what's coming in Q3 and Q4 2026, what lands in early 2027, and four things we will explicitly not build, with reasons.
A field-notes essay on three task families where frontier models reliably take different positions, and what that means for picking your debate roster.
session