Why MeowByte
Most AI chat products pick a model for you and stop there. That's fine for a quick question. It's a worse fit for a decision you actually want checked - where a second opinion, or a third, would change what you do next.
MeowByte started from two annoyances: being stuck with one lab's model regardless of whether it was the right one for the task, and wanting more than a single confident-sounding answer when a decision actually mattered.
What that turned into
Personalities instead of a fixed assistant
A personality is a name, a system prompt, a model, and a few parameters - routed through OpenRouter, so it can be any model worth using, not whichever one a vendor shipped by default.
The Council instead of one answer
For decisions worth debating, seat a few personalities together. They restate the question independently, analyze it, challenge each other's reasoning, and converge - with the full transcript kept, not just the final line.
How it's built
Entirely on Cloudflare's platform - Workers, D1, KV, R2, Durable Objects, and Queues - with OpenRouter as the single gateway to every model, and Stripe for billing. No separate backend infrastructure to run or scale.