Why we built the Brand Brain instead of another LLM wrapper

Most AI marketing tools are a prompt with a UI on top. We took the harder path: a per-customer model of voice, governance, and goals that compounds with every approved draft.

EngineeringApril 14, 2026 · 12 min read

Shipping LangGraph in production: lessons from a 10-node agent

When the agent is allowed to write to your customer's LinkedIn, every state transition matters. How we wired idempotency, audit, and crisis pause into a graph that never publishes without consent.

PlaybookMarch 30, 2026 · 6 min read

The 30-second content plan: how the agent decides what to ship next

From signals to slot picks: a transparent walkthrough of how Amelytic produces a 14-day content plan that gets approved 4× more often than the agency drafts it replaces.

StoryMarch 18, 2026 · 5 min read

Approval as the only human gate

Why we removed the second “publish” click. A study of 30 brands and the operator-time it freed up.

EngineeringMarch 4, 2026 · 9 min read

Voice extraction without fine-tuning: how the Amelytic Agent reads your last 90 days

A short tour of how we score voice on five axes from a corpus of past posts and a 200-word brand brief — and why fine-tuning was the wrong tool for this job.

Subscribe to the weekly digest at [email protected]

Fieldnotes.

Why we built the Brand Brain instead of another LLM wrapper

Shipping LangGraph in production: lessons from a 10-node agent

The 30-second content plan: how the agent decides what to ship next

Approval as the only human gate

Voice extraction without fine-tuning: how the Amelytic Agent reads your last 90 days

Build with us.