Personas
How the auto persona learns
Onboarding analyses your last 50 posts and 50 replies; every 50 new manual replies refine the model.
Initial assessment
When you install ThreadPilot, the extension reads your last 50 posts and 50 replies from your X profile (DOM-scraped — we never call the X API). An LLM produces a writing-style profile that becomes your auto persona.
Continuous learning
After onboarding, ThreadPilot watches the manual replies you write yourself on X. Every 50 new replies, your style profile is reassessed and your auto persona becomes a closer match to how you actually write.
We store the style summary, not the raw replies long-term. Writing samples are pruned after each assessment.
Did this article help?
If you’re still stuck, we’re one ticket away.