What does High Agents actually do?
Four things on a daily loop: (1) learns your writing voice from tweets you paste, (2) generates a market-intelligence brief from your topic + curated sources, (3) turns the brief into 40 ranked thread headlines, (4) writes drafts in your voice on the headlines you pick. You copy the drafts into X. ~10 minutes a day.
Do I need an existing X audience to use this?
No. Voice training works off any tweets you paste — yours, or accounts you want to write like. Bootstrap accounts use this exactly that way.
Do you auto-post to X?
Not yet — currently you copy the generated drafts and paste them into X yourself. That keeps the loop fast and gives you a final read-through before anything goes live. Direct auto-posting is on the roadmap.
Does it hallucinate?
It can. We work hard against this with research-grounded prompts, anti-fluff filters, and explicit reminders not to invent CVE numbers / stats / quotes. Always read each draft before posting. The tool drafts; you ship.
Which LLM does it use?
Claude (Anthropic) for voice profiling, research synthesis, and draft generation. Research briefs use Claude's built-in web search. We bring our own keys; you don't need an LLM account.
Can I use my own data sources?
Yes. Add RSS feeds or GitHub markdown URLs (awesome-lists, READMEs) on the Setup page. The brief weighs them alongside web search.
What if the drafts don't sound like me?
Re-train the voice profile with more tweets — 20-30 representative ones is the sweet spot. Filling in "About you / Audience / Goals" on Setup makes a noticeable difference too.
Free trial?
Yes — sign up free and run the full loop (voice training, brief, ideas, drafts) on the starter tier. Upgrade when you hit the daily-generation limit.
Can I cancel anytime?
Yes. Cancel from your account settings; you'll keep access until the end of your current billing period.
Is my data private?
Your tweets, topic, and generated content are scoped to your account via row-level security in Supabase. We don't share or train models on your data.