FanSim generates synthetic fan agents from your real audience segments and lets you A/B test copy, visuals, pricing and timing against a statistically modeled fan population. Decide with evidence, not guesswork.
FanSim is built and operated by bemyfriends — the team that has run fandom businesses for 1,000+ artists and IPs over more than 10 years. Its models are trained on bemyfriends’ hands-on know-how, proprietary data and real campaign cases, so they understand the artist and the fan at the same time.
Every candidate is judged by individual agents — not an average.
Copy, subject lines, announcements
Thumbnails, covers, photocards
Tickets, goods, membership tiers
Best release & on-sale windows
Stress-test sensitive moves for backlash
Which win-back offer retains at-risk fans
Turn segments into a population of statistically modeled fans, each with demographics, spending, affinity and churn signals.
Run live A/B tests and watch each agent react in real time, with per-segment winners and revenue estimates.
Channel intelligence grounded in real public data — including YouTube — with sentiment, desires and frustrations.
Qualitative, in-character reactions and 1:1 interviews with synthetic fans for the nuance numbers miss.
Record real results and feed them back so predictions sharpen against your fandom over time.
Generate, score and iterate messaging automatically until the winning variant emerges.
Ground every analysis in your own research, reports and brand guidelines via semantic retrieval.
Choose Claude, Gemini, OpenAI or MiniMax — switch engines and model versions from the admin.
Ask up to 1,000 virtual fans (agents) anything about your artist’s activities and get ahead of the response before you act — or sit down 1:1 with a single fan for focused, in-character analysis.
Pose a question or plan to up to 1,000 fan agents and read the room — in-character reactions, stance, and a synthesis you can act on.
Chat one-on-one with a single fan persona for deeper, focused insight into the “why” behind the numbers.
Add a fandom and let AI draft 3–5 fan segments with demographic and behavioral distributions.
Sample a synthetic fan population — hundreds of agents that genuinely reflect each segment.
A/B test your candidates, read per-segment results, and launch the winner with confidence.
FanSim continuously checks live public signals across channels and grounds every analysis in what’s happening right now — so recommendations stay in step with the trend, not last month’s snapshot.
YouTube · X · Weverse · Communities · News
Real-time fetch + grounding in 10 years of fandom data
Channel intelligence, simulations & strategy
FanSim is in private beta with a small group of partners. Have an access code? Sign in. Otherwise, get in touch and we’ll set you up.