Baze

Formerly FactChat. One account for 80+ AI models, plus agents for research, slides, images and video. We launched it for universities as FactChat in March 2024, then renamed it Baze in October 2026 and opened it to everyone from ₩5,900 a month.

Role
Product & Frontend Lead
Team
A small team I lead
Timeline
Mar 2024–now
Stack
Next.js · logician-ui
Outcome
750K users · 100+ universities
Links
baze.co

TL;DRPlaceholder: the product, your role and the result, in two lines.

Baze’s model picker: Super Agent at the top, then every model grouped by provider: OpenAI, Claude, Gemini, Grok, Perplexity, DeepSeek and more.
One account, every major model: the model picker in Baze.

The bet

Placeholder: the bet. Why one account for every model, and why universities first?

More in Team chemistry is something you can design.

Four separate requests (search, image, compare, PPT) beside one request that runs them all, for fewer credits. One request, a trip to Sydney as a short video and a slide deck, fans out to web search, video and PPT, all done.
Super Agent: one request instead of four. It searches, makes the video and the deck, and picks the model for each step.

The call

Placeholder: the hardest call. The rename is a candidate: FactChat for institutions becoming Baze for everyone at ₩5,900 a month.

FactChatrenamed Oct 2026
ChatsLecture notesGrant draftCode reviewReading list
Summarize this week's lecture in five bullets.
GPT-6 AstraHere are the five points from Tuesday's lecture…
Ask anything
A schematic, not a screenshot.
The rename, performed: FactChat becomes Baze.

The result

From 28 universities in October 2025 to more than 100 a year later, and 600+ institutions in all. 750,000 people have used it. In September 2026 they ran about 48 times as many tokens as a year before, and every university that rolled it out campus-wide renewed. Figures from Mindlogic's public releases, October 2026.

A row of slide decks in different styles: a travel budget, Field Notes, Space Form Light, Common Objects, a portfolio.
Sample decks from the presentation agent, each from one conversation.
NextThe writing