Answers worth quoting.
When askFinz answers a question, you can see where the answer came from. Cite it, share it, defend it.
askFinz isn't a single chatbot. It's a workshop — different rooms for different kinds of work, all sharing one memory of you and one way of doing things. A fleet of specialist agents on call, dozens of workspaces with the door open, the right brain picked for the question, and a paper trail you can defend.
Go deeper: guides · integrations · compare · models
workshop/
├── agents/
│ ├── researcher cites
│ ├── coder pairs
│ ├── writer drafts
│ ├── analyst charts
│ ├── summariser distils
│ ├── browser drives
│ ├── mailer triages
│ ├── trainer fits
│ ├── workflow schedules
│ ├── search retrieves
│ ├── socials posts
│ └── finance reads
├── workspaces/
│ ├── chat · research
│ ├── code · news
│ ├── search · data
│ ├── train · workflow
│ ├── socials · finance
│ └── … · more
├── models/
│ └── reasoning · chat ·
│ vision · code ·
│ long-ctx · audio
└── reach/
├── extension
├── desktop
├── workflow
└── apiWhen askFinz answers a question, you can see where the answer came from. Cite it, share it, defend it.
Your work isn't all the same. askFinz pairs you with the right agent for what's in front of you — and routes between them when the work needs more than one.
Different surfaces for different work, but a single home, a single login, and one consistent way of moving between them.
Sensitive work runs in its own room. We make this easy to verify.
If something is on the roadmap, we say so. If we change our minds, we'll write about it. We won't sell you what we haven't built.
Take your data with you, plug askFinz into your own tools. No quiet lock-in.
Every agent run takes the same shape, whether it's answering a question in chat, drafting a literature note, refactoring a service or sweeping across the news. You can watch each stage land. You can override at any stage. You can rewind any stage and re-run it without losing what came before.
The diagram on the right is alive — it walks through a single run on scroll-in. Watch the Run stage hand off to the satellite agents mid-pipeline; that hand-off is automatic, and you see who took the sub-task.
The long write-upThe model picking that happens at step 01 is its own discipline — see /models for how that decision gets made.
A featured slice of the roster — one job rarely sits with one agent. A research note pulls in the Writer for the draft, the Researcher for the sources, the Browser for the live check, the Analyst for the chart. Work routes between them automatically — you watch the hand-offs, you keep the editorial call.
The full rosterSynthesises web pages, PDFs, prior notebooks and live feeds — every claim attached to the page it came from.
Pairs in a full IDE — refactors, generates tests, runs the suite, writes the design doc next to the diff.
Long-form drafts, memos and briefs that hold their structure across thousands of words.
Explores data, runs regressions and builds the chart that actually answers the question.
Distils long meetings, long documents and long threads down to what someone needs to know in two minutes.
Drives a sandboxed live web session — fills forms, finishes sessions, returns the artefact with screenshots.
Triages the inbox, drafts the replies, schedules the sends — your voice, your timing.
Runs fine-tunes and eval sweeps end to end; you watch the curves, you ship the result.
Turns a sequence of moves into a visual pipeline that runs on a schedule — triggers, retries, the AI step are nodes on a canvas.
askFinz's own LLM-friendly search engine — clean answers with sources, ready to drop into a draft.
Drafts and schedules across multiple accounts — one voice per channel, the calendar in one place.
Portfolio-aware drafting and analysis — briefings, notes and charts that know what you hold.
The workshop has a room for the day-to-day, a room for serious writing, a room for code, a room for the news, a room for the numbers — and a long tail of specialist rooms for the work that doesn't fit anywhere else. Each one has its own door, its own tools, its own memory of what you did last time. The hallway between them is what makes it a workshop and not a tab graveyard.
Browse every workspaceWe work across model families instead of crowning one. Each family earns its turn on the kind of work it's actually best at — and we tell you which one wrote each answer, in a small badge you can click through.
Routing is intent-aware but never invisible. Pin a family per workspace, override per turn, or hand us a key for a model you already pay for.
Want even more choice? The browser extension adds 380+ more AI models, ready to use right alongside the cloud ones.
How we chooseThe slow, careful brains. Used when an answer has to defend itself.
Conversational, fast, broadly capable. The default for everyday turns.
Reads a screenshot, a chart, a whiteboard alongside the words.
Stronger at refactors, tests and reading unfamiliar codebases.
For when the question carries a hundred pages with it.
Read out a briefing, transcribe a meeting, draft a script meant to be heard.
Self-hostable families when residency and independence are hard constraints.
The same question gets a different shape in each room. Chat answers fast. Research cites four sources. Code drafts the diff. News converges on the three things that broke overnight. Pick the room that fits — the memory of you follows.
The mock on the right cycles between four workspaces every couple of seconds. Click a tab to pin it.
Reach the workshop from wherever you happen to be working. A side panel in the browser. A desktop client off the tab. A pipeline canvas for what should run on a schedule. A public API for the code you've already written. No quiet lock-in — take your data with you.
Request access and we'll route you to the workspace that fits the shape of your work first. If you'd rather read the long version before you reach for it, the manifesto and the trust write-up are linked below — both honest about what's in the room today and what isn't.