Talk to Your Data — Voice AI Data Scientist
OctOpus now talks. Tap the mic, ask your question out loud — 'forecast next month's revenue', 'why did that model fail?', 'which features mattered most?' — and the agent answers back in real time, speech to speech. It's the same autonomous AI data scientist, now with a hands-free conversational interface. Live voice is available to everyone, free to try.
How voice mode works
Tap the mic and start talking. OctOpus streams your speech to a real-time model, understands the intent, and responds out loud while it works — narrating the plan, the experiment it's running, and the result. You can interrupt, redirect ('focus on F1, not accuracy'), or just say 'go' to launch a full research run. When a low-latency realtime voice isn't available, OctOpus falls back to the browser's built-in speech so the conversation never breaks.
Why voice for data science
Data work is full of small questions that don't deserve a notebook — 'is this column leaking?', 'what's the holdout score?', 'rerun with class weights'. Speaking them is faster than typing, and hearing the answer keeps your eyes on the chart, the dashboard, or the meeting. Voice turns OctOpus into a data scientist you can think out loud with, hands-free.
The same agent, end-to-end
Voice is an interface, not a toy. Behind it is the full OctOpus research loop: profile the data, write per-experiment training code, rotate model families, diagnose failures, validate the winner on a holdout, and deploy a prediction API. Ask by voice, get a real, validated model — not just a spoken summary.
Key capabilities
- Real-time speech-to-speech — talk to OctOpus and hear it answer as it works.
- Hands-free: launch runs, steer experiments, and ask about results out loud.
- Automatic fallback to browser voice when realtime isn't available — the conversation never breaks.
- Same autonomous engine underneath: profiling, model rotation, holdout validation, deploy.
- Free to try — no credit card.
Frequently asked questions
Can I really talk to OctOpus?
Yes. Tap the mic and speak naturally — ask a question, describe a goal, or say 'go' to start a run. OctOpus understands the intent and responds out loud in real time, speech to speech, while it profiles your data and runs experiments.
Is the voice feature free?
Yes — live voice is currently available to everyone, including free and guest users, while we open it up for testing. You don't need a paid plan or a credit card to try it.
What can I do by voice?
Ask questions about your dataset ('is this column leaking?', 'what's the holdout score?'), launch a full autonomous research run, steer experiments mid-run ('focus on recall', 'try class weights', 'skip TabPFN'), and hear the agent narrate what it's doing and why.
What happens if real-time voice isn't supported?
OctOpus automatically falls back to the browser's built-in speech (Web Speech API) so the conversation keeps working. You still talk and listen — just turn-based instead of fully streaming. No setup required.
Does voice change what OctOpus actually builds?
No. Voice is just a faster interface to the same autonomous AI data scientist. Whether you type or talk, OctOpus runs the full loop — profile, write training code, run and diagnose experiments, validate on a holdout, and deploy a prediction API — and gives you the validated model, the train.py, and the metrics.