Hundreds of millions of taxi trips. One Mac. Ask freely on a flat plan.

Note: This is a public method demo. The shipping product is turbolapper - AFM (Apple Intelligence on-device; data stays on your Mac). See requirements and Technology.

Serious scale, still on your Mac

Wrestling hundreds of millions of trip rows used to mean warehouse tickets, cloud bills, and metered exploration. On a Mac with turbolapper - demo hardware was an M2 MacBook with 32 GB RAM - you can load, profile, and keep asking in hours of exploration, not weeks of queue time. Extra questions do not open a meter from us, because the analyst runs local. Airports, channel mix, broken timestamps - you can chase all of it without leaving the device.

Summary

• Data: New York City open taxi trips - hundreds of millions of rides depending on the cut.
• Method: Ingest, prep, ask operational questions on the Mac.
• Result: Scale that used to need a warehouse culture, now practical for private exploration and free follow-ups.
• You get: Better hygiene when looking is cheap; patterns when the second question is free.
• Next: Watch the session · Join the waitlist

What Local Analytics means at this scale

Spreadsheets die at this row count. Cloud exploration gets expensive the moment a team gets curious.

With turbolapper:

1. Serious data stays on your device while you explore.
2. Interactive answers as the corpus grows - see Technology for cold-start framing.
3. Flat-cost curiosity - turbolapper never meters question count, even when the table is huge.
4. Shipping product (AFM): Apple Intelligence on-device; data stays on your Mac.

What we ran

1. Downloaded public NYC taxi trip data.
2. Dropped it into turbolapper on a Mac (stage, profile, roles/grain) - M2 · 32 GB in the session.
3. Asked operational questions: airports, channel mix, quality issues.
4. Chased follow-ups the moment something looked wrong - because looking was free.

That is the opposite of “one dashboard, no more questions this sprint.”

What the free questions find

Session window patterns - re-cut anytime when curiosity is free.

When every query meters warehouse or AI spend, hygiene side quests die. When they are free on-device, you find broken timestamps, odd joins, and operational patterns you were not looking for. That is the unit economics of curiosity in practice - at a row count people used to call “enterprise only.”

Scale, still interactive

Answers should stay interactive on-device as data grows. Cold-start behavior and benchmarks live on Technology.

How to run this yourself

1. Stress-test on public scale before proprietary loads when you can.
2. Profile before chat: nulls, extremes, grain.
3. Document units and codes in prep.
4. Budget time for hygiene follow-ups - free questions are the feature.
5. Keep provenance so you can re-run as data refreshes.
6. Move the same method to private data when you are ready - still on-device.

FAQ

What does this demo show?

Local Analytics at real scale - hundreds of millions of open trips on a laptop, with room left to ask more. Watch: YouTube · Community.

Why emphasize free follow-ups on big data?

Big data is where people stop asking. Flat monthly turbolapper pricing and on-device processing make the second and third question rational again - that is how hygiene and insight both improve. Economics

What hardware ran the demo?

An M2 MacBook with 32 GB of RAM in the published session. Performance framing: Technology.

What should I do next?

Watch the session, read retail prep for trust and federal data for multi-series local-vs-cloud work, then join the waitlist.

Watch and go further

• Watch: NYC taxi session
• Related: Federal data · local vs cloud · Retail prep · Community
• Product: Waitlist · Technology · Economics