The Jettova API Is Open: Try It in Your Browser, npm Install It, or Hand It to Your AI Agent

A few weeks ago we introduced the Meet API and said it was coming to developers. Today it's simply open. You can try the fairness engine in your browser right now, with no key and no signup: go to our developer page, type in a few cities, and watch it rank where that group should meet. When you're ready to build, a free key takes thirty seconds and covers 500 calls a month. No sales call, no waitlist.

The playground is the fastest way to understand what makes this engine different. Give it Boston, Berlin, and Tokyo and ask for the least total flying, and it comes back with coastal Norway, because that genuinely is the flying-time middle of that triangle. Ask for the cheapest option for the whole group and it tells you the honest answer, even when that answer is one member's home city. Ask for the fairest, the option where nobody gets a brutal worst-case journey, and you get a different list again. These are three different questions groups actually argue about, answered with numbers instead of opinions.

The API has grown well past one endpoint. The deciding layer now includes the meet ranking itself, a batch endpoint that compares up to five scenarios in a single billed call (the current team versus the team plus two new hires, hubs only versus anywhere), a group-trip endpoint that folds in everyone's availability and returns the best travel window alongside the fairest cities, and a locations resolver that turns free text like NYC into the exact city and airport code every other endpoint accepts.

Around the deciding layer we've opened the practical trip toolkit. A visa endpoint answers the question that kills a fair destination in the real world: can everyone's passport actually get in without weeks of paperwork? One call returns each passport's entry status and a single group-ok boolean. An expense-splitting endpoint takes who paid what and returns the minimum set of who-pays-whom transactions, the same settle-up engine our travelers use after a trip. A packing-list endpoint returns a climate-and-activity-aware checklist as structured data. And a booking-links endpoint turns a decision into prefilled hotel, flight, and activity links, attributed to your key so bookings your users make are visible to you.

If you write TypeScript or JavaScript, the official SDK is one install away: npm install jettova. It's zero-dependency, fully typed, works in Node, Deno, Bun, and the browser, and surfaces errors the way you'd want them: unresolvable inputs come back as a list you can fix, rate limits tell you exactly how long to wait, and expensive calls support idempotency keys so retries never double-bill.

And if you're building with AI agents, we think this is the most interesting part. Agents are good at talking about travel and bad at computing it. No model can derive the fairest meeting city for six people from real flight-cost and travel-time estimates, or check five passports against a destination, or produce a minimum-transaction settlement. Our MCP server, jettova-mcp, gives any MCP-capable agent those exact tools with one line of configuration. Without a key it can even use the meet demo, so an agent can try the engine before you've set anything up.

One honest note on scope: the AI-generation endpoints, full itinerary generation, natural-language refinement, and planning from a video or photo, run real model inference per call, so they're enabled per agreement rather than on self-serve keys. Everything else, the fairness engine included, works on the free tier. Every figure the API returns is a directional estimate built for deciding, not a bookable quote; booking always completes with real prices on partner sites.

The deciding half of travel has never had public infrastructure before. It does now, and the fastest way to see it is to type your own group's cities into the playground and watch the argument end. If you build something with it, a Slack bot, an offsite tool, an agent, tell us. The roadmap gets shaped by the people actually building.