one subscription · one wallet · twenty-plus tools
Vantage bundles focused AI tools — resume scoring, document parsing, travel research, PDF chat, and more — behind a single subscription. A wallet of credits meters exactly what you run. Nothing else.
live metering — every run is quoted, held, and settled on a ledger
Most AI products want their own $20/month. Vantage flips it: subscribe once, get a monthly grant of credits, and spend them across every module in the catalog — ten credits on an email today, two thousand on a full travel research job next week.
Pick a plan. That's the last pricing decision you make this month.
Your wallet fills on renewal. Grants expire monthly; top-ups roll over.
Every module quotes its cost before it runs. No surprises on settle.
Twenty-plus products sounds like twenty codebases. It isn't. Every module in the catalog executes on one of five archetypes. Pick one below and watch it run.
The workhorse. A form goes in, a versioned prompt template shapes it, structured output streams back in seconds. Most of the catalog lives here.
Upload a PDF, invoice, or contract. A background worker parses it page by page, pulls out the fields that matter, and hands back clean structured data you can export.
Your document gets chunked and indexed a single time. After that, every question retrieves only the passages that matter and answers with page-level citations — which is why messages cost a fraction of the ingest.
The heavy hitter. You state a goal; an agent loops through live web search, reading, and reasoning until it has a cited report. A hard budget caps the loop, so it can never run away with your credits.
Long-running conversations with memory. The session keeps what matters — your goals, your characters, your streaks — and compresses the rest, so week three feels like the coach was actually there for weeks one and two.
A module declares its engine, form, price, and model policy — once, in the registry. At runtime your app sends only the module id and inputs; the platform resolves the engine from the registry, renders the UI from the schema, quotes the credits, routes the model, and meters the run. New product? New row in the registry.
// registered once, server-side
{
"id": "resume-ats",
"engine": "transform", // declared here — and only here
"input_schema": { "resume": "file", "job_description": "text" },
"pricing": { "mode": "flat", "credits": 125 },
"model_policy": { "default": "standard", "premium": "opus" }
}
// all a client ever sends
POST /v1/runs { "module": "resume-ats", "inputs": { … } }Every run follows the same contract: you see a quote, credits go on hold, the run is metered, and you're charged the actual — never more than the quote. Failures refund themselves. Try it below; the ledger on the right is exactly what our database does.
this demo persists locally — refresh the page and your wallet and ledger are still here. the real app treats your browser the same way: inputs, drafts, and run history live on-device first (localStorage/IndexedDB) and sync through the API when it's reachable.
Every module shows its cost before you commit. Metered jobs quote their maximum.
Holds capture actual usage. If a job uses less than quoted, the difference comes back.
A run that errors releases its full hold automatically. You pay for outcomes.
Monthly plan credits reset with your cycle. Credits you buy separately roll over.
Twenty-two modules at launch across five categories — and because a module is a manifest, the catalog grows weekly. Vote on what ships next from inside the app.
One number to care about: credits per month. Every plan reaches the full catalog.
need more mid-month? top-up packs: 2,500 credits / $6 — they never expire.