Fixed-line routing
Declare the path work takes through agents and tasks once; every run follows the same drawing.
Work infrastructure
Lay the pipe once. Let the work flow.
Pypeo connects triggers, agents and services into pipelines that keep running after you close the tab. Every job is metered in power tokens against a balance you top up — no plans, no renewal cycle.
runs / r_7c41 — invoice-chaseflowing
120+ service fittingsAnthropic · OpenAI · Gemini · xAIAppend-only ledgerREST API on every pipelineAES-encrypted credentials
Installation
Four fittings between a trigger and a result
1
Source
Pick the inlet: a schedule, a webhook, an RSS feed or a plain API call.
2
Assemble
Join agents and tasks in sequence. Each joint declares its model and the services it may touch.
3
Pressurise
Run it. The engine reserves tokens up front and streams progress joint by joint.
4
Account
Every debit lands in the ledger with the balance after it; failed runs return their reserve.
Flow test
One pipeline, followed end to end
The exact sequence a production pipeline goes through, condensed into five chapters.
I
Inlet Collector
claude-sonnet-5Pull every overdue invoice off the books
F
Flow Processor
gpt-5.2Draft firm, polite chase reminders
L
Line Dispatcher
claude-haiku-4.5Queue each reminder for delivery
01/05 Three agents are bolted in line: a collector, a processor and a dispatcher.
The fittings
Standard parts for non-standard work
Six fittings cover most installations. All of them draw from the same wallet and write to the same ledger.
120+ service fittings
Slack, Telegram, RSS, webhooks and raw HTTP thread straight onto any section of the line.
Model per section
Choose Anthropic, OpenAI, Gemini or xAI for each agent. Your own keys, held encrypted at rest.
Scheduled pressure
Cron-style schedules keep the line moving without anyone standing at the valve.
Metered accounting
An append-only ledger books every token that passes, with the balance after each movement.
External couplings
Every pipeline exposes a REST endpoint secured by hashed keys, billed from the same balance.
0+
service fittings on the shelf
0+
pre-drawn pipeline layouts
0%
of failed runs refunded by reserve release
No
subscriptions — the meter is the whole bill
Metering
Pay for throughput, not for access
Top up power tokens; pipelines draw them down as work flows. The table below is the entire price list.
Starter
€10
All 6 model providers · Sequential process · Built-in tools & actions
Maker
€25
widest boreHierarchical orchestration · Public API + 3 keys · Crew memory & templates
Pro
€50
Everything in Maker · Priority queue lane · Scheduled triggers
Scale
€120
Unlimited API keys · Priority queue lane · Low-balance alerts
Studio
€250
Everything in Scale · Team invites · Best per-token rate
Larger top-ups buy tokens at a lower unit rate. Balances never expire.
FAQ
Fitting questions
The single metering unit for the platform. Model usage, connector calls and engine time are all converted to tokens, so one balance covers every pipeline.
Never. There is no subscription and no auto-renewal; the only ways the balance changes are a top-up you make and a run you start.
The engine reserves tokens before opening the flow and releases the reserve if the run stops. Failures do not bill — the ledger shows the reserve going out and coming back.
Yes. The ledger is append-only and every row records the balance after it, so the account history is a complete, replayable record.
Yours. Add keys for Anthropic, OpenAI, Gemini, xAI or any OpenAI-compatible endpoint; they are AES-encrypted at rest and decrypted only inside the run engine.
Every pipeline has a REST endpoint. Keys are stored as hashes, can be revoked instantly, and API runs meter against the same balance as manual ones.
Put the first line in the ground.
Sign up, take the starter tokens and have work flowing through a pipeline today.
Get starter tokens