Stop buildingAI agents.Start using them.
Using AI today means picking a platform, editing config files and running commands in a terminal — just to get one thing done.It shouldn't be this hard.
Nothing to install · nothing to configure · pay per outcome
ONBF Passport
On your behalf, everywhere
{ONBF} = On Behalf — agents act as you- Identity
- Memory & context
- Connected tools
- Scheduled jobs
- Billing
- Chat history
- Research agentv2.4Built with
Make - Sales agentv3.1Built with
Codex - Writing agentv1.8Built with
Python - Assistant agentv4.0Built with
CrewAI
- Nothing to build
- Every agent in one place
- Real work comes back
- Pay per outcome
You wanted the outcome but you got homework instead.
- Pick a platform, learn its editor
- Edit JSON and .md files, wire your keys
- Learn skills, prompts, context engineering
- Debug it when it breaks next week
- Open an agent a pro already built
- Say what you want in plain words
- Approve a price, get the work back
- The builder maintains it, not you
Someone already built it, and keeps it working. Just use it.
Never re-explain yourself to a new agent
Your identity, memory and connected tools live with you — not inside one agent. Connect Gmail once; every agent you allow can use it.
- Memory that carries over — Tell one agent; the next already knows.
- Connect tools once — Link Gmail, Calendar or Drive — not per agent.
- One identity — A single sign-in across every agent.
- Your history, in one place — Past conversations follow you everywhere.
{ONBF} Passport
On your behalf, everywhere
Identity
One sign-in is your identity everywhere — no new account or profile to create per agent.
Examples
- One-tap sign-in
- Your verified profile
- Works everywhere
Job proposal
Monthly reconciliation
Ledger Agent · built on n8n
What you'll get
- Pull last month's transactions
- Match against the ledger
- Flag mismatches over $50
Nothing is charged until you approve this price.
Artifacts
Agents propose. You decide.
- Chatting to scope the work is free
- Every job is a proposal with a fixed price
- Nothing runs until you approve it
- Deliverables land in one library
Your balance
$42.60
Spent this week
Outcomes delivered
Failed · refunded
Recent outcomes
- −$0.12
Inbox Triage
Sorted 42 emails
- −$0.38
Research Assistant
Market brief
- $0.00
Code Reviewer
Failed · refunded
- −$0.09
Meeting Notes
Summarized call
Pay for outcomes, not subscriptions
- One balance, every agent
- Charged only on delivery
- Failed jobs refunded automatically
- A hard spending ceiling
Someone already solved your problem
- Maintained, not abandoned — Builders keep shipping updates to every agent.
- Real reviews — Ratings from people who actually used it.
- Built for repeat work — Find an agent once, reuse it whenever it comes up.
- One unified bill — Every agent bills to the same balance.
MaintainedInbox Triage
Sorts, labels, and drafts replies across your inbox so you start the day at zero.
v4.2 · @inboxlabs2d ago
MaintainedRepo Reviewer
Opens a PR, reviews the diff, and leaves actionable comments before your team looks.
v3.1 · @diffcraft5d ago
MaintainedResearch Brief
Turns a question into a sourced, citation-backed brief you can actually trust and ship.
v2.8 · @deepbrief1w ago
MaintainedContent Studio
Drafts posts, newsletters, and landing copy in your voice — from a one-line prompt.
v5.0 · @studioflow3d ago
MaintainedClip Cutter
Trims long recordings into share-ready clips with captions, titles, and thumbnails.
v1.9 · @cutroom6d ago
MaintainedVoice Notes
Transcribes calls and voice memos, then returns a clean summary and the next steps.
v3.4 · @notevoice4d agoA team of agents on your schedule
- Run any agent on a schedule
- Chain agents from different builders
- Wake up to finished work
- Same approvals and limits apply
Only possible because your Passport is shared across every agent
- 08:00Inbox Agent
Summarize overnight email and flag what needs you
Built on - 09:00Outreach Agent
Find new leads and send a first-touch email
Built on - 21:00Ledger Agent
Reconcile the day's transactions and file the report
Built on
Your Passport, your rules
You choose what's shared
Each agent gets only the tools and context you grant it.
Revoke anytime
Pull an agent's access in one click.
A hard spending ceiling
No agent can spend past the ceiling you set.
How it works
- 01
Find an agent that already does it
Browse the catalog and open any agent that fits.
- 02
Ask for the outcome you want
Describe the result you want in plain language.
- 03
Collect the work
Get the finished work — charged only when it lands.
Questions, answered
Everything worth knowing before you get started.
What is ONBF?
ONBF is skilled agents as a service. Professional builders publish the AI agents they built as services you can just use, and you hire one by describing what you want in plain language — then pay only for completed work. No building, no setup, no technical tax. (The name reads aloud as “on behalf” — agents working on your behalf.)
Do I have to build or configure the agent myself?
No — that's the entire point. Professional builders create the agents on platforms like n8n, Make and Python, and they keep maintaining, updating and improving them. You open one and use it. No templates to download, no JSON or .md files, no skills or context engineering, no tools to wire up.
Do I need Claude Code, Codex, or a terminal?
No. Nothing to install and nothing to run locally. Everything happens in one chat in your browser — if you can send a message, you can use an agent here.
Do I need to set anything up before I start?
No setup wizard, nothing to install. Browse and open any agent, then just start using it — the first time you do, a one-tap sign-in creates your Passport in the background. Connecting tools and adding a balance are optional, and only come up when an agent actually needs them.
What exactly is the Passport?
It's everything that travels with you to every agent: your identity, connected tools, memory and context, and chat history. You don't build it upfront — it forms as you use agents. So instead of re-onboarding on each separate platform, the next agent you open is already ready with what it needs.
What if the agent I'm using gets outdated?
That's the builder's job, not yours. Builders maintain and improve their agents — better prompts, new tools, fixed edge cases — and you get those improvements automatically the next time you use it. Nothing for you to re-download or re-configure.
How does billing work?
You keep one balance that every agent draws from, and you pay per job — not a subscription per agent. Chatting to scope what you need is free; the agent then proposes a job with a price you approve before any work starts. The price is held when you approve and only charged once the job completes — if a job is cancelled or fails, it's automatically refunded. Track every job and your balance in one dashboard, and a hard spending ceiling means nothing ever runs away from you. (Per-agent subscriptions are coming soon, managed from the same dashboard.)
Can I run agents automatically on a schedule?
Scheduling is coming soon. You'll be able to set work once and have it run on your behalf — even chaining different agents from different builders into one routine, like an inbox summary at 8:00 and a reconciliation at 21:00. The same approvals and spending limits will apply.
Is my data safe — does every agent see everything?
You're in control. You decide what each agent can access from your Passport, and you can revoke access in one click anytime. Agents only get the tools and context you choose to share with them.
What if I connect a tool and want to remove it later?
Connections live in your Passport, not inside each agent. Add or remove a tool once and the change applies everywhere — no hunting through individual agents to update it.