Fit your people and your budget
You ask: How do we fit this around our people, and keep costs down?
Your people have limited time, and your budget is tight.
Work is designed to follow your rhythm: your working days, your decision cycles and how much you can review. Not everything needs a live chat. Work that can wait is planned into batches that run while you are away, at lower cost. The provider's page says batch saves 50 percent. We have not measured it.
A batch job runs today, and a spending limit is refused before a call is made. Choosing real time or batch for each task is planned.
How it answers
- Each task would run now, or by a stated turn at a lower price. Planned, not built
- A batch job runs today. Anthropic's pricing page says batch saves 50 percent (tool summary, 2026-10-13). Nothing here measures the saving. Measured (job); price unverified
- Nothing would need an answer in real time. Waits have a stated turn. Planned, not built
- Recurring work starts itself, and a failure reaches your inbox. Measured
- A spending limit is refused before the call is made, and one switch stops all spending. Measured (spending only)
- A restart loses no batch work. Measured (batch); laptop sleep unverified
Reliability on this page
Spending is checked before it happens: a call that would pass your limit is refused (claim 5), and the cost of each run is recorded (claim 8). How we raise reliability
The honest limit
- The stop switch stops spending. It does not stop a job that is running now. Limits on steps, tokens and wall time for each job are not built (issue #58).
- A laptop that goes to sleep is unverified. Nothing measures yet what a person spends reviewing.
For NGOs
Decisions by committee, funding cycles, small budgets. The first offering is meant to be a guided setup into your own account or server.
What B Team costs
Decided The founder (Codrin) decided this on 2026-10-13. The amounts and the test for who counts as a verified non-profit are not decided, so each amount below is an empty placeholder. Not decided B Team is not offered yet. There is no price per user.
| What | Verified non-profit | Public body | Company |
|---|---|---|---|
|
The software
Free for verified non-profits, and for public bodies under the licence. Security fixes are free for everyone.
|
Free | Free under the licence (a lawyer confirms) | [price] |
|
Onboarding package
Start is the default way in: the guided setup, the data policy, the first project and an AI literacy session. Grow covers defining goals, writing test sets and adding projects. A day rate covers anything beyond.
|
[price] | [price] | [price] |
|
Project support, per active project
A project is one product with its own code repository. It is active in a month in which work was done on it. There is no price per user.
|
[price] | [price] | [price] |
|
Yearly maintenance and support
It pays for the service: help, guaranteed response times and supported upgrades. Security fixes themselves stay free for everyone.
|
[price] | [price] | [price] |
|
Hosting in Europe
Later, per project. We intend to offer a hosted service in Europe. It is not built.
|
Planned, not built | Planned, not built | Planned, not built |
- Who counts as a verified non-profit: for Dutch organisations, ANBI status or CBF recognition, checked by hand. Something comparable abroad.
- Public bodies pay full price for the services. Under PolyForm Noncommercial, the working direction for the licence, a public body may use the software free of charge, so the full price is for onboarding, support and yearly maintenance. A lawyer confirms the licence (issue #22).
- Companies pay full price for the services and also pay for a commercial licence.
- Security fixes are free for everyone. The yearly fee pays for the service around them, not for the files. That follows the Cyber Resilience Act's free security updates, and the lawyer confirms it.
- Model costs: you pay the provider directly, with your own key. Paying through us, as a one-stop shop, is later. It needs the providers' written approval and a processing agreement with us. Planned, not built
What cost can truthfully be shown now
- Spending limits are enforced before a call. Measured
- The model cost of one test run is known: the job reviewer's test set of 15 checks cost $0.0896 in model fees. That is one test run of one role, not a price for you. The job worker's test run of 13 cases has no cost figure. Measured
- Later, cost per accepted change, rejected work included. Planned, not built
The main benefit we aim at is coordination around real value, not a low price. No total cost of running B Team exists yet. We say so because hiding a price reads as evasive.
The technical detail
The claims behind this page, with their states as the evidence page records them.
| No. | Claim | State | What it serves | Check or issue |
|---|---|---|---|---|
| 4 | Bulk work goes through batch. Anthropic's pricing page says batch saves 50 percent (tool summary, 2026-10-13). Nothing here measures the saving on our jobs. | Measured (job); price unverified | Batch job type (F3) | #148 |
| 5 | A spending limit is enforced before a call. | Measured | Cost caps before a call (N3) | ledger_checks; test_budget |
| 6 | One switch stops all spending, per project or all. It does not stop a job already running. | Measured (spending only) | Kill switch (S13) | stop_checks (4 of 4) |
| 7 | Limits on steps, tokens and wall time per job. | Not built | Runaway limits, kill switch (S13) | #58 |
| 18 | A flood of items cannot flood the person. | Measured (stand-in only) | Cap on what may wait (F4, S13) | checks 7a, 8a |
| 20 | A test set must pass before a model or instruction change is used. The gate is a person's command. | Measured; gate manual | Test-set gate (F5) | golden_checks; #115 |
| 36 | Recurring work starts itself. A failure reaches the inbox. | Measured | Scheduled jobs (F14) | schedule_checks |
| 37 | A restart loses no work. | Measured (batch); laptop sleep unverified | No lost work (N2) | isolation_checks |
| 51 | Each task runs in real time or in a cheaper batch. | Not built | Real time or batch per task (F26) | #196, #65 |
| 52 | Five measures are recorded by the machinery and shown together. | Not built | The five measures (F27) | #290 |
| 56 | Nothing requires a person to answer in real time. Waits have a stated turn. | Not built | Work paced to people (N8) | #65, #196 |
The other four questions
-
Question 1
Build what actually matters
Start from your goal. End with real users. Judge by use.
Read the answer -
Question 2
Choose your models, keep your freedom
You choose your models and where they run.
Read the answer -
Question 3
People and AI, building together
Agents write. People decide, guide and check.
Read the answer -
Question 4
Use AI and stay compliant
B Team produces the records your compliance needs as you work.
Read the answer
Tell us you are interested
B Team is not offered to anyone yet. NGOs come first. If you want to hear when that changes, or you would like to be one of our first pilots, write to us.
Write to: [contact address to be decided]
This draft has no form and collects no data. Once there is an address, what happens next is meant to be this: we tell you whether we can work with your organisation yet. We promise no date and no price.