What it costs
What it costs
The Audit
A fixed fee, quoted after the discovery call, with the Blueprint included.
Each workflow we build
A fixed build fee, set by its level on the scale below and quoted per row in your Blueprint.
The monthly review
A monthly fee per workflow set up on Claude or ChatGPT. Each month we rerun your saved cases, check the approval settings, and tune the rules and prompts with you. You pay Claude or ChatGPT direct.
The retainer
A monthly fee per agent we build, from its first day of live work, with a floor per company, quoted in your proposal. It pays for monitoring, exception tuning, swaps to newer models, the 30 and 90 day measurements, the log and your control panel. You pay the model and licence vendors direct.
Where it runs
You choose where this runs: Claude or ChatGPT, our hosting, or your own hardware. You also choose how it starts, standard or controlled. A remote install is included; an in-person install is quoted.
The context session
A context session teaches your agent how you write and negotiate. It sits between the mapping session and go-live, and it costs nothing extra.
The workflow complexity scale
Level 1 Score 0 to 2
Routine
Fixed rules in one or two systems, and a person approves each action.
For example
- Invoicing and collections
- Lead follow-up
Level 2 Score 3 to 4
Guided
A few decision rules, and a person takes each case the rules do not cover.
For example
- Inbox and scheduling
- Purchase order to receipt matching
Level 3 Score 5 to 6
Judged
The agent acts within a threshold on cases that take judgement or arrive by the hundreds.
For example
- Collections and cash application
- Candidate screening and scheduling
Level 4 Score 7 to 8
Negotiated
Back and forth with an outside party: the agent drafts, and a person decides the hard cases.
For example
- Contracts and negotiations
- Support ticket triage
Level 5 Score 9 or more
Orchestrated
Many systems or a desktop-only tool, high volume, and agents that hand work to each other.
For example
- Purchase-order approvals across entities
- Month-end reporting across entities
Four kinds of agent carry no price, because we do not build them: an agent that approves spend, changes bank details, releases a payment, or signs a contract.
Starter path
For solo operators and small teams
Starter prices- 400for 90 minutes on video
A mapping session
You get a map of your business and a starter list with each workflow's saving, payback month and access level.
- 700 to 1,500, proposedto set up one workflow
Your first workflow
ProposedWe set it up on Claude or ChatGPT with your accounts, write its rules and approvals, and train your team. It must pay itself back within three to six months.
- 100 to 200, proposeda month per workflow
Monthly review
ProposedEach month we rerun your saved cases and adjust the rules. You pay Claude or ChatGPT direct, from 20 a month.
Runs on Claude or ChatGPT, our hosting, or your hardware. Standard or controlled rollout.
Need a hard stop in code, or more volume than these allow? We build and host it: 1,500 to set up, 250 a month.
Why start now
51,800 a month, or 155,400 a quarter.
Agents read, check and key; hours return.
Each month we check your setup still works.
Transparent transition, any time
If another firm or your own team takes over, we hand over the files and complete control within 24 hours, and the agents keep running. The handoff ends the retainer, with no exit fee.