For businesses · delivered by Cascadian
We go into your business and build the agents.
Free up your team’s time, increase revenue, scale without headcount, and raise what the business is worth. The audit is how that starts — two weeks, fixed scope, and you keep the plan whether or not you hire us.
Before you pay anyone for this
There are free scanners that will read your repo and score it. Run them. They are good, and we run them too as an input to this.
What they do not do is tell you what to fix first, what it costs, who should own it, or what to build once it is fixed — and then build it. That judgment is what you are paying for here. Not the scan.
What arrives
Five artifacts, all of them countable.
Two weeks from kickoff. Fixed scope, fixed price, and the date is held when you sign — not when we get to it.
01
The configuration read
Your CLAUDE.md, .claude/ directory, permission posture, MCP surface, hook coverage and CI integration — read, not asked about. Scored against 14 categories.
A scored report, one page per category
02
The spend picture
Where your token budget actually goes, itemised by project and by person, with the three largest recoverable line items named.
A cost breakdown and a projected saving
03
The risk register
What an agent can currently touch that it should not. Prod writes, secret exposure, unreviewed tool permissions, missing stop conditions.
A ranked risk list with the config that closes each one
04
The agent opportunity map
The recurring work in your business that should be an agent, ranked by hours returned per week against build effort.
8–15 candidate agents, scoped and estimated
05
The remediation sequence
Everything above, put in the order it should be done, with an owner and an estimate on each line. The part the free scanners do not do.
A 90-day plan you could hand to someone else
Setting expectations
What you are not buying.
Not this
This
A slide deck
A diff list
An interview about your setup
A read of your actual configuration
A maturity score and a handshake
A sequenced plan with owners and estimates
A pitch at the end
The report, whether or not you hire us
Price
One number, published.
Most firms in this category will not put a number on a page. We will. If it is not worth this to you, it is better that we both find out now.
Flat, any team size
$5,000
Two weeks, fixed scope, delivered remote. Same five artifacts whether it’s one repo or a multi-team surface with compliance constraints.
What we do
Four kinds of work, one team.
The audit is where most engagements start, because it is the cheapest way to find out what is worth building. It is not the only thing we do.
01
Training
Your team, their actual workflows, in a room. They leave with files that run, not notes about files. Half a day or a full day, on-site or remote.
02
Setup and automation
We get Claude installed properly against your systems — connectors, permissions, guardrails — and put the recurring work on a schedule so it finishes before anyone sits down.
03
Custom skills
The jobs that are specific to your business, written as files your team owns. Every engagement leaves them behind; they are yours, not licensed back to you.
04
Forward-deployed engineering
The work that does not fit in a workshop. We sit inside the team, build against your real systems, and stay until it is running in production.
After the audit, if you want it
Your next ten employees won’t be human.
We build the agents on the map, in the order the plan says. Fixed-scope projects, or a monthly pack if the list keeps growing — which it does.
Project
$15k – $40k
Fixed scope, fixed price, delivered.
Deploy
$7,500 / mo
3 agents a month, library access, maintenance.
Scale
$15,000 / mo
5–10 agents a month, priority queue, monthly architecture review.
Embed
$25,000 / mo
Weekly on-site. Eval harness and cost governance owned by us.
Three-month minimum on retainers. Website builds are a separate flat $10k–$25k, delivered in 30 days.
The cheapest way to find out if we are any good
Take the library first. It is free and it is the same work.
Every skill we run on client engagements is published. If the files are useful, the audit will be too. If they are not, you have lost nothing.
The assessment
How good are you at Claude, actually?
Eight questions. You get the level you actually test at, the one move that gets you to the next one, how many hours a year the job you name is costing you, and three skill files cut to your title, company and industry.
Take the testUnder two minutes · no account · change any answer
- 01The Beginner
- 02The Chat Master
- 03The Operator
- 04The Automator
- 05The Agent Conductor
Section surveyed 5,013 knowledge workers: 54% rate themselves AI proficient, about 10% test as proficient.