AI in your business, only where it helps
We sort what AI should handle from what a simple rule already handles, and put it to work with your team.
Is this you?
Everyone's talking about AI and you don't know where to start.
You tried a tool, and it became one more forgotten tab.
Your team spends the day copying data from one place to another.
If one of these sounds like your company, it's worth a call.
Let's talkThe same work, redrawn
- InA quote request comes inInA quote request comes in
- By handSomeone reads it and copies it into a spreadsheetAIAI reads the request and fills in the fields
- By handSomeone checks prices and stockAutomationThe system checks prices and stock
- By handSomeone puts the quote togetherAutomationThe quote comes out ready
- Person approvesA person checks and approves it
- OutThe customer gets the replyOutThe customer gets the reply
- InA quote request comes in
- By handSomeone reads it and copies it into a spreadsheet
- By handSomeone checks prices and stock
- By handSomeone puts the quote together
- OutThe customer gets the reply
- InA quote request comes in
- AIAI reads the request and fills in the fields
- AutomationThe system checks prices and stock
- AutomationThe quote comes out ready
- Person approvesA person checks and approves it
- OutThe customer gets the reply
Not everything needs AI
Every task goes down one of five paths. AI is just one of them.
Leave it as is
2- Negotiating with a big customer
- On-site visit
Get it organized
2- Commission spreadsheet in three versions
- An approval that goes through four people
Off-the-shelf tool
2- Sending invoices
- Team calendar
Automation, no AI
2- Payment reminders
- Copying store orders into the system
AI, with a person reviewing
1- Reading invoices and entering them in the system
Where AI usually helps
Customer service
Answers what keeps coming back and calls in a person when needed.
When every conversation is a negotiation.
Documents
Reads invoices and contracts and fills in your system, showing where each value came from.
When only a few come in each month.
Emails and requests
They arrive sorted by topic and priority.
When one person handles it in a few minutes.
Your team's daily work
A company AI tool, with rules for using it, to write and summarize.
Without a defined task for it to do.
How it works
Call
You tell us what takes up time. We tell you if it can be solved.
30 minutesfreeDiagnostic
What's worth doing, in what order, and what it costs. The document is yours.
1 to 2 weeksfixed priceTeam setup
The right tool, rules for using it, and a session with each team.
optionalTest on your data
How often it's right and what it costs. If it isn't worth it, we go on without AI.
before buildingBuild
One part per stage. You pay after you approve it.
1 to 2 week stagesLive
60 days of free fixes, and the result measured at 30 and 60 days.
optional support
Your data stays yours
We sign a data processing agreement and never use your data to train AI. Anything that affects a customer or money, a person approves first.
And your team?
AI takes copying and retyping off your team. What to do with the time that frees up is your call.
AI systems already running
We describe each one by the kind of business. Client names stay confidential.
- 01
Email triage
- Support emails came in with no category and no priority.
- Every email arrives with a category, a priority and a summary. Rules handle what they can, and a stronger model steps in only when needed.
- 02
Document reading
- Every credit decision depended on someone reading the documents by hand.
- The system reads each financial document field by field and shows the confidence and origin of every value.
- 03
Data from official sources
- The regulatory information that mattered was spread across government sources.
- The data comes out structured with its source cited, a second model checks it, and the team can ask questions of it.
Common questions
How much does it cost?
Cost depends on the size of the problem. We talk numbers in the first call, with no commitment. The diagnostic is paid, has a fixed price, and ends with the price of each stage.
My company is small. Is this for me?
It is when there's work that repeats every week. The diagnostic exists to find that out before you spend anything on building.
Will my data end up in ChatGPT?
Only what you approve, in a company account, on a plan where the provider states it does not use your data to train AI. That goes into the agreement we sign.
What if the AI gets it wrong?
It sometimes does. That's why the test measures how often it's right before anything is built, and a person reviews anything that affects a customer or money.
We tried it and it didn't stick. What's different?
Often the tool arrives without a defined task. Here it goes into a task picked in the diagnostic, with rules for using it and a team trained on that task.
Who owns what gets built?
Your company. Code, accounts and licenses are under its name from day one.
Where AI helps
Tell us in a few words what takes the most work today. You don't need anything prepared.
In a free 30-minute call, you'll find out whether AI helps and what the path would look like.