An AI assistant on your own business data
We build it for small and mid-sized businesses. The assistant knows your email, your documents and your business software. It finds what your team needs and gets the work ready. The data stays yours.
This is where your office's time goes.
- Searching
- A client asks what was agreed last year. The answer is in an email, in a quote in a shared folder or in a colleague's head. Someone goes looking.
- Retyping
- A request arrives in Outlook. Someone retypes the name, the company and the question into the CRM, and again later into the quote and the invoice.
- Following up
- On the phone, someone promises a quote by Friday. The promise isn't written down anywhere. If that colleague is ill or on leave, it waits.
What the assistant does for your team.
It works on the same data as your people and does the preparation. Anything that goes to a client or changes your systems, a person approves first.
Answers questions, with the source
Ask what was agreed with a client. The assistant searches email, call notes and your CRM, and shows the email or note the answer comes from.
Handles incoming email
A request in Outlook or Gmail becomes a record in your CRM, or a ticket for the right person. The assistant reads the name, company and question off the message. If the record exists, it updates it.
See how an email becomes a CRM recordLogs calls
After a phone call, the transcript sits with the right contact. What was promised, like sending a quote or calling back, is ready as a task.
See how a call lands in the CRMPrepares forms and documents
A form that takes fifteen minutes to fill in arrives pre-filled. A report gets a first version from your own data, and a person checks it.
Flags what is stuck
The assistant flags a stalled ticket or a follow-up nobody picked up. They show at the top for whoever has to handle them.
Your data stays in your own database.
- We mirror your email, documents and business software into your own database. Your team keeps using Outlook, Exact and the other tools.
- You decide which data may go to a commercial AI model, which stays on private infrastructure and which never leaves the database.
- The assistant gets only the access its task needs. Every question, answer and action goes into a log on your own infrastructure.
First an architecture review, then one task.
It starts with an architecture review, a few weeks of scoped work. You get a map of your data and a phased plan. The plan names the first task, for example the requests in the inbox.
We connect only the sources that task needs and build the assistant for it. A first version can be ready in as little as two weeks. When it works, the next task follows, on the same data. You don't rebuild the first one when the second arrives.
Questions about AI on your own data.
Do we have to replace Outlook, Exact or our CRM?
No. Everything that works stays. We sync the data to a database you own, and the tools stay where they are.
Can the assistant send email or change data on its own?
Only what you allow. It gets scoped access per task, and every action is logged under its own identity, so you know what happened, when and on whose authority.
Which AI model does the assistant use?
The one that fits the task. On your own backend you choose per task, and switching provider is a configuration change. Data that may not leave your infrastructure goes to a private model.
What does an assistant like this cost?
That depends on the task and on the systems we connect. You get an indication in the first call and a firm figure once we have scoped the task together.
Start with the task that takes the most time.
Describe it below. After a 30-minute call you have a one-page plan and a ballpark price.