04 · AI
Company AI Assistant
AI suggests. Humans approve. Systems execute.
In short: A customer-facing assistant that answers from your own data, or an internal one that knows your files and policies. Safe, permissioned, improving over time.
See it working
The assistant proposes, you approve — try the button yourself.
The problem
Customers ask the same questions all day.
Your team can't find the policy, the price list or last year's file.
You tried a chatbot and it made things up.
What you get
| Component | What it does for you | Result |
|---|---|---|
| Customer assistant | Answers from your content on your site or WhatsApp, 24/7 | Fewer repeated questions |
| Internal assistant | Searches your documents with the user's own permissions | Answers in seconds |
| Guardrails | Cites sources, says 'I don't know', escalates to a human | Trust |
What it's not
- Autonomous actions on your systems without approval
- Training a model from scratch
FAQ
Which AI do you use?
Claude by default, server-side only, with a key and spending limit dedicated to your project.
Is our data safe?
Your documents stay in your database. Personal data is never sent un-anonymised, and access follows your existing permissions.
How does it get better?
The monthly plan covers reviewing unanswered questions and updating the knowledge base.
Start with a 15-minute call.
Tell us what eats your week — we'll tell you, honestly, what the right next step is.