A working assistant
Deployed where your team already works, answering from your sources with citations.
01 Service / Custom tools
Custom AI Tools & Assistants
Internal assistants and purpose-built tools trained on your data and grounded in your actual process. Not a generic chatbot bolted onto your website.
02 What it is
Built on your data, shaped by your process.
A generic chatbot knows nothing about your policies, your pricing, or how your team actually decides things. We build assistants grounded in your documents and your process, so the answers cite your sources and stay inside your rules. When the tool does not know, it says so instead of guessing.
Before anything reaches your team, we test it against real questions from your business and measure the accuracy. And if an off-the-shelf product already does the job, we tell you to buy it. A custom build has to earn its cost.
03 Who it's for
When custom beats off-the-shelf.
A custom tool is worth it when the knowledge is yours and the generic options keep getting it wrong.
- Your team answers the same internal questions over and over.
- Institutional knowledge lives in a few people's heads and a scattered drive.
- Off-the-shelf AI tools don't know your policies, pricing, or process.
- You tried a generic chatbot and nobody trusted its answers.
04 How it works
Five steps to a tool your team trusts.
Trust is the deliverable. Every step exists to make the answers dependable enough to use.
- 01
Map the questions
Collect the real questions your team asks, and what a good answer looks like.
- 02
Ground it in your data
Connect your documents and systems so answers come from your sources.
- 03
Build and constrain
Set the guardrails: what it may answer, what it must refuse, what it cites.
- 04
Test with your team
Measure accuracy against real questions before anyone relies on it.
- 05
Deploy and monitor
Roll out where your team already works, with usage and accuracy tracked.
05 What you get
What you get.
The grounding pipeline
The connection between your documents and the tool. Your data stays in your accounts.
An evaluation report
Measured accuracy against a test set of real questions, run before rollout.
Admin controls and docs
Update the knowledge base and guardrails yourselves, without calling us.
06 Timeline & investment
What it costs, and how long.
Scope depends mostly on the state of your source documents and how many systems the tool must reach.
| Phase | Timeline | Pricing | What it covers |
|---|---|---|---|
| Scope & question mapping | 1 week | Fixed fee | The question set, source inventory, and accuracy bar for the build. |
| Build | 3–10 weeks | Scoped per tool | Grounding pipeline, guardrails, and integration into your stack. |
| Evaluation & rollout | 1–2 weeks | Included | Accuracy testing with your team, deployment, and admin handover. |
Figures are illustrative. Actual pricing is quoted after a short scoping call.
07 Questions
About this service.
Where does our data go?
It stays in your accounts. We build inside your stack with your access controls, and your documents are not used to train anything outside your business. The data-handling setup is documented as part of the build.
How do you stop it from making things up?
Three ways: answers are grounded in your documents and cite their sources, the tool refuses questions outside its scope, and we measure accuracy against real questions before rollout. If it cannot meet the accuracy bar, it does not ship.
Should we just buy an off-the-shelf tool instead?
Sometimes yes, and we will tell you when. The scoping week exists partly to answer build versus buy. If a product you can subscribe to for a fraction of the cost does the job, that is the recommendation you get.
08 Related
Where to go next.
Other services
Get answers your team can actually trust.
Book a call and bring the questions your team keeps answering by hand. We will tell you what a grounded assistant takes to build.