Central AI knowledge hub

Your knowledge. Useful together.

Your business already knows a lot. But that knowledge is scattered across people, documents and systems. A central AI knowledge hub makes selected knowledge useful for daily work, with the agreements and checks your team needs.

01 · A work example

From source to useful proposal.

See how sources and working agreements help prepare an answer together. This is an illustration with example content, not a live connection to business systems. The demo does not send or save data.

Business memory at work

Fictional, prewritten example · nothing sent or saved

Retain → recall → reflect

Business question

Can we start in November?

Retain

Keep agreements and experiences, with source and date.

recall

Recall the relevant, current context for this question.

Request · fact
Maintenance for two locations.1
Agreement · fact
Intake first; project schedule follows.2
Current limit
6 Oct: capacity not confirmed.3

reflect

Weigh the evidence and draft a proposal with sources.

Draft email · proposal, not a fact

Check capacity before promising a date.

For both locations, let’s scope the work in an intake first.2 I cannot yet confirm a November start: capacity still needs checking.3

Then a person checks the proposal.

Draft · awaiting your review

Selected memory, not model retraining.

No live AI. Changes and approval reset on reload or when switching tabs.

Source 1 · Sales

Prospect’s question

Email · customer request ·

We need maintenance for two locations. Can we start in November? We want a clearly scoped proposal first.

Fictional example source. No customer data or live connection.

Source 2 · Sales

Service agreement

Service sheet · approved ·

We first scope the work in an intake per location. Only then do we confirm the quotation and project schedule. An intake is not a project start.

Fictional example source. No customer data or live connection.

Source 3 · Sales

Initial planning note

Meeting note · initial status ·

Available intake and project capacity have not been confirmed. Do not promise a November start. Check planning before offering a date.

Fictional example source. No customer data or live connection.

Source 4 · Sales

New approved intake plan

Planning decision · approved by planner ·

Approved by the planner on 9 October: an intake slot is available on 12 November, 10:00–11:00, to scope both locations. This replaces the unconfirmed intake capacity in the 6 October note. The customer still needs to accept the slot. The project start is not confirmed; quotation and project planning follow the intake.

Fictional example source. No customer data or live connection.

Source 1 · Marketing

Maintenance service sheet

Offer · approved ·

Our maintenance service provides one contact person and agreements per location. We have no evidence for a fixed time-saving claim.

Fictional example source. No customer data or live connection.

Source 2 · Marketing

Brand voice agreement

Style guide · approved ·

Write directly and humanly. Avoid superlatives and unsupported promises. Close with a question that invites a conversation.

Fictional example source. No customer data or live connection.

Source 3 · Marketing

Lessons from sales calls

Sales notes · observed pattern ·

In recent sales calls, prospects asked who keeps track of maintenance when everyone is busy. This is an observation from those calls, not proof of what all customers want.

Fictional example source. No customer data or live connection.

Source 1 · Reporting

Open quotations

CRM export · dated snapshot ·

Quotation A awaits a response. Quotation B still has an unclear scope. Neither quotation is recorded as closed revenue in this export.

Fictional example source. No customer data or live connection.

Source 2 · Reporting

Project planning

Planning · provisional ·

A provisional start slot is reserved for B. It is not confirmed because the scope is still unclear. Do not present the reservation as a confirmed start.

Fictional example source. No customer data or live connection.

Source 3 · Reporting

Weekly meeting decision

Action list · approved ·

Sales will clarify quotation B’s scope this week. The project lead will only confirm the schedule afterwards. This action is agreed, not yet completed.

Fictional example source. No customer data or live connection.

02 · Four building blocks

How knowledge becomes usable.

01 · Context

The right knowledge

Relevant information, definitions, examples, permissions and agreements for each workflow.

02 · Way of working

A consistent route

Repeatable ways to find information and prepare output.

03 · Review

A person who decides

A designated owner reviews output where errors have meaningful consequences.

04 · Handover

A solution you understand

Documentation, training and recovery steps make the solution understandable for your team.

03 · Sources and oversight

Your systems remain authoritative.

Chat can be an entry point for finding knowledge or drafting a proposal. Your people and existing systems remain the starting point. For each workflow we agree what AI may prepare and where human approval is needed.

  • Your CRM, ERP or accounting package remains the official data source.
  • Only selected information gets an agreed purpose and access.
  • Autonomy and approval are decided for each workflow.
  • Privacy, security and measurement goals are linked to the specific architecture.

04 · The first workflow

Start where work gets stuck.

For one business this starts with shared marketing and sales context. For another, with reporting, knowledge searches or one repeatable process. The opportunity scan helps choose an entry point focused enough for a first trial.

See what the scan delivers

Choose the first workflow

Explore where a central AI knowledge hub can help first.

Email one specific work situation. Pieter helps decide whether it is focused enough for a first trial.

Email Pieter about your AI opportunity scan

45 minutes · selected problem · recommended next step

Prefer to call? 0473 80 38 03