An example from our own work
Our own proposal and sales-document system.
What we did
Our internal system for proposals and sales documents. Shown as an example of how we work, not as a client project.
AI & automation
AI to understand the information. Clear rules to act on it.
We use AI to understand information and support decisions, and defined rules for checks and actions that need to be consistent. That way an emailed order, a document or a request becomes an orderly step in a process — with approval points in the right places.
Connections in stages
You can start with reading and preparing drafts, and add authorised actions only after testing.
Read
The system reads and summarises information from agreed sources.
Draft
It prepares a draft — an order, a reply or a summary — for a person to review.
Approve
A person approves or corrects it. Exceptions and unclear cases stay with the team.
Authorised action
Only after testing do we add writing to systems, within the permissions defined.
What it looks like on a normal working day
An order arrives by email
The document's details are captured and checked; missing information is flagged; the authorised action happens in the business system.
A customer needs service
The request is collected with the details needed, assigned for handling and given a status you can follow.
A manager needs the picture
Data from agreed sources is gathered into a short report with exceptions and items that need a decision.
Production is done
An update from the factory floor is linked to the order details and to the next delivery step.
An example of a process that can be scoped. What gets automated depends on the systems, permissions and needs of each business.
What we check before expanding
- The usual case, and also missing data, the wrong permission and a failed connection.
- Which sources the system reads from, who may use them, and how missing or stale data is spotted.
- How we know an action finished, what happens when it fails, and who follows up.
- A business measure that fits the process, with a baseline before any comparison.
Read the story
Common questions
Does the agent do everything on its own?
No. We define the limits of what it may do, its permissions and its approval points. Sensitive processes and professional requirements need the right controls.
What about privacy and our business data?
Data sources, permissions, storage and processing services are defined as part of the solution. Not every system or model works the same way.
Do we need to replace our systems?
Not necessarily. We first look at what already works and what can be connected. If a new component is needed, we define how it fits with the systems you already run.
Can we start with a small project?
Yes. We can define one focused process and test it before expanding. Scope, cost and success criteria are agreed up front.
Which piece of work keeps coming back too often?
Tell us what happens today, which systems you use and where things get stuck. We'll look at whether to connect what you have, build a dedicated tool or bring in an AI agent.