What we need to see
The questions your team asks repeatedly, source systems and a simple list of who can see or change each record.
AI work pattern · every industry
An AI Assistant That Knows Your Business.
ChatGPT knows the internet. It doesn't know your jobs, your customers, your stock, or your shop floor. We build internal AI assistants that connect to your actual systems (MES, CRM, accounts, shared drives) and answer questions in plain English. Shop-floor staff use it on a tablet. Office staff use it on their desktop. Both ask "where's job 1647?" and get an instant answer.
Tell us the task, your current tools and where the work gets stuck. We will help you choose a sensible first scope.
Prove it on your own work first.
How we judge the pilot
Test useful answers, source accuracy and permission boundaries with real staff before enabling actions.
What stays in your hands
Start read-only. Require confirmation for agreed changes and retain an audit trail and manual route.
The work in context · Illustrative scenes
Map the inputs and the handovers.
Start with representative examples, the tools involved and the point where the work slows down.
Make review part of the workflow.
Show the prepared work to the right person, flag uncertainty and keep a clear route to correct mistakes.
The problem, in plain English.
01
Information is locked in systems
Job status in the MES. Customer history in the CRM. Stock in the ERP. Drawings on the share. To answer "is job 1647 ready to ship?" someone has to open four tabs. Most people just walk over to the shop floor and ask.
02
Your team can't use ChatGPT for this
Public AI doesn't know your jobs, your customers, or your data. And you don't want your business data in a public model. That's the whole problem with "just use ChatGPT" as an answer.
03
Shop-floor staff get left behind
The office uses spreadsheets and dashboards. The shop floor uses a noticeboard and a clipboard. Bridging that gap with traditional software means custom screens, training, change management. AI chat does it in plain English, anyone can use it.
04
Knowledge walks out the door
"How do we usually handle that with this customer?" lives in one person's head. An assistant trained on your past quotes, jobs, and customer notes makes that knowledge available to everyone, including the new starter.
What actually happens.
01
Connect to your real systems
API connections to your MES, CRM, accounts, ERP, document store. Read-only by default. Write access for specific actions you approve (logging a quality issue, moving a job stage, starting/stopping a task).
02
Retrieval-augmented, not memorised
The assistant queries your live data when asked, it doesn't memorise it. Answers are always based on the current state of your business, not a cached snapshot.
03
Action-capable where it matters
Beyond Q&A, the assistant can take actions: start a task, log a defect, send a customer update, raise an internal ticket. Every action confirms before executing, and every action is logged.
04
Safety rails for the shop floor
Voice-first interface for noisy environments. Confirmation prompts before any destructive action. Role-based permissions, operators can't change pricing; estimators can't edit production schedules. Built for real workshops, not air-conditioned offices.
Fixed fee, phased delivery.
£5,000-£18,000 for build, ~£300-£800/month for hosting + model usage
From £5,000excluding VAT
Internal AI assistant
- Workflow audit + scope
- Build, integrations, and tuning to your real data
- Deployment, handover and agreed access controls
- Hosting, AI usage and ongoing support quoted separately
What people ask us.
01Isn't this just ChatGPT with a system prompt?
No. ChatGPT plus a system prompt is a chatbot. What we build is connected to your live business data, can take real actions, has audit logs, has role-based permissions, and is hosted in your environment. Different problem, different solution.
02What about hallucinations?
The assistant only answers from retrieved data, if it can't find the answer in your systems, it says so. We don't let the model "make up" facts about your business. For action-taking, every action confirms before executing, so even if the assistant misunderstands, the human has the final say.
03Will the shop floor actually use it?
Yes, if it's designed for them, voice-first, big touch targets, plain-English answers, and integrated into the workflow they already follow (e.g. on a tablet at each work station). The technology is the easy part; the UX is what makes adoption stick.
04How do you keep our data private?
We agree where data is processed, which providers are used, retention settings and access permissions before the build. We document the processing terms and test the chosen setup against your requirements. Private or on-premise deployment is scoped where feasible.
05Can it integrate with the systems we already have?
Almost always, yes. We've connected assistants to bespoke MES, Sage, Xero, Microsoft 365, Google Workspace, SharePoint, custom databases, and email. If the system has an API or even a file-based interface, we can integrate.
Which industries this suits.
Other AI work we build.
Keep exploring
Build this for your business.
What could work
better for you?
A new website. More customers. Less admin.
Tell us where you want to get to. Chris or Kay will reply within one working day.