Happy Webs
№ 01 / SolutionInternal AI assistant · Fabricators

A Shop-Floor AI That Knows Every Job.

Every fabrication shop has the same daily question: where is job X right now? The answer's in the MES, in the production schedule, on the shop floor, on a clipboard, or in someone's head. We built an internal AI assistant for Kingsland that takes the question in plain English — voice or text — and answers from the live data, then lets staff take actions like starting a task, logging a defect, or moving a job stage.

№ 02The promise

The promise

Live shop-floor deployment at Kingsland
< 8 wk typical build time
On your data private, not shared with public AI
№ 03Where it hurts

Where it hurts

The fabricators reality.

Shop-floor staff are the people closest to the work but furthest from the systems. The MES, the schedule, the customer notes — all on a desk that's nowhere near the bay. They walk over, ask, get a partial answer, go back. Multiply across the day across the team and it's hours of friction. Customer-service staff hit the same wall from the other side: every status query means walking to the shop or pinging someone.

№ 04What we actually build

What we actually build

Tuned for fabricators.

01

Connect to the MES and job systems

Read-only by default. Write access for specific actions: start/stop a task, move a job stage, log a defect, raise a quality issue. Every write confirms before executing.

02

Voice + text on a tablet

Designed for noisy shop floors and bay-side use. Voice-first, big touch targets. Same assistant works on desktop for office staff — different interface, same brain.

03

Answers from live data, not memorisation

The assistant queries the MES when asked — it doesn't cache. "Where's job 1647?" gets the current answer, including the current production stage, who's working on it, and the next scheduled step.

04

Knows your customers and your conventions

Tuned on your past quotes, jobs, and customer notes. "How do we usually handle that with Acme?" pulls relevant historical context — what we've quoted, what we've delivered, what their preferences are.

№ 05Real engagement

Real engagement

The outcome.

Kingsland Fabrications

Conversational AI assistant in daily use by shop-floor staff. They ask "where's job QU-1647?", "what's due this week?", and get instant answers. The assistant can also start/stop tasks, log quality issues, and move jobs between production stages — all through natural conversation with built-in safety confirmations.

№ 06Typical stack

Typical stack

We integrate with what you have.

  • Custom MES
  • Microsoft 365 / Google Workspace
  • SharePoint / Drive (for drawings)
  • Sage 50 / Sage 200
  • Tablets / shop-floor terminals

If your stack isn't listed, ask — almost always we can integrate. We've connected to bespoke MES, ancient on-prem systems, and email-only interfaces.

№ 07Pricing

Pricing

Fixed fee, phased delivery.

£5,000–£18,000 for build, ~£300–£800/month for hosting + model usage

From £5,000

Internal AI assistant for fabricators

  • Workflow audit + industry-specific spec
  • Build, integrations, and tuning to your real data
  • Live deployment in your environment
  • Ongoing tuning + accuracy monitoring
№ 08Questions

Questions

Things people in this industry ask us.

01 Isn'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 MES and job data, can take real actions, has audit logs, has role-based permissions, and runs in your environment. Different problem, different solution.

02 Will the shop floor actually use it?

Yes, when it's designed for them. Voice-first, big touch targets, plain-English answers, integrated into the workflow they already follow. The tech is the easy part — the UX is what makes adoption stick.

03 What about hallucinations?

The assistant only answers from retrieved live data. If it can't find the answer in your MES, it says so. For action-taking, every action confirms before executing — even if the assistant misunderstands, the human has the final say.

04 How do you handle permissions?

Role-based. Operators see job status and can log defects but can't change pricing or schedules. Estimators see customer history and quotes but can't move production stages. Managers see everything. We configure roles to match your real org.

05 What's the cost over time?

Build is one-off. Ongoing: ~£300-£800/month for hosting and AI model usage at typical shop-floor query volumes. That covers the model API, the hosting, and tuning as your data grows.

№ 09The pattern

The pattern

See internal ai assistant for other industries.

Use case hub

Internal AI assistant

ChatGPT knows the internet.

Let's see if we can help.

A 15-minute chat with Chris & Kay. No slides. No pitch deck. You tell us what's on your plate; we follow up by email with real thinking.

Studio81 Penny Meadow, Ashton-under-Lyne, OL6 6EL
HoursMon–Fri · 09:00–17:00 GMT