Vimpex | gecco

Overview

34+ hours freed every week. That is the measured impact for Vimpex, a UK manufacturer of fire safety and rescue equipment, reported by six members of staff six months into a partnership with gecco that started as a short, focused engagement and has since continued as an ongoing one.

Fire safety manufacturers run on data spread across ERP systems and processes that have quietly done the job for years without needing to change. At Vimpex, the daily production meeting couldn't start until the team had pulled the latest figures together by hand, and closing the books at month end took four full days. gecco built a custom connector, using a standard called MCP, giving Claude a live, read-only line into Vimpex's Epicor system, then trained the team to build on top of it themselves.

The result now touches every part of the business. Finance closes the month in a day and a half instead of four. A live morning meeting tracker has replaced hours of manual retyping every week, and staff across the business, several with no AI experience six months ago, have gone on to build more than 40 tools of their own.

A whiteboard, a spreadsheet, an old PC

Manufacturers running on ERP systems generate huge volumes of production and sales data, but that data often stays locked inside the system, or scattered across whatever paper, spreadsheet or ageing PC first solved the problem years ago. Getting it into a usable, shareable form usually falls to whoever has the patience to do it by hand.

At Vimpex, the team pulled the latest Epicor figures together by hand into a shared spreadsheet every morning before the daily production meeting could start, a job that took four to five hours a week. Month-end reporting was bigger still: closing the books took four full days, every month.

Sales had its own version of the problem. Tenders, many of them from Fire and Rescue Services, were tracked on a physical whiteboard that got wiped clean once a job closed, taking the history with it. Alerts from a paid tender-finding service arrived daily, but only a handful in a hundred were ever relevant, and winning responses were often rewritten from scratch because nobody could find what had worked before.

The wider picture was similar. A survey of 14 staff found six had never or rarely used AI at work, and the few tools that did exist tended to live in one person's account rather than the business. When a member of staff left, his projects left with him.

A custom MCP into Epicor

gecco's first move was to build a custom connector, using an open standard called MCP, that gives Claude a live, read-only line into Vimpex's Epicor system across both the UK and Sweden operations. In plain terms, staff can now ask Claude for live production, sales and finance data directly, without waiting on a report or retyping a single row.

The clearest example is the morning meeting tracker. Built over several sessions with the two people who run the meeting day to day, it now pulls live Epicor data across both companies automatically, and every exclusion or anomaly is shown on screen rather than hidden. What took hours of retyping a week now takes the team straight to the actual discussion.

Quality benefited too. Vimpex's ISO 9001 calibration register, tracking 67 pieces of equipment, used to be a spreadsheet with no view of what was due or what it cost. gecco rebuilt it as a standalone app that tracks both automatically. Alongside the tools gecco built directly, Vimpex's own staff were trained to build their own, which is where the story gets more interesting.

A 12-hour job, done in a minute

34+ hours a week. That is the floor of what six Vimpex staff reported saving through AI, taking the bottom of every range they gave in a formal review. Across a year, that is close to one full-time role, from six people alone. Month-end reporting, which used to take four full days, now takes a day and a half.

The most striking example did not come from gecco at all. Vimpex manufactures voice sounders, and filtering every recording used to be a manual job on a legacy system that had served the team reliably for years, taking up to 12 hours. One of Vimpex's own engineers modernised it himself, building an AI-powered dashboard that does the same job in about a minute, entirely on his own initiative.

The same engineer went on to build an online configurator for one of Vimpex's fire alarm sounder ranges, turning a fiddly, error-prone paper order into a simple form, and a dashboard comparing UK and US fire safety standards for a market-entry review that would previously have meant weeks of manual cross-referencing. Neither needed gecco in the room.

From using AI to running it

Vimpex's programme has grown from a short, focused start into an ongoing partnership, and the next phase is about consolidation rather than more building. Six AI champions are already in place across the business; the next step is giving them the structure to run department sessions themselves, so new skills spread without a consultant in the room every time.

The rest of the plan closes gaps that fast adoption opened up. That whiteboard of tenders is next in line to be digitised, automations currently spread across individual accounts are moving onto one company-owned platform, and a single hub will show every AI tool in use, who owns it, and what it touches.

What we learned

Start with the task people already hate doing every single day. The morning meeting retype was not the biggest opportunity on paper, but fixing it first built trust fast, and that trust carried into every department that came after.

Training the team to build, not just to use, made the biggest difference. The best example built itself: once staff had the confidence, one engineer quietly modernised a long-standing manual process without being asked, proof that adoption had stopped depending on gecco showing up.

Fast, wide adoption creates its own next problem: knowing what exists, where it lives, and who owns it. Organisations moving at this pace should plan for that consolidation step early, rather than treating it as an afterthought.