The whole business mapped, before a tool was chosen.
The core was sound. The joins around it were manual.
BEX Group is the Dublin group behind ROBUS lighting, FuseBox circuit protection and ATC electric heating: founded in 1984, more than 300 people, more than 2,500 wholesale customers, selling into more than 40 countries, with a UK headquarters opening in 2026. A business of that size runs on a mature stack and on established ways of working in every team.
The systems of record were trusted. The work around them was not automated: orders, quote requests, rebate terms and product data arrived as documents, and a person turned each one into an entry. Volumes were rising and every team reported being at capacity. AI use was individual rather than organised. Three teams had built prototypes of their own, and there was no written policy on what AI could and could not touch.
The Chief Commercial Officer sponsored a whole-business discovery before any build, rollout or subscription was committed to: where does AI earn its place across the group, what should come first, and what should not be touched at all.
Ask every function, sit with the teams, map every connection.
Eight questionnaires
One instrument per function: leadership, general, internal sales, design, warehouse, finance, marketing and IT. Each team gave its own account of its work, the tools it spends its day in and how it already uses AI, and the leadership team scored its own readiness and confidence in its data.
Interviews and the map
Discovery interviews across leadership, operations in Dublin, the UK warehouse, finance, marketing, internal sales, purchasing and lighting design, one of them a live keystroke walkthrough. From the interviews and the teams' own working documents we built an operating model of the group: 582 connections between people and systems, each classed by how it moves, 204 of them re-keyed or assembled by a person.
The readout
A findings report and a recommendations document, delivered on 15 September 2026 at an in-person readout. Every figure in the report is the team's own, given in interview or seen on screen and checked back to the transcript. Nomad timed no process and extracted no data from any system.
Nine findings, one pattern.
The same shape repeated across the group. The system of record is trusted, and the step that turns a document into an entry is done by a person. The constraint was not the core but the connections around it, and most of the risks the report set out were about ownership rather than technology: a named owner per data area, written down.
- 01
Order and quote capture
Orders and quote requests arrive as documents and a person keys them in.
- 02
Demand planning
Forecasting runs by hand in spreadsheets beside the stock record.
- 03
Management reporting
Several teams assemble their packs by hand from more than one system.
- 04
Customer rebates
Terms arrive on each salesperson's own template and are typed into a file.
- 05
Lighting design
A desk at capacity, with volume rising and the schedules counted by hand.
- 06
Quoting
Custom boards quoted by hand from a request email.
- 07
Product data
Two versions of the product record, re-keyed at each step to the catalogue.
- 08
Knowledge and governance
What the business knows sits with individuals, and no AI policy was written.
- 09
Integrating an acquisition
Each acquisition runs as a new project; the data step is not yet standard.
A champion in every department, nothing written into the core systems.
Nomad recommended an AI Adoption and Enablement Programme, running from late September to the end of January, and BEX Group agreed it. It puts an AI champion in every department, settles the AI policy and the security and data-access position first, and runs a weekly champion session. Then it works department by department on real workflows: discovery with the champion, a working day with the team, use, refinement and measurement against a baseline taken in the first weeks.
Nothing is written into the systems of record during the programme. Each department keeps what it built, and each cycle records what has outgrown a general tool and belongs in a system of its own. Those backlogs become the roadmap for Orchid, the model of the business, to follow.
The programme starts in late September 2026. We publish outcome figures when we have measured them, so there are none on this page yet.
If you run a group where the systems are trusted and the work between them is done by hand, that picture is what an Operational Audit gives you: every connection mapped, a written readout you can act on, and the scope of whatever comes next.