Locked corporate IT is common in construction. Many firms run inside a closed ecosystem (think Samsung-style, or any approved-device and approved-app estate). That protects the business. It also makes it hard to keep pace with AI.
If your people cannot use non-approved tools, you are not “behind because you are lazy.” You are blocked by a real control. The risk is that competitors and clients keep moving while your team waits for a policy that never arrives.
Accendo Consultants works with construction and contracting SMEs on ISO and HSEQ systems that still have to work in the real world. The same pattern shows up with AI: security is not optional, and neither is learning. You need both.
The problem is speed versus control
Approved ecosystems are great at containing risk. They are slow at admitting new AI tools. Procurement, cyber, and information security all have good reasons to say no. Meanwhile the tools (ChatGPT, Grok, Claude, and the rest) change weekly.
So site and office teams either:
- wait and fall behind
- or use personal accounts on phones and create a shadow IT problem that is worse than a managed trial
Neither option is a strategy.
Accendo’s recommendation: name an AI / effectiveness owner
Someone has to own the question. Not “become the IT department.” Own effectiveness: what is worth trying, what is safe enough to trial, what should be blocked, and what top management needs to hear.
That person (or small pair) should:
- understand how the business actually delivers work
- sit close enough to HSEQ / quality / ops that ISO and client requirements are not an afterthought
- translate AI experiments into plain English for directors
- keep a short register of trials, risks, and decisions
Without an owner, AI stays hallway chat. With an owner, it becomes a controlled improvement stream.
Silo a local environment when the main estate cannot move
If the corporate network will not allow GPT, Grok, Claude, or similar tools, do not pretend the need disappears. One practical pattern Accendo recommends is a siloed local environment: a separate, tightly controlled setup outside the main production estate where named people can trial approved AI use-cases.
That sandbox is not a back door into client data. It is the opposite of chaos:
- clear purpose and written rules
- no live project or personal data unless deliberately allowed and logged
- tools and accounts that are known, not random browser extensions
- time-boxed experiments with a written outcome
- a reporting line back to top management
Think of it like a controlled method trial on a construction site. You do not rebuild the whole programme of works to test one new approach. You fence it, supervise it, and bring the evidence upstairs.
What top management should get back
The AI / effectiveness owner should report in management language, not hype:
- what was tried
- what worked for estimating, method statements, toolbox talks, document control, or client packs
- what failed or created risk
- what would be needed to bring a use-case into the approved ecosystem
- what must stay blocked
That gives directors a choice based on evidence, not fear or FOMO.
How this fits ISO and HSEQ thinking
ISO 9001, 14001, and 45001 already expect organisations to manage change, competence, and information. A unmanaged AI free-for-all fights those controls. A named owner plus a siloed trial environment supports them: documented decisions, risk awareness, and improvement without inventing a second paperwork factory.
UKAS-credible systems are lived systems. AI adoption should be lived the same way: proportionate, owned, and evidenced.
If your IT lock is real, start there
If this sounds like your firm, start with ownership and a safe sandbox, not with buying every tool on the market. Accendo can help you frame the role, the rules, and how the trial reports into top management, alongside the lean ISO and HSEQ work you already need for clients and certification.
Contact us today at andrew@accendo.org.uk.
Accendo Consultants: lean ISO, UKAS-credible, built for how you work.
