When does an SME need an AI consultant?

Signals that outside help can accelerate the work

The trigger should be an operational problem, not fear of missing out. Look for recurring friction with an owner, volume, and a result the business can observe.

• Pilots exist but never reach daily use.
• Manual work grows with every new customer or employee.
• Several tools hold parts of the same process.
• The team needs temporary execution capacity rather than another workshop.

When to wait

Do not start a consulting project only to 'do something with AI'. First name the process owner, collect a baseline, and make time for employees to test and adopt the change.

Bring in expertise when uncertainty blocks progress

External support is most valuable when opportunities are visible but the team cannot confidently prioritise, design, govern or deliver them. An experienced consultant can prevent months of disconnected experimentation and build internal capability. Even the strongest consultant will struggle without a sponsor, process owner and access to users.

• There are many ideas but no evidence-based order or business case.
• A pilot works, while production, security and integrations remain unclear.
• The company wants an independent choice between products and custom work.
• Internal knowledge exists but delivery capacity or experience is missing.

The right moment after an early experiment

A recruitment firm has used a standalone product to summarise CVs. Staff see value, but information is split between the ATS, inbox and documents. A consultant models the full selection workflow, defines privacy boundaries and designs safe integration. The early experiment is not discarded; it supplies examples for the evaluation set. Outside expertise helps because the journey from demo to dependable operation is substantial.

Prepare the engagement

A small amount of preparation produces better proposals and prevents the first weeks being spent entirely on orientation.

• Describe the business goal and current friction in ordinary language.
• Gather volumes, systems, sample data and known exceptions.
• Assign an executive sponsor and daily process owner.
• Ask for method, assumptions, outputs and knowledge transfer.
• Begin with a defined phase and explicit continue-or-stop decision.

Keep the first assignment assessable

An initial phase should reduce uncertainty and enable a concrete decision within weeks. Judge practical understanding as well as presentation quality.

• A credible baseline and agreed prioritisation.
• Working evidence using representative examples.
• Clarity on cost, risk, dependency and next phase.
• Knowledge and documentation retained by the company.

When not to hire yet

Pause if nobody can make time, the goal is merely to 'do something with AI', or the consultant is expected to make every internal choice. Start with a short leadership session or process review instead. A full implementation without ownership and decision authority usually creates delay rather than momentum.

Run a short readiness check before hiring

A consultant can add expertise and momentum, but cannot replace absent ownership. Confirm that a business sponsor can make decisions, users can contribute time and representative process information is accessible. The technical solution does not need to be known, but the business objective must be concrete enough to test trade-offs. If it is not, commission a short exploration rather than committing to a full implementation team.

• One process owner can set priorities and explain real exceptions.
• The team can provide examples, volumes, current costs and failure scenarios.
• Management has set a decision date, budget range and acceptable risk boundary.

Match the engagement to the uncertainty

A scan or roadmap suits unclear priorities. When direction is known but feasibility is uncertain, a discovery or prototype is more useful. An interim consultant works well when the company needs ongoing internal coordination across several initiatives; a project team fits a bounded implementation. Keep the first engagement no larger than necessary. It should end with evidence and a decision, not an automatic commitment to a long follow-on contract.

• Roadmap: priorities, economics, risk and the sequence of investment.
• Discovery: process design, data assessment, architecture and tested concept.
• Implementation: production code, integrations, monitoring, adoption and handover.

Prepare the first week before the contract starts

Consulting budget is often lost while people wait for accounts, diaries and decisions. Schedule interviews, system access, security requirements and weekly demonstrations in advance. Share documentation, but let the consultant observe how work is actually performed. Agree who removes blockers and how quickly. This makes it clear whether external expertise is accelerating delivery and keeps the internal team involved in decisions and ownership.

• Book the kick-off, user interviews and decision meetings before day one.
• Maintain one shared backlog for questions, risks, assumptions and actions.
• Require weekly demos using real scenarios, including unresolved failures.

Check after two weeks whether external help is accelerating decisions

Measure reduced uncertainty, not just visible activity. A strong consultant should have secured access, validated the process with users, exposed critical assumptions and started a practical test. If questions remain vague or discussion centres mainly on tools, reset the engagement early. External expertise should make the internal team faster and more capable, rather than create a growing dependence on consultancy.

• Decisions, open risks and required client input are visible to everyone.
• Internal colleagues understand why options have advanced or been rejected.
• The next deliverable and its acceptance criterion are unambiguous.