What an AI Strategy Consultant Should Deliver in the First Two Weeks

The short version: by the end of week two, a proper AI strategy consultant should have handed you a written audit of where AI fits in your business, a prioritised list of three to five use cases with rough cost and time estimates, and one small thing already live or ready to go live. If you're two weeks in and all you have is a slide deck full of buzzwords and a "we'll circle back", you've hired the wrong person.

Why the first two weeks matter more than most people admit

I've watched a lot of businesses hire AI consultants over the last two years, and I've been the consultant myself more times than I can count. The pattern is depressingly consistent. Weeks one and two get spent on "discovery workshops" that produce a lot of sticky notes and no output. Then week six arrives and the client is still waiting for something they can point at and say "that's ours now."

Here's the uncomfortable bit nobody likes to say out loud: if a consultant needs more than two weeks to tell you where AI fits in your business, they either don't know your industry or they're padding the invoice. A competent consultant who's done this before can walk into a business with 20 to 200 staff, look at the workflows, and within ten working days tell you exactly what to automate first, what to leave alone, and what it'll cost. That's not arrogance, it's pattern recognition built from doing this dozens of times. Novelty is rare. Your invoicing problem is not special. Your customer service backlog is not special. The specifics differ, the shape doesn't.

Day one to three: the actual audit, not a chat

A real audit means the consultant sits with your actual systems, not just your org chart. I did this for a 14-person accountancy practice in Leeds last spring. Day one, I asked for read-only access to their practice management software, a look at three months of client emails in the shared inbox, and thirty minutes with whoever does the bank reconciliation. Not a workshop. Access.

By day three I had a list: 40% of incoming client emails were the same six questions asked in different words, their onboarding process took four separate manual data entry steps across three systems, and their monthly management accounts took one senior person two full days to compile because nobody had ever automated the pull from source data. None of that came from a brainstorm. It came from watching the work happen.

If you want a sector-specific version of what this looks like, the breakdown in what accountants should automate first and what to leave alone covers exactly this kind of triage for finance teams, and the logic transfers whether you're in accountancy, retail, or professional services.

Day four to seven: use cases with numbers attached, not ideas

This is where most consultants go soft. They'll give you a list like "improve customer response times using AI" with no number attached. That's not a deliverable, that's a wish.

What you should get by day seven is something closer to this, using the accountancy example again:

• Client email triage: an AI-assisted inbox that drafts responses to the six repeat question types, saving an estimated 6 to 8 hours a week across the team, build cost roughly £2,000 to £3,500 depending on integration with existing email software, live within three weeks.
• Onboarding automation: connecting the CRM to the practice management tool so new client data enters once instead of four times, saving about 45 minutes per new client, one-off setup around £1,500.
• Management accounts pull: automating the source data extraction, cutting the two-day compile down to roughly half a day, higher complexity, six to eight weeks to build, £4,000 to £6,000.

Three options, ranked by effort versus payoff, each with a real number. That's what "prioritised" is supposed to mean. If your consultant hands you ten ideas with no ranking and no cost, they've handed you homework, not a plan.

Day eight to ten: the thing that goes live

This is the part I feel most strongly about, and it's the part most consultants skip. Somewhere in the first two weeks, something small should be built and working. Not "in progress." Working.

For that accountancy client, by day nine we had a simple AI-drafted response system running on their three most common client email types, connected through their existing helpdesk software, reviewed by a human before sending. It wasn't glamorous. It saved about five hours in its first full week. But it was real, it was measurable, and the team could see it working before they'd committed a penny to the bigger projects.

Why does this matter so much? Because trust in AI projects is earned in weeks, not built in decks. If a business owner has been burned before, and most have been sold something by an agency that overpromised and underdelivered, a small working win in week two rebuilds more confidence than any strategy document ever will.

Day eleven to fourteen: the roadmap and the boundaries

By the end of the fortnight you should have a document, not a conversation, that covers three things.

First, a 90-day roadmap with what gets built when, who owns it internally, and what it depends on. Second, a clear statement of what won't be automated and why. This is the part consultants avoid because it sounds like admitting limits, but it's the single most useful thing in the whole document. For that accountancy firm, I wrote plainly that client-facing tax advice conversations should never be handed to AI, not because the tech can't draft something plausible, but because the liability and trust cost of getting it wrong outweighs any time saved. A consultant who tells you AI can do everything is either lying or hasn't thought hard enough about your business.

Third, a straight answer on ongoing involvement. Do you need this person for another six months, or did they just save you from hiring someone full-time for a problem that needed ten days of proper attention? The distinction between hiring someone for a fixed sprint versus an ongoing role is worth understanding before you sign anything longer term, and it's covered well in the piece on fractional AI officer versus AI consultant, because the two-week deliverable should tell you which one you need.

What good costs, and what cheap costs you

I get asked constantly what a proper two-week engagement like this costs. In the UK, expect somewhere between £3,000 and £8,000 for a genuine audit plus one small live deployment, depending on the size of the business and how tangled the existing systems are. Under £2,000 and you're likely getting a template report with your company name pasted in. Over £10,000 for two weeks with no live output and you're paying for a brand name, not results.

For a fuller breakdown of how pricing works across different scopes, the numbers in what an AI consultant costs are worth reading before you agree a fee, so you're negotiating from a position of knowing what's normal rather than guessing.

The bit that separates consultants from agencies

One thing I'll say plainly because it affects what you should expect in week two specifically: a solo consultant or small firm can move faster in the first fortnight than a big agency, purely because there's no internal handoff between the person who did the audit and the person who builds the thing. I've seen agency clients wait three weeks just to get past the "account manager translates findings for the delivery team" stage. If speed in the first two weeks matters to you, and it should, that's a real factor in choosing between the two, and it's worth reading through before you pick. The comparison in AI consultant versus AI agency for small business lays out when each structure serves you better.

The same speed question shows up differently if you're running a marketing agency yourself and considering bringing in outside AI help for client work rather than internal operations. The needs there, faster briefs, cleaner reporting, less time lost to admin, are covered specifically in the guide on AI consultants for marketing agencies, and the two-week deliverable expectations shift slightly because the "client" inside that engagement is often your own account team.

What to check before you sign anyone

None of this matters if the contract doesn't spell out what week two looks like. I've seen contracts that say "strategic review" with no defined output, no date, and no ownership of whatever gets built. If the audit produces code, integrations, or automations, who owns that IP when the engagement ends? Is the small live deployment yours to keep and modify, or does it break the moment the consultant walks away? These questions belong in the contract, not in an awkward email three months later, and the checklist in what should be in an AI consultant contract before you sign covers exactly the clauses that protect you here.

If you run a coaching or course business rather than an agency or a firm with staff, the shape of the first two weeks looks slightly different again, because the "audit" is often about content and enquiries rather than internal systems. That specific version is worth a look in the piece on AI consultants for coaches and course creators, since the deliverables there focus more on content workflow than operational automation.

The question to ask on day one

If you take one thing from this, take this question and ask it before you pay a deposit: "What will I have in my hands by day fourteen that I didn't have before?" A good consultant answers with specifics, an audit document, a ranked list with numbers, one live piece of automation, a 90-day plan. A weak one answers with process words like "alignment," "roadmapping session," or "discovery sprint" and no actual date attached to an actual output. Ask the question. Write the answer down. Hold them to it.

I do this work for owner-led businesses: the process is set out at hiring an AI consultant and the service detail at AI strategy consultant.

Every industry guide I have written is collected at AI consultant by industry.

Related: ai consultant coaches course creators what to automate first for small businesses.

Frequently asked questions

What should I have received from an AI consultant by day fourteen?

A written audit of your current workflows, a ranked list of three to five AI use cases with cost and time estimates attached, one small automation already live or nearly live, and a 90-day roadmap with clear ownership of what happens next.

Is it normal for the first two weeks to just be meetings and workshops?

No. Discovery should happen through access to your actual systems and short conversations with the people doing the work, not extended workshops. If two weeks pass with only meetings and no written output, that's a sign to raise it or walk away.

How much should a two-week AI strategy engagement cost in the UK?

Expect roughly £3,000 to £8,000 depending on business size and complexity, covering a full audit plus one small live deployment. Under £2,000 usually means a template report; well over £10,000 with nothing live built by day fourteen suggests you're paying for reputation rather than output.

What's a red flag that an AI consultant isn't going to deliver anything real in the first two weeks?

Vague language without dates, no request for actual system access, no mention of a small live deployment, and reluctance to name what AI shouldn't be used for in your business. A consultant who claims AI can automate everything hasn't thought hard enough about where it fits.

Where to check the details

• Anthropic's own documentation
• the NIST AI Risk Management Framework