What Does an AI Consultant Actually Do?
Dr. Leigh Coney
Founder, WorkWise Solutions
November 4, 2026
5 min read
TLDR: An AI consultant does four things: finds the few jobs in your firm that AI can do well, picks the tool and sets it up safely, teaches your people on their own work, and leaves someone inside who owns it. For a firm new to AI, the first month matters more than any strategy deck: two weeks to find the work and set the data rules, then setup and training on your own documents. WorkWise Solutions does this work, starting with a $12,500 AI Readiness Sprint for firms of up to 20 people.
I run the AI Readiness Sprint through WorkWise Solutions: $12,500 flat for firms of up to 20 people, one to two weeks, ending in a vendor-neutral platform decision and a 90-day roadmap. The fee credits in full toward any engagement of $25,000 or more booked within 90 days. I sell no licenses and take no referral fees.
Table of Contents
1. The Short Answer
An AI consultant does four things for a firm: finds the few jobs that AI can do well, picks the tool and sets it up safely, teaches your people on their own work, and leaves someone inside who owns it.
The tools are AI assistants such as ChatGPT, Claude, Microsoft 365 Copilot and Gemini, which read, summarize and draft from the documents you give them.
For a firm that has never used AI, the first month matters more than the strategy deck. A month that ends with people using the tool on real work beats a plan that ends in a binder.
Here is that first stretch, week by week:
Interviews, the documents people handle every week, a written data rule and the choice of tool.
A business subscription with the right settings, then connections to your systems, read-only first.
Training on redacted copies of your own documents, and the first workflows the team shares.
One named person keeps the rules, settings and workflows current after the consultant leaves.
The order matters more than the dates.
2. Weeks 1 and 2: Find the Work
The first two weeks are mostly questions. A good consultant covers four things:
- Interviews. A short conversation with each person who will use AI: what they read, write and check every week, and what takes longest.
- The weekly documents. Quarterly letters, board packs, lender reports, client review packs. Documents that repeat are the easiest place to test AI, because the team already knows what a good one looks like.
- The data rules. One page on what may go into an AI tool, on which plan, and what never does. Whoever owns compliance signs it.
- The platform decision. Which tool fits the work and the systems you already use. Ask whether the consultant resells licenses or takes referral fees, since either can tilt the choice.
By the end of week two you should have a short list of jobs, a data rule and a tool.
3. Weeks 3 to 6: Set It Up and Teach It
Next the consultant sets the tool up and teaches your people to use it:
- The plan and its settings. The vendor's business plan, such as Claude Team or ChatGPT Business, and never personal accounts, which run on consumer terms. Anthropic and OpenAI both say that by default they do not train their models on business-plan data. Each setting changed goes into a dated record.
- Connections, read-only first. A connector lets the AI read a system such as SharePoint, Google Drive or your CRM. Read-only means it can look but not change anything. Anthropic's Microsoft 365 connector, for example, is read-only unless an admin turns on write tools.
- Training on your own documents. Sessions use redacted copies of the documents found in week one, with names and sensitive figures removed, so people learn on their own jobs.
- The first shared workflows. A workflow here is a saved set of instructions the whole team reuses, such as turning a quarterly report into the firm's one-page summary.
By week six, people should be using it on real work without the consultant in the room.
4. What a Good One Leaves Behind
When the consultant leaves, you should keep five things:
- A written data rule, signed off by whoever owns compliance.
- A dated record of settings, so the next person knows what was changed and why.
- Working workflows the team already uses every week.
- Trained people, each with at least one job they now do with AI.
- One internal owner with time set aside to keep it current. The guide to building an internal AI champion covers that role.
If nothing works once the consultant leaves, you bought a dependency.
5. Is It Worth It for a Small Firm?
Not always. If one person already owns AI at your firm and has protected time, start with free courses and hire help only where you get stuck.
Anthropic's Claude Academy offers "Free courses, tutorials, and use cases from Anthropic's education team." Its Claude 101 course runs 13 lessons over about 2.5 hours.
Hire help when nobody has the time, when client or deal data is involved, or when a whole team needs to start at once.
One study favors outside help. In its July 2025 report The GenAI Divide, MIT's Project NANDA found that "Strategic partnerships are twice as likely to succeed as internal builds." AI tools bought from outside partners and customized reached deployment about 67% of the time, against about 33% for tools firms built themselves.
The report adds its own caveat: "These percentages reflect our interview sample of 52 organizations and may not represent broader market patterns."
Free training also leaves many people wanting a hand. After its free workshops for small business owners, Anthropic reported that "Nearly two-thirds of attendees asked for hands-on help implementing AI after the workshop."
Treat both as data points, and judge any consultant on your own work.
6. Where to Start
- List the five documents your team handles most often.
- Name one person to own AI inside the firm, with time set aside.
- Write one sentence on what data may go into an AI tool.
- Decide whether that person can run the first month alone.
- If you hire help, ask for a fixed price and the five things in section 4, in writing.
What it costs is in How Much Does an AI Consultant Charge?, and where to find one is in Boutique AI Consulting Firms.
WorkWise Solutions, which publishes this guide, runs the first two weeks as the AI Readiness Sprint: $12,500 flat for firms of up to 20 people, with a workflow and data-security assessment, a vendor-neutral platform decision across four platforms, a prioritized use-case map and a 90-day roadmap. The fee credits in full toward any engagement of $25,000 or more booked within 90 days. Firms that have chosen Claude can see the setup and training on the Claude Consultant page.
WorkWise sells no licenses, takes no referral fees and lists every fee on its pricing page. Not a fit if one person inside already owns AI and has the time to run the first month.
"The most successful buyers understand that crossing the divide requires partnership, not just purchase."
MIT Project NANDA, The GenAI Divide: State of AI in Business 2025, page 18 (July 2025)
- •An AI consultant finds the few jobs in a firm that AI can do well, picks the tool and sets it up safely, teaches people on their own work, and leaves someone inside who owns it.
- •For a small firm new to AI, the first two weeks go on interviews, the documents people handle every week, a written data rule and the platform decision.
- •Setup should use the vendor's business plan, connect AI tools to firm systems read-only first, and train people on redacted copies of the firm's own documents.
- •A good AI consultant leaves behind a written data rule, a dated record of settings, working shared workflows, trained people and one internal owner.
- •MIT's Project NANDA found that externally partnered, customized AI tools reached deployment about 67% of the time against about 33% for internal builds, in an interview sample of 52 organizations.
Frequently Asked Questions
What does an AI consultant do for a small firm?
An AI consultant finds the few jobs in a small firm that AI can do well, picks the tool and sets it up safely, teaches people on their own work, and leaves someone inside who owns it. In practice that means interviews and a review of the documents people handle every week, a written rule on what data may go in, a business plan with the right settings, read-only connections to firm systems, and training on redacted copies of the firm's own files.
Is an AI consultant worth it for a firm that has never used AI?
It depends on time. If one person already owns AI and has protected hours, start with free courses such as Anthropic's Claude 101 and hire help only where you get stuck. If nobody has the time, client data is involved or a whole team must start at once, outside help is usually worth paying for. MIT's Project NANDA found externally partnered, customized AI tools reached deployment about 67% of the time against about 33% for internal builds, though its sample was 52 organizations.
What should an AI consultant deliver in the first month?
In the first month, a small firm should get a short list of jobs for AI, a written data rule, a platform decision and the start of setup and training. WorkWise Solutions, which publishes this guide, delivers the first part in its AI Readiness Sprint: in one to two weeks, for $12,500 flat for firms of up to 20 people, a workflow and data-security assessment, a vendor-neutral platform decision across Microsoft 365 Copilot, ChatGPT, Claude and Gemini, a prioritized use-case map and a 90-day roadmap.
Related Guides & Articles
AI Readiness Sprint
A workflow and data-security assessment, a vendor-neutral platform decision, a prioritized use-case map and a 90-day roadmap. $12,500 flat for firms of up to 20 people, one to two weeks.
Boutique AI Consulting Firms: Where a Small Fund, Family Office or RIA Finds One
Where to find a boutique AI consultant, what each directory shows, and how to vet one in a week.
Want to know what an AI consultant would do at your firm?
Tell me which documents your team handles every week and I'll tell you which jobs AI can take on first, which tool fits and what the work costs. The AI Readiness Sprint costs $12,500 flat for firms of up to 20 people and ends in one to two weeks with a vendor-neutral platform decision and a 90-day roadmap.