AI for consultants: faster synthesis, no shortcuts on accuracy

Consulting is packaged judgment. Much of the labor around that judgment, including research, note synthesis, drafting and formatting, is exactly what AI compresses. The risk is that faster output makes it easier to skip the checks that protect your credibility.

Where AI helps consultants

Getting up to speed. Before kickoff, build a briefing on the client, their market and competitors. Search-plus-citations tools such as Perplexity are useful here; see Perplexity vs Google. Open the sources that will appear in your deliverable. A recurring research workflow can refresh the brief weekly.

Synthesis. After ten stakeholder interviews you have pages of notes. Ask for themes, supporting quotes, points of disagreement and gaps. Then read the raw notes for what it missed or flattened, because a model tends to smooth over the minority view that turns out to matter.

Drafting. Proposals, statements of work, workshop agendas, discussion guides and update emails start as templates and notes. The assistant assembles a draft, and you set scope, price and tone. Good prompts matter; see how to write better AI prompts.

Client analysis prep. Client spreadsheets arrive messy. Cleaning them is a good use of AI, as the Excel guide shows. Verify with control totals before you analyze.

Productizing your method. If you repeat the same diagnostic, you can turn it into a template or a small custom tool for clients, built by describing it in plain English to a tool like Claude Code.

Where AI does not help

• Original insight. It reflects existing material. Your pattern recognition across clients is the product.
• Verified facts. Invented statistics and misattributed quotes are the fastest way to damage credibility. See hallucination.
• Stakeholder politics. It cannot read the room or know who actually decides.
• Client-ready without review. Anything with your name on it needs your judgment.

Risks specific to consulting

Client confidentiality. Most engagements come with NDAs and data-handling terms, some of which restrict third-party processing or require client approval. Read them. Where the answer is unclear, use redacted or synthetic material, or ask the client. The considerations are covered in can I use AI for client work and is it safe to put company data in ChatGPT.

Disclosure. Some clients expect to be told when AI is used, and some contracts forbid it. Decide your practice and put it in your engagement letters.

Cross-client leakage. Keep separate workspaces or projects per client, and do not reuse one client’s material as an example for another.

Accuracy liability. If your deliverable contains a wrong figure from an assistant, you own it. Keep a source for every number.

Starter prompts

Interview synthesis:

Below are notes from 8 stakeholder interviews about [topic].
Produce: (1) five themes with the number of interviewees who
raised each, (2) one verbatim quote per theme, (3) points where
interviewees disagreed, (4) topics nobody mentioned that I
might expect. Mark any theme supported by only one person.

Research brief:

Research [company]. Give me a one-page brief covering business
model, recent developments, competitors and likely strategic
priorities. Cite a source link for every factual claim, and
list what you could not verify.

Your first week

• Day 1: Review your engagement letter and NDA template for data-handling language and decide your AI policy for clients.
• Day 2: Run a research brief on a company you know well and grade it for errors.
• Day 3: Synthesize notes from a past project and compare with your original findings.
• Day 4: Draft a proposal from a template and check scope and price yourself.
• Day 5: Choose one repeatable part of your method to turn into a template.

A worked example: the two-week discovery sprint

At the start of an engagement, you interview a dozen people and read a pile of client documents. If the client’s terms allow it, or the material is redacted, use the assistant to build a running knowledge base: a summary of each document, a table of claims and who made them, and a list of contradictions between sources. After each interview, add the notes and ask what is new, what conflicts with earlier interviews and what questions remain.

By the end you have a structured picture built faster, and a list of contradictions to raise with the client. The interpretation, the recommendation and the story you tell are still yours.

Common mistakes to avoid

• Reusing one client’s material as another’s example. Keep workspaces separate.
• Citing a statistic from a chat answer. Trace it to the original publication.
• Assuming the client is fine with AI use. Ask, and record the answer.
• Delivering an unedited draft. Clients can tell.

What the sources say

Labor statistics for management analysts, the BLS term for consultants, and field-experiment evidence on AI and knowledge work.

• The Bureau of Labor Statistics projects employment of management analysts to grow 10 percent from 2025 to 2035, much faster than the average for all occupations.
• In the Harvard Business School field experiment, subjects using AI completed 12.2% more tasks and completed them 25.1% more quickly on average for tasks selected to be within the AI frontier.
• In the same Harvard Business School experiment, subjects using AI were 19% less likely to produce correct solutions for a complex managerial task selected to be outside the frontier.
• The Bureau of Labor Statistics says government agencies are expected to seek the services of management analysts as they look for ways to reduce spending and improve efficiency.

To build your own toolkit

Learn AI Fast is a two-day in-person class in Boston for non-developers, with a hands-on research session on a provided scenario and a build lab on a provided specification.

Sources

• U.S. Bureau of Labor Statistics, Occupational Outlook Handbook: Management Analysts: “projected to grow 10 percent from 2025 to 2035, much faster than the average for all occupations”
• Harvard Business School, Navigating the Jagged Technological Frontier (working paper, 2023): “subjects using AI outperformed those not using AI, completing 12.2% more tasks and completing them 25.1% more quickly on average”
• Harvard Business School, Navigating the Jagged Technological Frontier (working paper, 2023): “subjects using AI were 19% less likely to produce correct solutions compared with those without AI”
• U.S. Bureau of Labor Statistics, Occupational Outlook Handbook: Management Analysts: “expected to seek the services of management analysts as they look for ways to reduce spending and improve efficiency”