IESE AI Club

Role playbook

AI for consulting.

Consulting sold senior judgment and a pyramid doing research, analysis and slides. AI compresses the pyramid. Early-career tasks disappear first, but problem structuring and defending an answer can now be learned faster.

This assumes you have the base. Start there if not

The job

What this role actually does.

Six things every week. Model releases do not change them; this page should still hold in three years.

01Scoping 02Research 03Analysis 04Interviews 05Storyline 06Delivery

Most examples are vendor customer stories. Their self-reported, unaudited numbers show something was built and used—not a benchmark to expect.

01

Proposals and scoping

Drafting
The use case
Drafting the proposal and stress-testing the scope before it reaches the client.
Job to be done
When I write a proposal, I want a fast first draft and an honest critique from the client's side of the table, so I find the weak assumption before they do.
Value to the business
More proposals per partner-hour, and fewer engagements scoped into a loss.
How to evaluate it
Proposal turnaround, win rate, and margin variance against the scoped plan. Pricing and commitments stay a human decision.

What AI does here

Keep credentials, cases, biographies and methods in one library; generate the tailored draft. Use AI upstream to challenge the stated problem and expose questions before you commit to a scope.

Workflow map

  1. InputCapture the client brief, discovery notes, relevant credentials and commercial constraints.
  2. AI-assisted processRetrieve relevant case material, draft a proposal structure and identify unanswered scope questions.
  3. Human checkpointEngagement lead tests feasibility, economics and the client’s actual decision need.
  4. OutputA tailored proposal with a clear problem statement, workplan, team and assumptions.

What is still yours

Scope honestly. Overpromising is paid for by an analyst in month three. Decide what can be delivered and tell the client when they have asked the wrong question.

02

Research and market sizing

Research
The use case
Building the fact base at the start of a case: market structure, players, sizing build-ups, regulatory context.
Job to be done
When a case starts on Monday, I want the landscape assembled by Tuesday, so the team argues about the answer rather than about gathering.
Value to the business
Days recovered at the front of every engagement, where they are worth the most.
How to evaluate it
Time to a usable fact base, and the proportion of figures traceable to a named source. Sizing assumptions get stated explicitly, not buried in a generated paragraph.

What AI does here

AI accelerates landscape scans, competitor profiles, regulation, sizing and benchmarks. Insist on citations and open them: an unverified sizing build-up is worse than none.

Workflow map

  1. InputSet the market definition, decision question, time period and acceptable source types.
  2. AI-assisted processSearch, cluster findings, extract cited figures and assemble alternative sizing build-ups.
  3. Human checkpointConsultant opens primary sources, validates logic and rejects weak or incompatible data.
  4. OutputA traceable market view with assumptions, ranges and source notes.

What is still yours

Source quality and structure. Check whether a market number is real or recycled. Choose the build-up logic: sizing is a modelling decision before it is research.

03

Data analysis and modelling

Analysis
The use case
Interrogating client data conversationally, building the first cut of a model, and testing what the numbers actually support.
Job to be done
When I have the client's data, I want to ask questions of it directly, so I test more hypotheses than a hand-built model allows in the time available.
Value to the business
More options explored per week, and a shorter path from data room to recommendation.
How to evaluate it
Hypotheses tested per week, time to a first model, and errors found in review. The analyst reconciles output to source: a wrong number in a client deck is the fastest way to lose the client.

What AI does here

AI speeds cleaning, reconciliation, exploration and first models on messy client data. Use it to find faults in your logic, then verify every claimed fault.

Workflow map

  1. InputSecure the client extract, data dictionary, business question and known data-quality limitations.
  2. AI-assisted processProfile fields, propose cleaning steps, generate exploratory cuts and draft model logic.
  3. Human checkpointAnalyst reconciles outputs to source data, tests edge cases and documents every assumption.
  4. OutputA reviewable model and evidence-led finding suitable for client challenge.

What is still yours

Every number that leaves the building. You are liable for plausible calculations on misread data. Defend any cell to the client CFO.

04

Interviews and expert calls

Meetings
The use case
Capturing and synthesising expert calls and client interviews into evidence the team can search.
Job to be done
When I am in an expert call, I want the record made for me, so I can follow the thread instead of transcribing it, and find it again three weeks later.
Value to the business
A searchable evidence base per engagement instead of notes in twelve notebooks.
How to evaluate it
Time from call to shareable synthesis, and how often the team retrieves a prior call rather than re-running it. Settle what may be recorded and processed before the call, not after.

What AI does here

AI synthesises interviews: where stakeholders agree, executives contradict each other and themes recur without priority. It also drafts sharp, interviewee-specific follow-ups.

Workflow map

  1. InputAgree consent, interview objectives, participant context and a focused discussion guide.
  2. AI-assisted processTranscribe calls, extract themes and contradictions, and suggest evidence-based follow-ups.
  3. Human checkpointInterviewer checks quotes against recordings and interprets incentives, tone and omissions.
  4. OutputA coded interview record and a defensible cross-interview synthesis.

What is still yours

The interview itself. Getting an employee to tell a stranger what is wrong is a trust exercise. So is judging which stakeholder is telling the truth.

05

Storylining and deck production

Deck
The use case
Turning an agreed storyline into pages: exhibits, drafts, formatting and consistency across a hundred-slide pack.
Job to be done
When the story is agreed, I want the production done, so the overnight goes on the argument rather than on alignment and fonts.
Value to the business
Hours returned per deck, and packs that are internally consistent.
How to evaluate it
Hours per deck and revision rounds before partner sign-off. The storyline stays the consultant's: a generated narrative reads generic to anyone who has seen a hundred decks.

What AI does here

AI accelerates formatting, chart building and page cleanup. For storylining, use it to critique the pyramid, expose gaps and identify pages doing no work.

Workflow map

  1. InputStart with the decision, evidence base, audience stakes and approved visual template.
  2. AI-assisted processPropose an outline, tighten slide headlines, draft charts and flag gaps in the argument.
  3. Human checkpointLead confirms the recommendation, sequencing and political implications with the client context in mind.
  4. OutputA concise decision deck whose claims link back to validated evidence.

What is still yours

The answer and its order. A storyline persuades one audience with known politics. A machine checks logic; it does not know the CFO needs cost before growth.

06

Client communication and delivery

Synthesis
The use case
The connective tissue of delivery (status notes, meeting summaries, follow-ups, handover material), and helping the client's people actually use what you built.
Job to be done
When an engagement is running, I want the routine communication produced consistently, so attention goes to the work and the client is never guessing where things stand.
Value to the business
Fewer surprises, cleaner handovers, and adoption that survives the team leaving the building.
How to evaluate it
Adoption of delivered tools after handover, client satisfaction, and time spent on delivery admin. Bad news is delivered by a person.

What AI does here

AI assembles status updates, steering packs, notes and follow-ups. The gain is time returned to difficult conversations.

Workflow map

  1. InputBring the meeting record, decisions, risks, action owners and prior client commitments together.
  2. AI-assisted processDraft notes, status updates and steering-pack sections; identify unresolved actions or conflicting messages.
  3. Human checkpointProject lead verifies commitments, adjusts tone and delivers difficult messages directly.
  4. OutputA timely, accurate client update with decisions, owners and next dates.

What is still yours

The relationship and difficult message. Consulting is bought on trust and renewed when you tell the client what they need to hear in time.

Start this week

  1. Build one library holding your credentials, case studies and methodology sections, then generate your next proposal from it.
  2. Take the interview transcripts from your last engagement and ask where stakeholders contradicted each other.
  3. Take a deck you are proud of, ask which page is doing no work, and then take the criticism seriously.

Where it fails in this role

  • Anything that leaves the building unchecked. You are liable for the number, and plausible is not the same as correct.
  • The client relationship. Trust transfers when a person takes a risk on your behalf, which is not a deliverable.
  • Genuine novelty. It is trained on how problems have been solved before, and clients pay most for the framing that nobody had.

Other playbooks

All nine playbooks Product management Sales Marketing Venture capital Operations Forward deployed engineer Founder / entrepreneur AI product manager

Now do it somewhere real.

A playbook is a map. The Industry Fellowship is the terrain: a term spent talking to professionals who are implementing AI in one industry, working out where it actually creates value, and building a working prototype against what you find. Most fellows start as beginners.

Explore the Fellowship Apply for the Class of 2028