Skip to main content
ConsultantAI home

The State of AI in Consulting · 2026

How is AI actually being used in consulting?

Three minutes, mostly multiple choice. We are writing a short study on how boutique firms use AI, what they deliberately will not use it for, and who checks the numbers before a client sees them. Your answers go into the aggregate; your name does not.

Anonymous by default. No account needed. Email is optional and only used to send you the published study.

0 of 24 answered · 22 needed for the study · about eight minutes

The firm

Which best describes the firm?
Where is most of the work delivered?
Which service lines matter most to the firm? (select up to four)
Typical engagement value?

Policy and practice

What is the firm’s formal position on AI in client work?
Which tools are approved for client work? (select any)
Roughly what share of consultants use AI in client work, regardless of policy?

This is the number firms rarely have. Anonymous answers only.

Do you tell clients when AI has contributed to a deliverable?

The delivery chain

Which stages does AI touch today? (select any)
Which stages is it explicitly not allowed to touch? (select any)
What share of research and first-draft work does AI do?
What share of final deliverables contains AI-drafted content?

Verification and error

Who checks the numbers in AI-assisted work before it goes out?
How many hours per engagement go on checking or correcting AI output?
In the last 12 months, has an AI-assisted error reached a client?
At the moment of delivery, what share of the numbers have a traceable source?
Six months after delivery, how would you evidence where a specific number came from?
Has a client asked you to evidence the source of a number in the last 12 months?

Economics

What has AI done to the realisation rate on analytical work?
What has it done to the pyramid?
How is pricing changing? (select any)
What is the biggest constraint on scaling AI in delivery?

Two last things

Which benchmark would you most want to see against peer firms? (optional)

This shapes the report. One line is enough.

What would have to be true for AI analysis to reach a client without being checked? (optional)

The report

The full study covers firm type, region and service line: where AI sits in the delivery chain, who checks the numbers, how often an error reaches a client, what share of figures can be traced six months later, and what it is doing to realisation and pricing. Respondents get it free. It will be priced for everyone else.

Optional. We publish aggregates — no firm names, no individual answers, and never anything attributable to you.

The study is written up as aggregate percentages. We do not publish firm names, and we do not publish individual answers.