I recently went back and reviewed one of my early GenAI-assisted consulting engagements, not to produce a marketing number or claim some AI magic but to focus in on a more useful question.

Where did we save time?

After reviewing my timesheet entries, comparing the work against a typical manual consulting approach, and discussing the engagement pattern with the client, the estimate came out at roughly:

• 56% total time saved overall

• 55% reduction in consulting time

The math looks interesting when I looked at the data because a fully manual process would have included more client resources but we were able to keep a smaller team due to being able to cycle through iterations more quickly and require less downstream resource time.

A second, very similar project then saved more than 20 additional consulting hours because the process, tools, and prompts had already been worked out - and also saved significantly on duration as well as effort.

What is it?

This was a real consulting engagement, not a theoretical productivity exercise and the work involved the usual consulting pattern:

• Kick-off and scoping the engagement along with interviews,
• Preparing and running workshops
• Analysing data and reviewing documents
• Researching external options
• Preparing reports
• Regular client meetings
• Final presentation and closing out the work.

The biggest time reductions did not happen everywhere, but showed up in specific areas:

• Workshop preparation
• Data analysis and thematic mapping
• Internet research
• Report preparation
• Final presentation preparation

It saved time in those areas where consultants often spend a lot of time turning messy inputs into useful structure.

• Interview notes
• Workshop outputs
• Search results
• Strategy and Policy document reviews
• Thematic analysis
• Option development and recommendations
• Drafting reports

Having said this it did not remove the need for judgement - if anything the opposite happened - it created more space and time for judgement to be applied. It did not replace the time spent talking with the client or the requirement for document reviews - or the associated accountability.

Real productivity gains are rarely smooth across the whole engagement. They appear unevenly, depending on the type of work, the quality of inputs, the maturity of the process, and how much reuse is possible.

What does it mean from a business perspective?

The business value is real with a few practical lessons.

• GenAI savings are strongest where the work involves synthesis. Analysing interviews, identifying themes, summarising source material, preparing reports, and turning raw input into structured outputs are natural fit areas.
• The human work does not disappear. Kick-off meetings, interviews, feedback sessions, client review, and decision-making still matter. In many cases, they matter more because the consultant has to validate, interpret, and shape the AI-assisted output (and we have more time for that).
• The first project creates the learning curve. The second similar engagement saved even more time because the process, tools, and prompts were already in place. That suggests the value increases when organisations treat GenAI as a capability, not as a one-off.
• The savings are not evenly distributed. Some tasks barely changed. Some tasks changed significantly. A few may take longer, especially where extra quality control, handover, or documentation is needed.
• The consulting model starts to shift. If a consultant can deliver the same or better outcome in materially less time, the value conversation changes. It becomes less about hours consumed and more about outcomes, judgement, reuse, and speed to decision.
• There is a governance angle. Faster output is only useful if it remains accurate, defensible, and appropriate. The consultant still needs to manage quality, assumptions, source material, client context, and risk.
• Clients may benefit multiple times. They may see lower effort on the immediate engagement, but they may also gain a reusable process that improves future work as well as engagements that take less time to complete.

What do I do with it?

For consultants, I would suggest a simple approach:

• Map the work before applying AI. Break a typical engagement, process, or internal workflow into stages. Look at where time is spent, where people get stuck, and where outputs are repeatedly created from similar inputs.
• Separate judgement from production. GenAI may be very good at producing drafts, summaries, mappings, and options. Humans still need to make decisions, validate findings, manage relationships, and own the outcome.
• Look for repeatable patterns. The second project is often where the bigger gain appears. If you build prompts, templates, review steps, and reusable workflows, the value compounds.
• Track time honestly. Do not rely on vague productivity claims. Compare actual effort against a reasonable baseline. Include both consultant time and client time where possible.
• Expect uneven results. Some tasks will improve dramatically. Others will not. Some may even take more time because of review, documentation, or governance needs.
• Build quality control into the process. Faster is not better if the output is wrong, shallow, or unsupported. Review steps, source checking, assumptions, and human sign-off still matter.
• Think about pricing and value models. For consultants, this raises interesting questions. If the work takes less time but delivers strong outcomes, hourly pricing may become a poor reflection of value.

For me, the useful lesson from this review was not that GenAI saved time, which it clearly did, more that the savings were specific, explainable, and repeatable. GenAI did not replace the consulting engagement, it changed the shape of the work. Less time spent grinding through preparation, synthesis, search, and report drafting - more emphasis on judgement, structure, review, client context, and knowing what good looks like. Something I expect to compound and change with the introduction of Agents into the consulting process, such as leaving consulting outputs and insights behind in the form of Agents, delivering even greater value.

That feels like a much more realistic business conversation than the usual “AI will replace consultants” narrative.