This is what transformative AI means for consulting.
Hope, Hype, Heresay part 3.
Over the last two weeks (part 1, part 2), Joel has shared our pragmatic perspective on where AI is for the consulting industry now.
The tl;dr version:
Some great productivity wins but few examples of AI truly transforming how we work.
Because… Clients use consultants to deal with the last 20% of problems. Specific and niche. Even when the problem statement is common, it's resolution (in the client’s context) is unique.
Dealing with these problems requires a flow of interdependent knowledge tasks. Which consulting teams dance through over the changing beat of their projects.
To help their teams, big firms are building their own versions of ChatGPT. Using their vast and ever growing reserves of knowledge (and cash) to do so.
In our view, boutiques should target the most common services they offer. Break down those workflows and focus on solutions for the end-to-end process.
Let’s get on with explaining what that means and why!
An example - the qualitative analysis process
The process
When we talk about the qualitative analysis process we mean some variant of…
Gather information in different formats (interview notes, transcripts, docs, PDFs, etc) in a single place. Review it to build your understanding. Code/ tag the relevant bits.
Theme the information based on the industry, project type and the client’s specific aims / constraints. Usually means copy and paste relevant bits of information from different sources into a single place (Excel, Miro, etc.). On some projects it may be what you do in your head!
Analyse the themed information. Draw out grounded insights that you can quantify, evidence and present.
While these steps overlap, the final step is where clients see greatest client.
Yet 70-80% of project time is danced away in the steps before. Hours, days, weeks of tedious, manual work.
Transforming that process…
A transformed qualitative analysis process reverses this ratio. Giving teams 70-80% of their project time to work on that final step. I.e., the ‘so-what’ insights, and the critical discussions with clients.
Achieving this transformation needs a solution tuned for the qualitative analysis process.
A single solution that can:
Understand all the different sources of information that you have to work with
Suggest themes for the project that you can amend to reflect what you know and understand
Find information relevant to the themes, in each source (irrespective of file type);
Give you tools to cut, recut, analyse and visualise that information as you need to draw insights.
In short, get you to the point that needs human brains to do the hard, creative thinking. In the space of a few hours not a few weeks.
AI solutions exist to help each part of this process;
Ingestion AI that can gather different data formats in a way that enables users to query them side by side;
AI classifiers that can recall information relevant to themes (even as themes change). These are easy to train for the niche, specific context of your project. Giving you more accurate results, when you need them (e.g., rushing to respond to a stakeholder);
Prompt chains that help your consultants speak to the dataset. To retrieve more accurate information, with fewer queries;
Large language models to process the information and generate responses that are easier to work with.
Caveat. To move from patchwork productivity gains to transformative... these solutions need to come together. Designed and applied for the specific process you want to use them for.
This is good news for boutique consultancies and consultants
Why?
The AI can train quicker on much lower volumes of data. You can get going, by yourself, in a couple of hours. You don’t need the reams of knowledge, resource and cost that building your own ChatGPT requires.
It skips the need for technical data science or ML expertise.
It is easy to adopt in a solution built with UI/ UX designed for a specific process. I.e., it happens ‘in the flow’ of the project not as a separate activity.
In short, it overcomes the biggest barriers to adopting AI in ways that truly transform how you work.
Qualitative analysis is an example of this. It’s one close to our hearts. But it isn’t the only example.
We see a trend of more products like Discy. Designed to fix workflows by matching the right type of AI use case to the process.
With this trend we see a shift away from the opposite. Generic, overhyped AI products in search of a problem!
As this happens, there will be more opportunities to transform your consulting processes. And it’s with this shift that the promise of AI for consultants turns from hype to hope to reality.