What is Generative AI?

By now, I’m guessing you’ve seen (or at least heard of) ChatGPT. If you haven’t played with it yourself, I recommend you stop reading this article right now and have a go here.

You’ve seen the countless LinkedIn posts proclaiming the new wave of AI is here and how automation is going to put us all out of work. Maybe you’ve seen the announcements (here, here and here are some) that Big Tech is pushing generative AI into all of its products. Or heard politicians in the news calling for AI regulation.

What’s going on and why is everyone talking about AI all of a sudden?

In a nutshell, there’s been a step change in the quality of the technology. We can do things now with AI that we couldn’t before, and it’s worth your time to understand how this is going to impact your business.

Generative AI refers to a type of artificial intelligence that is capable of creating something new. Unlike other types of AI, which rely on feeding in pre-existing data to be transformed, generative AI uses algorithms to generate entirely novel content, such as music, images, video and text. And it does this with a very simple natural language interface (i.e. you can give it instructions as if you were writing to a co-worker).

I worked in consulting for ~10 years. I’ve spent the last ~6 months at South Park Commons delving into Large Language Models (LLMs) specifically and generative AI more broadly. There is no doubt in my mind that generative text and images represent a tidal shift for our industry.

How Can You Apply It to Your Business?

If you speak to 100 consultants and ask them what they do, you’ll get 100 different answers. So don’t bite my head off for this take. But one way of thinking about the industry is that we tailor research insights to a specific problem our clients have - designing, executing, synthesising and communicating that research is what we get paid a lot of money to do.

Out of all these phases, research execution is low value-add, repetitive and time-consuming. It also tends to be donkey work. Ask any analyst at a consulting firm and they’ll have plenty of war stories of late nights in the white-collar trenches aligning boxes, cleaning interview transcripts or building out yet another scenario in the market model. But it’s necessary to get from a blank page to a final report.

A lot of that research execution involves information processing. Writing things down, summarising them, expanding on them. Cleaning, classifying and grouping data points. Explaining charts in text and quantifying spoken word. And a whole lot of coordination: amongst the team and across vendors. For these use cases, generative AI is fantastic.

At Junior, we’ve spent a lot of time thinking about how we can apply generative AI to consultants’ workflows. And we’ve started with helping you run interview workstreams more efficiently. In our view, this is the most common, highest value-add and most disorganised work that consultants engage in - we’re on a mission to make it 10x faster and easier for you.

We think generative AI will be applicable to all of the work consultants do: web surveys, secondary research, data analysis + visualisation, modelling + forecasting, pitch writing, project management slide production etc. Our vision is to build out modules that cover all of the consultant activity stack, cutting out the grunt work endemic to all of those activities to let you focus on the bigger picture.

Now - you might say - these use cases are interesting, but why should I bother? Ultimately, what’s the bottom line for my business?

• As a result of doing less research execution, the quality of work delivered will drastically increase. Not because the AI is doing it for us but because consultants will be able to test more hypotheses to a deeper level in faster iteration cycles.
• Consulting will become materially more profitable as firms change the way they staff teams. We estimate that for firms that apply gen AI to their operating model, gross margin on a per project basis will increase by 10-15pp over the next 3 years.
• The industry will see higher staff retention rates and overall morale improvement, as the worst parts of the job are eliminated.
• Competition will rapidly accelerate even while the market for advisory will continue to grow robustly. Firms who do not embrace this technology will find themselves at an existential disadvantage to AI-first consultants - slower to market, weaker in delivery, and more expensive than competitors.

What Should You Be Concerned About?

It can’t all be sunshine and roses. There are very real risks associated with AI adoption that you should start thinking about asap.

• Information security: I know it might go without saying, but please do not use ChatGPT in conjunction with sensitive client information. Having spoken to many consultants over the last 6 months, I can assure you that some on your team are. OpenAI has changed their data policies when it comes to use of the model APIs in a B2B application (like Junior) - they no longer collect this data for training of future models. This is not the case for ChatGPT — anything you put in there, consider it leaked.
• Hallucinations: Generative AI models are probabilistic, not deterministic. That means there are cases — particularly when it’s not quite sure what the answer is — where it generates incorrect or partially correct information. There are a number of methods for hallucinations to be reduced to near zero, but they are nevertheless a risk. You should not rely on the AI to give you the right answer — always double check its work! At Junior, we’ve built data lineage and audit capabilities into the application at all stages, so you know exactly where the AI got its answer from.
• Appropriate training: a very common refrain we hear from the market is what the impact of this is going to be on the professional development of new consultants entering the industry. Some version of ‘well if they aren’t summarising call notes, making slides, or building the model, how will they learn X or Y skill’.

We can have a spirited debate about whether these rote tasks are actually what hones the advisory skillset. But more broadly, this is not a new fear nor is it limited to the consulting industry - there is legitimate anxiety in all sectors of the economy as to how jobs will change with gen AI. From our perspective, gen AI is an enhancement, not a replacement, of human capabilities. At the end of the day this technology is not going anywhere, so we need to work out how to harness it positively. And that will involve lots of staff training and iteration.

AI is not the panacea to all our problems. What it can’t do (thankfully!) is what consultants excel at:

• lateral thinking: ’I’ve seen this in another market’
• reasoning and exercising judgment: ’this advice would not be appropriate for our client’
• prioritisation: ’this data point is not important’
• contextual inference: ’reading between the lines the expert conveyed this’

So we get to keep all the fun bits of the job, and dispense with the donkey work. Sounds like a win-win to us!

How Junior Can Help

Consultants have enough on their plate managing their existing client base without allocating time and resources to experimenting with AI tooling. That’s where we come in.

We know how the consulting industry works inside-out and we’re building software to help you do all of this. Drop us a line if you’re interested in strapping rocket boosters to your teams.

dimitris@myjunior.ai | nico@myjunior.ai