Accenture: Dead Man Walking?
Like most transformational technologies, the new class of generative AI technologies (e.g., ChatGPT) is simultaneously overhyped and underestimated.
The hype I think I can safely skip--it's taken the tech world and media by storm, and hardly a day goes by without a big announcement. If you really want to stay abreast of just how fast things are going, check your Twitter feeds. It's insane and cool and fun and scary and just about everything else all at once.
For underestimation, though, don't look at Twitter. Look at Accenture's stock price over the last six months.
Clearly, Wall Street thinks Accenture is doing just fine.
But what if Wall Street understood that over half of Accenture's revenue was in jeopardy? I wonder if their stock would be doing quite so well.
Could the new AI tools really jeopardize that much of Accenture's revenue?
Let's take a step back and look at this. First, I'm using Accenture as the exemplar. Many other large firms with similar business models are similarly in jeopardy, but we can understand their risk just as easily by examining Accenture.
By and large, Accenture is a well-respected and well-run company, with over $63 billion in revenue last year selling outsourcing and consulting services. At the risk of oversimplifying, the vast bulk of their revenue fundamentally is billable hours (or effectively derived from billable hours). Indeed, one of the key metrics Accenture shares with investors is the utilization rate of its staff (91% in the last quarter).
Moreover, those billable hours are for knowledge work--computer programming, business operations processing, compliance services, etc. The exact type of work that new AI technologies are disrupting.
I've written about the phenomenal productivity improvement that AI can bring to software developers: https://www.thoughtfulbits.me/p/ai-fragile-systems-the-death-of-brittle. While I've personally seen a tenfold increase in my own programming productivity, even a 20 or 50 percent increase in productivity on average would be incredibly significant--that could translate directly to lower costs (and thus lower prices). What happens when a competitor to Accenture bids on the same software outsourcing projects but 20 percent or 50 percent less expensive?
If the competitive pitch is just price, I would bet Accenture would keep most of its business. Accenture is not the cheapest outsourcing vendor today, not by any stretch of the imagination. With over $63 billion in revenue, they are clearly winning on more than just price!
End to end software projects
There's more to the AI revolution than just programming productivity--doing the same activity more efficiently. AI is also transforming both the how and the what of many business processes. In keeping with the outsourced programming example above, consider the full scope of what's involved in those projects. An engineer physically typing on a keyboard and writing code is just one piece. But there is also project management, documentation, testing, regulatory compliance certification, user training, etc.
Some of these processes, like regulatory compliance, can be incredibly laborious and time-consuming. I've personally been involved in many steps of compliance at various companies and experienced this firsthand. Legal starts with the quarterly requests for a compliance update. This request, in turn, goes to a set of compliance managers, who then go around to the various parts of a company asking for updates. For aspects of compliance involving software, the compliance managers ask software program managers for updates. Those software program managers then ask the engineers for the latest updates.
Writing compliance reports is, of course, the absolute most fun thing any engineer can possibly do.
However, what if that compliance report could be generated at the push of a button? Moreover, for any issues identified, what if the report also showed the engineers how to fix the code to correct those issues?
It'd be a game-changer for compliance management. This capability would be more than just a matter of doing the same activity faster. One could completely rethink the process and eliminate countless hours of drudgery that exist today.
Needless to say, compliance is not the only aspect of software development that is being transformed. New AI developer tools can automatically document entire code bases--and keep that documentation up to date. Tests can be automatically generated, and the often discussed "shift-left" cybersecurity goal (fixing cybersecurity issues in code versus trying to remediate after the fact) becomes dramatically simpler with AI tools. The latest AI developer tools not only automatically find cybersecurity bugs but also provide fixes to correct the issues.
On the most recent earnings call, the Accenture CEO, Julie Sweet, spoke at length about their work with legacy systems. Historically that's been a source of competitive advantage for Accenture--they have the teams and expertise to deal with older and often outdated technologies. But what if AI tools could rewrite legacy software into more modern technologies?
These aren't theoretical "what ifs?". These AI-driven tools are available now (full disclosure--my company Polyverse makes some of them!), and the tools are getting better at a phenomenal rate--literally weekly in some cases.
Of course, the Accenture leadership team knows about these advances in AI capabilities--Julie discussed this on the aforementioned investor call, for example. Accenture's challenge, though, is what to do about it.
Right now, Accenture is doing a lot of talking and hand-waving about AI, but they have yet to make any fundamental changes to their business.
Somebody else is going to do it to them.
Here is my prediction. There will be a number of smaller, nimbler outsourcing firms that fully and aggressively embrace these new AI technologies. They will use those newfound superpowers to bid against Accenture and other "legacy" outsourcers. Only these new pitches are not just on price--they are on the complete package. An AI turbo-charged outsourcing provider could offer better software delivered faster, fully compliant, and better tested and documented--all at a dramatically lower price than legacy providers like Accenture.
At first, these competitors will just nibble at the edges. The pitches will seem too good to be true! And even when the pitch is believed, enterprise sales is still a time-consuming and lengthy process--at least so far, I have not seen any AI technologies that make the enterprise sales process any faster!
At some point, though, likely within a year, those nibbles will turn into a full-on piranha feeding frenzy.
Accenture and the other big public firms will do a lot of press, marketing, and other hooplas about their own adoption and embracing of AI. Ultimately they are trapped by their success. If being competitive in the future means dropping revenue by half, can they do it? Can they find enough new customers and enough new projects to make up for the shortfall fast enough?
It's not just a financial question, either. Culturally, these firms have billable hours ingrained in their corporate DNA. If you are an employee there, that's how you get paid, get your bonuses, get promoted into management, and so on. Shifting that billable hours mindset to a "how do you do this faster for less cost" mindset could be challenging.
Remember, this AI revolution is not a one-shot deal "Oh, let's learn how to use this new hammer." AI is progressing at a furious pace. In software programming, last year, AI tools were essentially at a level of better auto-complete. By this past winter, they were able to write large blocks of code. Now state-of-the-art is full-on code conversion, security testing, and compliance checking. Where will these tools be a year from now?
It's not just AI programming that is progressing rapidly, either. In November of 2022, ChatGPT 3.5 could pass the legal bar in the bottom 10%. In March 2023, ChatGPT 4.0 passed the bar in the top 10%. Similar rapid advances are being made in image generation, video generation, and so forth. Where will be a year from now?
Delivering value to customers as an AI turbo-charged provider is an entirely different mindset than how many billable hours. It's about continuously improving both efficiency and capability.
With Polyverse, we're fortunate to be working with a number of partners that are fully embracing this new AI-driven mindset. There is a palpable excitement and drive--they all see billions of dollars of opportunity from legacy providers ripe for disruption.
ThoughtfulBits: Ideas that Matter is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.