As a tech professional living in San Francisco, I’m often asked for my take on AI. Is enthusiasm continuing to accelerate or flatten out? Or are we entering the “trough of disillusionment?”
Here’s where the opinions of VCs and tech practitioners may diverge.
While optimism hasn’t bottomed out just yet, many of the shiny objects have lost their luster. Capital expenditures on AI have skyrocketed as companies invest in new AI tools, but we have yet to see those investments pay off in a serious way. Despite this, I’m still bullish on the long-term implications of the latest AI breakthroughs.
Reminiscent of the Cloud
What gives me this confidence? Put simply, it’s history repeating itself. Let’s go back to 2006 when most tech companies housed their data and software in large, expensive on-premise data centers. Then, enter Amazon Web Services with its first offering: A rentable and flexible service for on-demand compute (aka, the cloud). Two years later AWS was still met with skepticism – with the exception of a few companies, including Netflix, Reddit, Dropbox, Airbnb and Pinterest.
Fast forward a few years to 2012 – the peak of the cloud hype. Companies are advertising themselves as “the [fill in the blank] cloud company.” Jump forward another decade and the cloud is powering virtually every facet of our tech-enabled lives.
What’s most interesting to me is that the companies who were most successful at leveraging this emerging technology weren’t those that marketed themselves as cloud companies. The most successful cloud adopters were those who delivered a revolutionary user experience and amazing product leveraging the strength of the underlying technology. (Think of Netflix moving from shipping DVDs to streaming, or Reddit handling insane user growth with ease.)
With AI, We’re in That 2012 Moment
Put simply, here’s what I’m seeing across the tech space:
- Shoe-horning AI into the tech stack, when it isn’t better than the legacy approach. At the end of the day, we just want stuff that will work, whether it’s AI-powered or not. Given a choice between a product that’s 80% good and relies on AI and a product that’s 100% good and doesn’t, any sane person would choose the latter. But the hype cycle can distort these decisions. Twelve months ago, companies were terrified of getting left behind and found budget for AI-based tools. Now, with a better understanding of what is “real” and what isn’t, we’re back to focusing on value.
- Commoditization of AI: I’ve joined a number of meetings with AI-centric companies selling us a range of services. Having listened to the pitch, it seems to me that almost all companies with an AI-heavy offering are playing in a saturated space since the cost to build is so low. AI and the underlying models themselves are no longer a meaningful product differentiator – it all comes down to the product itself. Conversely, the best pitches we’ve heard have been by smart founders, who identify a hard problem and offer a solution – in some areas, enabled by AI. And in their pitches, their product is so good that they typically only mention AI if asked.
- The strength of AI as a brand differentiator is rapidly waning. People are skeptical of products that claim to be “AI for […]”. If you chart the strategic value of AI over time [fig 1], you’ll see that the component layers of the “value pie” look very different as time goes by. A year ago, the brand value of AI was at an all-time high. We’re at, or maybe past, the inflection point at which the value of the technology outweighs the hype.
What’s Next?
So, where do we go from here? The companies that will win are those focused on building a revolutionary product and user experience and leveraging AI as a means to an end. If you’re building products and tools with AI, here’s my take:
Use first principles thinking for your product: Find a hard and valuable problem to solve, then build a great product that users love – using whatever technology is the best fit. If your pitch or value proposition includes “AI,” you might be focused on the underlying technology rather than the value of your product.
Is the tool/product usable? To give a recent example, we evaluated five AI-forward companies essentially offering the same product. What it came down to was not, “Who has the best AI?” but instead, “Who has the best product?” Is it usable? Do they have APIs that we can use to connect to the rest of our tooling? Is it 10x better than what we have today? Most simply, do people internally like using the tool?
Consider today and tomorrow: You have to be differentiated in the medium term. Ask questions like: Are you building a platform that can hook into your other traditional and AI-based tooling? Do you have the strongest roadmap for the next two to three years? Remember, you’re not just building a product for today, you’re building a product for three years down the line.
My Optimism Remains
I’m optimistic about the impact of AI, particularly in doing work that takes slow, low-level, mundane work off people’s plates – translation, meeting summaries, and knowledge capture for example. I’m also excited to see how people will build revolutionary products and product experiences, like ChatGPT and what we’ve seen in the code generation space.
Now, let’s fast forward two years. It will be extremely interesting to see which “big bets” on AI struck it big. Which of them will have a real impact on our day-to-day work? That remains to be seen.
