Hello there 👋
Welcome to today's edition of the Breve - the only newsletter you need to get product and growth insights.
🔔 A quick bit of news before we get into today’s topic.
I am hosting a virtual event for product people in Abuja, Nigeria, in November. Register here to attend.
For speaking engagements, please email me at durotoluwaolumide@gmail.com
A lot has been happening in the AI space recently. There’s a new thing to catch up with daily. One thing I have noticed is that AI is fast becoming a commodity (at the developer & application layer) rather than a differentiator.
The barrier to entry is low (thanks to the APIs of large language models), which means that many new entrants can quickly build AI products or incorporate AI into their existing solutions.
When technology is commoditized, distribution becomes super important. If a new entrant is building an AI product likely to be built by an incumbent, it is going to be fighting a hard battle. This is because incumbents have:
Distribution to get the product in front of existing users;
Proprietary/ customer data to make or train good models;
Capital to spend on expensive LLMs;
Talent to build quickly.
Take Intercom’s AI chatbot (Fin), for instance; they easily leveraged their distribution to get the AI chatbot to millions of users. Their AI model could also digest existing customers' content because of their distribution.
They don’t have to find new audiences or spend seed capital experimenting with various channels. Their distribution became a defensible moat for the AI feature.
This does not mean that new entrants don’t have an opportunity. But they have to play the AI game differently. I wrote about the mistakes startups make when implementing building AI features here. In addition, new entrants don't have the reputational risks. So, they have better chances by focusing on AI products that:
invalidate the existing user experience of incumbents,
force incumbents to rework their infrastructure and distribution,
need data the incumbents don't have,
destroy value for incumbents,
focus on customer segments that incumbents cannot serve or
pursue an AI-first strategy that introduces new technology that doesn’t exist before
An example of this is Synthesia, an AI video platform where users can choose from among 140+ image avatars and 120+ voice options with different looks, languages and accents to create videos.
With this, users don’t need to go through the traditional process of creating a video recording of a human speaking and editing the video. Loom, for instance, will struggle to create this kind of AI-first product because that will invalidate the existing user experience on its platform.
When you look at Loom AI (the AI product of Loom), the AI feature is simply optimizing the legacy user experience (by creating title, summaries, and tasks) for videos created.
It is not reworking or redefining the user experience like Synthesia is doing. This is one way new entrants without distribution can play the AI game differently and stand a chance to win.
That’s all for today.
If you enjoy reading this, share it with your friends. Have a great week ahead!
Olumide ✌️