#PowerToTheConsumer: Insights from 30 Leading Consumer AI Founders, Operators and Thinkers
By Vivian Cheng and Ipshita Agarwal
We recently hosted an intimate consumer AI dinner with AGI House, bringing together some of the most prominent product and engineering leaders from OpenAI, Google X, Notion AI and Adobe Firefly, with 20 incredible early stage consumer AI founders.
Throughout the evening, we tackled topics like:
- Building a moat in consumer AI.
- Ways to go-to-market.
- Valuing AI products’ Discord communities.
- The importance of proprietary data.
- Use cases for consumer AI, beyond the obvious use cases of search and personal assistants.
Here are five of our key consumer AI insights sourced from our audience:
- Network Effects, Customer Experience and Brand Are Still the Moat:
Some founders half-jokingly said “there is no moat.” Most believe data, even proprietary data, is not a long-lasting moat even though it may be an effective barrier to entry. To build a long term moat, network effects are still key — proprietary data may help engineer these network effects by helping quickly facilitate go to market. OpenAI not only had the best data set, it was also first-to-market and built a brand on the back of its papers and founders’ profiles.
Other than proprietary data, being first-to market (e.g., JasperAI, Character.AI), having engaged communities (e.g. Midjourney’s Discord) and delivering a magical customer experience are other viable pathways to building long term moats.
2. Incumbents Have Distribution, Startups Have Shipping Velocity: Who Wins?
Some view this as “incumbents adding a feature” is never as good as “reimagining ways of doing things” using AI. Another way of framing this might be “spreadsheets with AI” versus, “Does this even need to be a spreadsheet?” In the initial wave, incumbents’ distribution advantage seems to have the upper hand. Beyond incumbents, OpenAI’s plugins have eliminated many chatGPT-wrapper startups. A true platform shift, one which enables startups to reach millions of consumers (like mobile or the App store), can upend this in startups’ favor. Shipping velocity matters.
3. Language is Not Enough and Compute is a Limiting Factor
Models trained on words and text alone cannot match human understanding. Language is a way of expressing understanding and knowledge, but it is not the same. The knowledge contained in the context, emotion and images cannot be replaced by language. Language in LLMs is being used as a target or even as a goal when, in fact, it should be a measure. Further, compute is a bigger limiting factor than we realize… especially as it relates to startups capturing proprietary data sets.
4. The Future of Work: No company will need more than a thousand people, generative AI ()) will create more jobs than it will eliminate and there will be an explosion of entrepreneurship.
On the one hand, GenAI tools like GitHub Copilot and AutoGPT are exponentially boosting productivity — one founder said they’re already thinking about, “Does this job need two engineers or can I use Copilot?” Their view is that future companies will have fewer employees. In the same vein, democratization of coding will lead to an explosion in entrepreneurship. On the other hand, some think that GenAI will create more jobs than it eliminates — especially ones that require context, subjectivity and reasoning.
5. And some hot takes from our attendees… (Our discussion was under the Chatham House Rule, so identities are cloaked.)
“You don’t need an AI bestie — tech still needs to transcend the uncanny valley where it feels just as good as talking to another human.”
“People are using AI search engines to find friends and dates… every consumer product culminates in a dating app.”
“Vertical software is dead since a general purpose AI can do everything.”
“Most consumers use chatGPT to ‘goof around’ — use goofing around to acquire and educate consumers.”
“User interfaces will no longer exist, your device is used to interact with agents only.”
There is a ton of energy in infrastructure and enterprise (check out another recent post from Team CRV for more on that front, #PowerToTheAIBuilder — AI Will Be Core to Every Software Application), but platform shifts like AI present an upheaval as well as an opportunity for consumer investing. Consumers are typically early, first adopters — and we’re excited about the application layer for both consumer and prosumer/SMB products. We can’t wait to see how creative founders innovate in this space beyond the early, obvious use cases. As we all know, the next big thing will look like a toy, and we embrace the slightly crazy/weird. If you’re building in consumer AI, our team would love to hear from you.
