YC’s latest batch was certainly a lot of ‘maybe AI can do this?’

by Ana Lopez

By watching hundreds of startups at YC Demo Days, you’re not always sure if you’re actually observing patterns or if your brain, as coffee battles monotony, invents them in some sort of business plan pareidolia. This year, however, the theme was pretty obvious: “AI can probably do that! Maybe.”

Today’s AI models are certainly more capable than those of yesterday and before. But we’ve seen time and time again how these systems demonstrate well but fall under systematic requirements or as tools with reliable and repeatable results.

It’s hard not to think of this batch as the forerunners of a coming wave of AI-powered shovelware. Choose a use case, do a little fine-tuning of an available model (no one builds their own model), choose a few good examples for screenshots and set up a prefab UI. Congratulations, you are now the first ever AI social media content generation platform for independent bars and restaurants in the Middle East and North Africa. Buy a few hundred 5 star reviews and you’re on your way!

Now it’s not that restaurants in Cairo and Beirut can’t use a handy tool to gain some traction online and attract new customers. It’s that AI, as it currently exists, doing something for you is like admitting it doesn’t matter.

Creating an AI-powered call agent that picks up the phone at your business sounds great if you describe it as a way to never lose a customer. But what will the customer think if the company they call decides AI is the take they deserve? Personally I would hang up and try someone else. What about a craftsman who gets an AI call to make an appointment? The same.

Realizing that an email to you has been trivially “personalized” by AI is like being told that we don’t bother personalizing our emails, but we want you to think we do. Wouldn’t you feel cheated? It is a systematic deception for the customers.

If your first interview with a company is with an interlocutor or a person clearly reading generated clues from the knowledge base or whatever, do you feel like a person joining a team or part being measured for installation? You are not even worthy of the full attention of a qualified human being.

That’s not necessarily the vibe I got from every AI startup in this YC batch, but I certainly got it from a few of them. Here’s a partial (!) list of the “AI can, probably” companies I’ve written up.

  • Type – AI first document editor.
  • Iliad – Generate game art items.
  • Lay up – Build workflows across apps with a single line command, such as onboarding a new employee.
  • Core – AI-powered onboarding orchestration that understands “the true nature of a business”.
  • Hadrius – SEC compliance robo advisor.
  • Rapid fire – Generated marketing content for SMEs.
  • Quazel – Learning language with an AI teacher.
  • Stand.ai – Generative AI “photographer” for e-commerce.
  • squat – Natural language accountant tools.
  • Berri.ai – Create ChatGPT apps as a service.
  • semantic – Financial news insights “enriched” by AI.
  • Credal.ai – ChatGPT-like interface for employees that references company documents but protects trade secrets
  • Defog – Add AI data assistant to your app.
  • Link handle – Suggests things from the knowledge base and adds to chat or notes live in the browser.
  • Sail – Automated sales emails.
  • Airflow – Automate market research based on reviews and feedback
  • Ten no – Turn knowledge base into a custom LLM.
  • Truewind – AI-powered accounting and financial processes.
  • Flair labs – Collect insights from customer service call records and emails.
  • Just paid – Automate the payment of bills, collect overpayments to suppliers.
  • Kyber – Automate insurance industry tasks such as answering questions and underwriting.
  • Meru – Platform for training your own LLMs.
  • The same day – AI that calls employees such as plumbers and roofers to make appointments
  • Zenfetch – Analyze live customer conversations and surface talking points.
  • in sync – AI to analyze customer emails.
  • Couple AI – Video courses generated using AI.
  • Latent – Automation of electronic health records.
  • avocado – AI receptionist to answer missed calls at SME.

Until about 30 seconds ago, I had added thoughts about the companies to these brief and probably inadequate descriptions. But I realized that the list was in danger of becoming a litany of complaints (not to mention way too long). No one likes to read someone who blasts ideas left and right, especially when many of those ideas are being worked hard on by people to whom they matter. It’s easy to criticize. So easy that someone in the summer batch could try to automate it!

But I challenge you to look at that list and not wonder about some of the entries: Is That really what is needed? Doesn’t that require a lot of supervision? Doesn’t this lead to accountability or less transparency? Has anyone asked customers if they want this? Who verifies and checks the results – another AI? Who is being displaced by these tools? Who trains people on it?

Virtually every company that presented said they went live a few weeks earlier and miraculously already had a healthy ARR. But a few weeks is barely enough time to install a major automation tool and read its documentation, let alone review its performance and judge whether it’s worth the price tag. I can’t imagine even half of this being used, actually used, by a potential customer.

An example I can’t help but share: a company with generative marketing images in its slide had the following prompt to get the system to cooperate: Our Classic Ketchup is made only with sweet, juicy, red, ripe tomatoes for America’s Favorite Ketchup’s signature thick and rich flavor. The copy of the AI: SWEET AND JUICY KETCHUP FOR EVERYONE! If I were a marketer at Heinz and that was in the demo I got, I would stand up, thank them for their time and open the door.

Some companies admitted that they turned around halfway through the program and recently wrote their first line of code for this new application. Of course, we have to consider the adventurous and free-spirited nature of early-stage startups, that’s part of the fun and excitement of the space. But do these companies really feel “innovative” to you? Rather, they seem to be big fans of innovation, sneaking into his room and trying on his clothes. (“Nice…here, try it, fintech.”)

I know I’m underestimating the amount of work it takes to build even the most superficial AI-powered B2B SaaS service, but a lot of this feels like our old hackathons where someone would release an API and everyone would try to put it in shoehorning to the most realistic-sounding application, hoping to get that $1,000 gift card from SAP or whatever. There’s fun in the creation process, but the results don’t really stand alone.

I’ll probably be proven wrong if one of these companies goes unicorn and everyone laughs at the businessupdates.org writer who doubted them. But I can’t shake the concern I felt as I heard founder after founder say with such conviction that their AI could do something better, when I suspect that belief has been cultivated under false pretenses.

Related Posts