Sitting through hundreds of startups at YC Demo Days, you’re not always sure if you’re really picking up patterns or if your brain, while coffee struggles with monotony, is making them up in some kind of pareidolia for business plans. This year, however, the theme was pretty obvious: “AI can probably do that! Maybe.”
Certainly, today’s AI models are more capable than yesterday’s and yesteryear’s. But we have seen time and again how these systems have good demos but fail to meet systematic requirements or as tools with reliable and repeatable results.
It’s hard not to see this batch as the harbingers of an upcoming wave of AI-powered shovels. Pick a use case, tweak an available model a bit (nobody really builds their own), pick some good examples for screenshots, and bolt on a pre-made UI. Congratulations, you are now the first AI social media content generation platform for independent bars and restaurants in the Middle East and North Africa. Buy a couple 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 just that having AI, as it currently exists, doing something for you is like admitting that it doesn’t matter.
Creating an AI-powered chat agent who answers the phone at your business sounds good when you frame it as a way to never lose a customer. But what does the customer think when the company they call decides that AI is the reception it deserves? Personally, he would hang up and try someone else. What about a business worker who gets a call from AI to make an appointment? The same thing.
Realizing that an email to you has been trivially “personalized” by AI is like being told we can’t be bothered to personalize our emails, but we want you to think we do. Wouldn’t you feel cheated? It is a systematic imposture on customers.
If your first interview with a company is with a conversation agent or a person who obviously reads leads generated from the knowledge base or whatever, do you feel like a person joining a team or a piece that is it being sized for installation? You don’t even deserve the full attention of a qualified human.
That’s not necessarily the vibe I got from every AI start in this batch of YC, but I sure got it from some of them. Here’s a partial (!) list of “AI can do that, probably” companies I noted.
- Guy – AI-first document editor.
- iliad – Generate game art assets.
- tray – Create workflows across apps with one line command, like onboarding an employee.
- Core – AI-powered onboarding orchestration that understands “the true nature of a business”.
- hadrio – Roboadvisor in accordance with the SEC.
- speedybrand – Marketing content generated for SMEs.
- Quazel – Language learning with an AI tutor.
- stand.ai – “Photographer” of generative AI for e-commerce.
- squawk – Accounting tools in natural language.
- Berri.ai – Creation of ChatGPT applications as a service.
- Semantic – Information on financial news “enriched” by AI.
- credal.ai – ChatGPT-like interface for employees that references company documents but protects trade secrets
- Pious – Add AI data assistant to your app.
- linkgrep – Suggest things from the knowledge base and add to chat or live notes in the browser.
- Browse – Automated sales emails.
- air flow – Automate market research based on reviews and feedback.
- can not – Turn the knowledge base into a personalized LLM.
- truewind – Financial and accounting processes powered by AI.
- style labs – Collect information from customer service call and email data.
- just paid – Automate the payment of invoices, detect overpayments to suppliers.
- cybernetic – Automate insurance industry tasks like answering questions and underwriting.
- Meru – Platform for training your own LLMs.
- The same day – AI that calls workers like plumbers and roofers to make appointments
- Zenfetch – Analyze live customer calls and surface talk points.
- sync up – AI to analyze customer emails.
- peer AI – Video courses generated using AI.
- Latent – Automation of the electronic medical record.
- Avocado – AI receptionist to answer missed calls in SMEs.
Up until about 30 seconds ago, I had actually added thoughts about companies to these brief and probably insufficient descriptions. But I realized that the list was in danger of becoming a litany of complaints (not to mention too long). No one likes to read someone just throwing ideas off left and right, especially when many of those ideas are being worked on by people who matter to them. It’s easy to criticize. So easy that someone on the summer lot can try to automate it!
But I challenge you to look at that list and not wonder about some of the entries: Is it that really what is needed? Won’t that need a lot of supervision? Doesn’t this introduce accountability or decrease transparency? Has anyone asked customers if they want this? Who verifies and audits the results? Another AI? Who is displaced by these tools? Who trains people in them?
Virtually every company that came forward said they had gone live a few weeks earlier and, miraculously, were already at some healthy ARR. But a few weeks isn’t enough time to install a major automation tool and read the documentation, let alone evaluate its performance and whether it’s worth the price. I can’t imagine even half of these have been used, actually used, by a potential customer.
One example I can’t help but share: a generative marketing imaging company on their slide had the following prompt to get the system to work: Our Classic Ketchup is made only from sweet, juicy red ripe tomatoes for the characteristic thick, rich flavor of America’s favorite ketchup. The AI copy: SWEET AND JUICY KETCHUP FOR EVERYONE! If I were a marketer at Heinz and that was in the demo they gave me, I would stand up, thank you for your time, and open the door.
Some of the companies admitted that they had pivoted midway through the program and wrote their first line of code for this new application recently. Of course, we must take into account the adventurous and carefree nature of early-stage startups, that’s part of the fun and excitement of the space. But do these companies really seem “innovative” to you? Rather they seem to be big fans of innovation, sneaking into her room and trying on her clothes. (“Nice… here, try it on, fintech.”)
I know I’m underestimating the amount of work it takes to build even the most superficial AI-powered SaaS B2B service, but a lot of these feel like our old hackathons where someone would make an API available and everyone would try to shoehorn it. to the most realistic-sounding app, hoping to get that $1,000 gift card from SAP or whatever. There is joy in the creation process, but the results do not stand on their own.
I’ll probably be proven wrong when one of these companies goes unicorn and everyone laughs at the TechCrunch writer who doubted them. But I can’t help feeling the concern I felt hearing founder after founder say with such conviction that their AI could do better, when I suspect that conviction has been cultivated under false pretenses.
