With hundreds of startups at YC Demo Days, you’re not always sure if you’re actually catching patterns or if your mind, while coffee grapples with monotony, invents them in a kind of business plan pareidolia. However, this year the theme was very clear: “AI can do it, probably! Maybe.”
Today’s AI models are certainly more capable than the models of yesteryear and yesterday. But we have seen time and time again how well these systems are demonstrated but fall under methodological requirements or as tools with reliable and reproducible results.
It’s hard not to think of this collection as a precursor to the next wave of AI-powered shovel tools. Pick a use case, do some fine tuning of the available template (nobody really builds their own), cherry pick some good example screenshots, and install the ready UI. Congratulations, you are now the first AI social media content creation 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 useful tool to gain some online traction and attract new customers. Having AI, as it currently exists, do something for you is a kind of acknowledgment that it doesn’t matter.
Creating an AI-powered conversational agent who answers the phone in your business sounds good when you frame it as a way to never lose a customer. But what does a customer think when the company they name decides that AI is the reception it deserves? Personally, I’d hang up and try someone else. What about a trade worker who gets a call from Amnesty International to set up an appointment? The same.
Realizing that an email sent to you has been trivially “personalised” by AI is like telling us that we can’t be bothered to personalize our emails, but we want you to think we do. Don’t you feel cheated? It is a systematic hoax on customers.
If your first encounter with a company is with a chat agent or someone who obviously reads signals generated from a knowledge base or something else, do you feel like someone who joins a team or part of a team that gets built? You don’t even deserve the full attention of a qualified human being.
This isn’t necessarily the feeling I got from every AI startup in this YC group, but I’m sure it got from a few of them. Here is a partial (!) list of companies where I noted “AI can do it, probably”.
- He writes – AI-first document editor.
- The Iliad – Create art assets for the game.
- lies down Create workflows across apps with a single line command, such as assign an employee.
- nucleus AI-powered internal organization orchestration that understands the “real nature of business.”
- Hadrios – Robo-adviser for SEC compliance.
- Speedbrand – Creating marketing content for small and medium businesses.
- Quazel – Learn the language with an artificial intelligence tutor.
- Booth – Generative AI illustrator for e-commerce.
- sack Natural language accountant tools.
- Wild – Create ChatGPT applications as a service.
- pertaining to semantics – “Personalised” financial news insights powered by AI.
- Credal. ai – ChatGPT-like interface for employees that references company documents but protects business secrets
- defog – Add an AI data assistant to your app.
- linkgrip – Suggests things from the knowledge base and adds to chat or notes directly in the browser.
- feather Automated sales emails.
- aiflo Automate market research based on reviews and comments
- Tenner Converting the knowledge base into a dedicated Master.
- fact Bookkeeping and finance operations supported by artificial intelligence.
- Flair Labs – Gather insights from data of customer service calls and emails.
- I just paid – Automating invoice payments, capturing overpayments to vendors.
- kipper Automate insurance industry tasks such as answering questions and underwriting.
- Miro – A platform to train your LLMs.
- Sameday – Artificial intelligence that calls workers such as plumbers and roofers to make appointments
- Zenfitch – Analyze live customer calls and surface talking points.
- synchronous – Artificial intelligence to analyze customer emails.
- AI pairing – Video courses created with artificial intelligence.
- potential Automation of electronic health records.
- Avoca – An AI receptionist to answer missed calls in SMEs.
Up until about 30 seconds ago, I had already appended thoughts about companies to these brief and likely inadequate descriptions. But I realize that the list is in danger of becoming a litany of complaints (not to mention very long). No one likes to read someone who is just dropping ideas left and right, especially when many of those ideas have been worked so hard on them by people who are important to them. It’s easy to criticize. It’s very easy for someone in the summer batch to try to automate it!
But I dare you to look at that list and not wonder about some of the entries: Is that Really what is needed? Wouldn’t that need a lot of supervision? Doesn’t this lead to liability or less transparency? Has a customer asked if they want this? Who checks and reviews the results – another AI? Who displaced these tools? Who trains people for them?
Practically every company that submitted said they started working a few weeks ago and miraculously were already in some healthy ARR. But a few weeks isn’t even enough to install a major automation tool and read the documentation, let alone evaluate its performance and whether it’s worth its price. I can’t even imagine half of it has already been used by a potential client.
One example I can’t help but share: A marketing image generator company had in their slide the following prompt for the system to work with: Our Classic Ketchup is made with only ripe, sweet, juicy, and red tomatoes for the thick, rich taste of America’s favorite ketchup. AI version: SWEET & JUICY KETCHUP FOR ALL! If I were a marketer at Heinz and it was in the demo I received, I would stand up, thank them for their time, and open the door.
Some companies admitted that they moved halfway through the program and wrote the first line of code for this new application recently. Of course we should allow for the free and adventurous nature of early-stage startups, and that’s part of the fun and excitement of space. But do these companies really feel “innovative” to you? They seem rather big fans of the innovation, sneaking into his room and trying on his clothes. (“Nice…here, you can try it, fintech.”)
I know I underestimate the amount of work it takes to build even the most AI-powered B2B SaaS service, but a lot of it feels like an old hackathon where someone makes an API available and everyone else is trying to dump it. To the most realistic application, hoping to get a $1000 gift card from SAP or something. There is joy in the creation process but the results don’t really stand on their own.
I’m probably wrong when one of these companies goes unicorn and everyone laughs at a TechCrunch writer who doubted them. But I can’t shake the anxiety I felt upon hearing founder after founder say with such conviction that their AI could do something better, when I would suspect Conviction was developed on false pretenses.