My Startup Idea Got Into Y Combinator

And why you could build a better version with AI

On Saturday I shared a deep dive on the 5 jobs of a founder/CEO with members. Get a 14-day free trial to get access to my library of 60+ deep dives, and a new one in your inbox each Saturday.

Today at a glance:

  • Opportunity → Ad-hoc Payment Reconciliation

  • Framework → Support-driven Development

  • Tool → No Code AI Toolkit

  • Trend → Layoffs

  • Quote → A Powerful Hiring Question

💡 Opportunity: Stripe Ad-Hoc Payment Reconciliation

This comes from an actual problem I came across this week while building Megaphone, my engagement network to help founders grow their followings and generate leads on X and LinkedIn.

PS: this is a fresh idea. If you’re interested in the one that got into Y Combinator (and is actually even more viable today) scroll down to the trend section).

Anyway… we built a custom payment gateway so that users could pay within the app instead of being taken to a Stripe Checkout page. We then realized that custom payment gateways work well for subscription payments but not one-off payments (we have options for both).

The surprising issue: there’s no direct way to attach a payment to a non-subscription product you’ve set up in your Stripe account. For a business that sells multiple types of products this makes reporting challenging.

I didn’t believe it at first but after asking some friends each one said they’d also run into this when using Stripe’s payment intent API. Their solutions were to either:

  • Pass the name of the product to Stripe in the payment’s metadata

  • Using Zapier as an intermediary

  • Custom-building a data pipeline and internal dashboard to reconcile payments

Unless we’re missing something (entirely possible — LMK if so), there’s an opportunity for a startup that ties one-off purchase revenue to Stripe products and handles reporting which can then be fed into bookkeeping software.

Initial target market would be the thousands of e-commerce companies that are living with this issue and trying to address it with their own janky, custom solutions.

🧠 Framework: Support Driven Development

Support driven development aligns the incentives of the people building products. Y Combinator partner Kevin Hale coined it:

I love this — everyone from the founders to the engineers should do customer support. Nothing will help a founder or engineer learn about and empathize with user issues faster than doing support.

The last thing you want as a founder is for your team to be removed from the pain points your user’s experience — you want the team to feel that pain intensely, so that they’re motivated to find the right and best solutions (rather than just a “good enough” solution).

Transform your business with the No-Code AI Toolkit, effortlessly integrating AI automation into your workflow. Ideal for professionals, this user-friendly solution simplifies tasks and saves valuable time.

Get ahead with instant, free access to revolutionary efficiency.*

📈 Trend: Layoffs

I went viral for the first time in 2019 after a bunch of my former teammates at Uber got laid off by creating an early version of what’s now become common — a post-layoff Google Sheet that recruiters could pick through to find vetted candidates from that company to hire.

The experience made me think deeply about the importance of vetting and high signal within hiring.

I nearly used that momentum to pursue a startup that lets high applicant volume companies creates prompts for candidates to create short video responses to. This would filter out low intent candidates and provide higher signal for recruiters and hiring managers before they invest in time on screening calls.

I didn’t pursue it since the time spent watching the videos to vet candidates would likely be too high to easily prove the value initially and get it off the ground. Ironically, a friend from Uber ended up going forward with the idea and joined Y Combinator’s W20 batch with it.

They pivoted, but with the rise of transformer-based AI models there’s an opportunity to tackle this since we’re still seeing a high level of layoffs in tech and elsewhere even though they’ve decreased from their Q1 peak.

Here’s what I’d do:

  1. Offer companies a bare-bones ATS that lets them create prompts for candidates to record short video responses. Don’t limit the length of the responses.

  2. Run the videos through computer vision and NLP models to understand sentiment (are they passionate about this role? are they evasive?). This removes the need for recruiters to watch each video — a huge bottleneck in the old version.

  3. Surface only the best 1% of applications to recruiters. I bet you could even pitch companies on needing to hire fewer recruiters to manage their recruiting pipeline, or just let hiring managers vet candidates for their own roles entirely without needing a recruiter in the loop (sorry recruiters!).

After an MVP, you could also let companies create target personas for each role that are more in depth than seniority, experience, etc.

You can get hyper specific here — what type of temperament are they looking for? Work closely with early users to learn what vectors matter most for “culture fit” across companies and you’ll be one of the first platforms that can actually vet for that upfront, further increasing signal and decreasing time wasted in interviews that aren’t a fit.

Over time you will be able to improve the system’s recommendations by training it on which videos lead to hires for each type and seniority of role.

💬 Quote: A Powerful Hiring Question

A classic way to understand what type of person you’re hiring is to ask them about their failures.

Most failures happen due to multiple factors. Chances are if you failed at something, part of the reason for it was because of something you did or didn’t do. I’ve failed enough times to know this is true :)

So if the person you’re considering bringing onto your team doesn’t acknowledge that they were a driving force behind their own failures you may be dealing with someone who is at least one of the following:

  • Vindictive, bitter, or dealing with emotional baggage

  • Inexperienced or scared and potentially not confident they’re ready for the role you’re planning to task them with

  • Genuinely delusional, dishonest, or lacking self-awareness

Even if the candidate just beats around the bush or gives a vague high-level answer it’s a red flag. The best candidates will own the mistake, wear it, and tell you what they learned from it. Those are people who are coachable and have a growth mindset.

🔗 Houck’s Picks

  • Free download: 100 leading SaaS landing pages audited. Identify landing page trends, get started with benchmarks, and know what your competitors are doing. Click here to receive it.*

  • How to price your SaaS using the Pareto power law (Link)

  • The three core reasons startups don’t find product-market-fit (Link)

Upgrade to see my full list of picks this week:

  • How the best board members set themselves apart

  • How this founder got a16z to invest in his startup

  • 3 big mistakes founders make post product-market-fit

💡 How I Can Help

Become a member with a 14-day free trial to join the community, get access to all 60+ deep dives, and fireside chats with experts.


Grow your audience + generate leads with my growth service.


Share your round with hundreds of investors in my personal network.


Hire curated candidates from top startups and communities.


I’ll help solve a specific challenge you’re facing with your startup.

🚀 Advertise in my newsletter to get in front of 90,000+ founders.

“*” indicates sponsored content.