Founder Insights and Contrarian Takes
Welcome back to Founder Mode!
The longer I work on building an AI company, the more I realize that most of the best strategies don’t sound impressive at first. They don’t show up well on slides. They don’t fit neatly into the usual startup playbook.
What actually works often feels backward. It’s about removing friction, giving up control where others try to lock it down, and doing the hard, boring work so customers don’t have to.
Lately, a few ideas keep coming up in conversations with founders, operators, and partners. These aren’t theories. These are patterns we’re seeing in the real world as we build.
Here are some surprising lessons. They changed how I see monetization in AI companies. They also change onboarding, GTM, and long-term value.
The Third Bucket Monetization Strategy
Many platforms struggle because they try to charge the wrong people. End users are price-sensitive. Business partners are running on thin margins. Everyone feels squeezed.
In situations like that, the smartest move is often to stop pushing for revenue from those two groups. Instead, give them great technology for free or close to it. Use that to capture the market and build engagement.
The real money often sits in a third bucket. The surrounding ecosystem. Hardware companies. Big brands. Sponsors. These groups will pay a premium for access to a highly engaged audience and clean data. A founder friend had a great two-sided marketplace with high engagement but no willingness to pay on either side. Adding a third side, and soon a fourth, opened up many ways to collect.
Not an issue at Pretty Good AI, but this way of thinking has helped us stop forcing revenue early and focus on building leverage or goodwill. Serve the patients. Support the providers. Monetize the ecosystem around them.
Radical Contract Simplicity Beats Lock In
Most enterprise buyers have been burned by long contracts. Seven-year terms. Hidden fees. Expensive onboarding. Once they’ve been trapped, trust is gone.
We’ve learned that removing the handcuffs is often the biggest disruption. Month-to-month pricing. Free pilots. No setup fees.
The logic is simple. A paying customer who hates you is worse than a customer who leaves.
When people know they can walk away, they relax. They engage honestly. They judge you on results, not fear.
For me, this method has made things smoother. It has shortened sales cycles and eased long-term support issues. Trust scales better than lock-in ever will. In some cases, you open up deals that might be harder to convert - i.e., let in "tire kickers"- but for me, this is an opportunity to learn and to better understand customers of all types.
I met a founder many years back who had built a $100M business with all month-to-month deals. Super old school B2B saas type company, and against all the sales wisdom and "growth advice," but he was the first to know when his product or offering wasn't meeting the mark, as customers voted with their feet. As he told me, it's way better to know now vs waiting until month 11 of a longer-term contract for them to leave anyway, with no feedback on why or when they became unhappy.
Automating the Messy Middle of Onboarding
Onboarding is where momentum goes to die.
If your process needs busy customers, it will stall. They can't fill out long forms, read emails, or answer many questions. They won’t do it. Not because they don’t care, but because they’re overloaded.
The better approach is to do the hard work quietly. Use AI to ingest sales calls, transcripts, notes, and messy inputs. Turn that chaos into a first draft configuration.
Then bring the customer back for the last five percent. Confirm these rules. Approve these settings. Done.
At Pretty Good AI, this change helps customers see value quickly. It also reduces friction when deals often slow down. This way, every minute we spend with a customer is spent learning from them and using anything we gather to improve and shorten the time to pilot. AI lets us hyper-customize the most detailed configuration with much less work. In a sense, building a highly custom solution, but at the lower cost/complexity of a very standardized approach.
From Rearview Data to Windshield Data
Many data products fail because they only tell users what has already happened.
That’s interesting for a day or two. Then engagement drops off fast.
To build something that lasts, the data needs to point forward. Use past performance to project future outcomes. Training plans. Simulations. Forecasts. Recommendations.
This turns your product from a report into a tool. From a one-time check-in to something people rely on every week.
At Instacart, this mindset had us focused on helping customers make better decisions. We would predict out-of-stock items and slower-delivery stores, and, with that up-front information, massively increase customer satisfaction by getting the items they needed when they wanted them. At the time, these were handcrafted and pre-trained models. Today, with LLMs and much lower-cost approaches, we can take feedback much more quickly, in near real-time, and retrain and tune models many times per user, or at least with per-user context.
Why This Matters
These ideas all point to the same thing. The future belongs to products that remove work rather than create it.
Free tools where others charge. Simple contracts where others lock down. Quiet automation where others demand homework. Forward-looking data where others stop at reporting.
None of this is flashy. All of it compounds.
5 Key Takeaways
- Monetize the ecosystem. The third bucket often holds the real value.
- Remove the handcuffs. Trust grows faster without long-term contracts.
- Hide the complexity. Do the hard onboarding work for the customer.
- Point data forward. Predictive tools beat historical dashboards.
- Simpler scales better. Fewer demands lead to deeper adoption.
Final Thoughts
Building Pretty Good AI keeps reinforcing one lesson for me. The best strategies don’t try to extract more from customers. They try to help more while asking less.
When you lower friction, remove fear, and quietly handle the messy parts, customers don’t just stay. They bring others with them.
That’s how durable businesses get built. Not by locking people in, but by making it easy to say yes and even easier to keep going.
See you next week,
-kevin
Recent Social Posts
Recent Podcasts
2810 N Church St #87205, Wilmington, DE 19802
Unsubscribe · Preferences
