Why We’re Betting on “Pretty Good” Instead of Perfect
Welcome back to Founder Mode << NEW SITE
I’ve been building companies in the Valley for a long time. Long enough to see hype cycles come and go. Right now, the loudest one is Artificial General Intelligence. Everyone is racing to replace humans entirely.
But when you enter the real world, especially in healthcare, that vision crumbles quickly.
At Pretty Good AI, we don’t spend our time chasing perfect systems. We answer phones, fix workflows, and help businesses navigate their chaos. What we’ve learned is simple: the winning approach is not full automation. It’s “pretty good” AI backed by humans who know when to step in.
This is the thinking that’s shaping how we build, who we work with, and why we’re comfortable going against the crowd.
The Trap of Perfect AI
There’s a belief that the end goal of AI is total replacement. No people. No handoffs. No exceptions.
That sounds great in a demo. It does not work in real operations.
In healthcare and other high-stakes services, emotions run high. Situations can be complex and unpredictable. Patients panic. Staff make judgment calls. Systems fail in ways no model can predict.
At Pretty Good AI, we see this every day. The AI can handle the bulk of routine calls quickly and accurately. But when something unusual happens, the ability to warm-transfer to a human is not a failure. It’s the product.
The tech accelerates the workflow. The human earns the trust. Together, they close the loop.
Pure automation breaks trust. Human-backed automation builds it.
Capital Is Not Validation
One of the biggest misconceptions in startups is that raising money means you’re right. It doesn’t. It just means someone else wants you to move on their timeline.
We’ve had investors chase us because they feel pressure to deploy capital into AI. They see our background and want to write a check. We’ve said no. Not because we can’t raise, but because we don’t want a boss.
At Pretty Good AI, we’re self-funding on purpose. That gives us control. Control over pace. Control over customers. Control over quality.
I've seen competitors raise huge funds but then struggle with expectations. Big teams. Slow decisions. Flashy demos that don’t survive contact with reality.
Ownership gives you the power to say no. And saying no builds credibility faster than any pitch deck.
The Reality on the Ground
Healthcare moves slowly. We were told it can take 17 years for a new process to become standard. Tech moves in weeks. That gap is where everything breaks.
We work with providers who still use paper forms, old phone systems, and incomplete workflows. Then we’re asked to deploy advanced voice AI into that environment.
This is where most startups fail. They assume the customer is ready. They never are.
At Pretty Good AI, we’ve received encrypted files for which no one has the keys. We’ve been sent call recordings we couldn’t listen to. We’ve dealt with systems no one remembers setting up.
AI doesn’t fail because it isn’t smart enough. It fails because the environment around it is broken. Fixing that environment is the real work.
Tech Is Easy. Implementation Is the Product
Anyone can spin up an AI demo in an afternoon. That’s no longer impressive.
What’s hard is deploying that AI into a live business with real consequences. Cleaning data. Mapping workflows. Sitting with operators helps me see what happens on Tuesday afternoons when the phones light up.
Many companies say they are tech platforms. However, they often just offer services that depend a lot on people, with only a little software on top. Hundreds of agents are doing manual work, with AI bolted on later.
Our bet at Pretty Good AI is the opposite. Start with code. Use humans only where the code fails. Build the system so it can scale without hiding labor behind the curtain.
That approach is slower at first, but it compounds.
Stop Negotiating and Start Building
Strategic partnerships sound great until you’re stuck in legal for six months.
We’ve learned the best way to test a partnership is execution, not paperwork. Skip the joint venture talk. Pick one ugly, messy customer. Try to fix them together in thirty days.
Shared pressure reveals everything. How decisions get made. How fast teams move. How problems get handled.
At Pretty Good AI, this mindset has saved us months of wasted effort. One shared win or one shared failure tells you more than any contract ever will.
Date before you marry.
The Bar Is Shockingly Low
This might be the most surprising lesson. The existing technology in healthcare is so bad that showing up with a modern stack feels like magic.
We don’t need perfection. We just need to be better than the systems built in the 1990s.
We've seen integrations that usually take months get approved in just half an hour. We know the right questions to ask. Not because we’re smarter, but because we’re focused.
In the real world, “pretty good” is revolutionary.
5 Key Takeaways
- Human in the loop builds trust. AI should accelerate work, not remove accountability.
- Control beats capital. Ownership gives you the freedom to build for the long term.
- Implementation is the moat. Fixing messy systems creates real value.
- Execution beats contracts. One shared launch is worth months of negotiation.
- Pretty good wins. You don’t need perfection to change an industry.
Final Thoughts
We aren’t trying to build a spaceship. We’re trying to answer the phone.
At Pretty Good AI, every hard lesson keeps reinforcing the same idea. The most valuable work is often the least glamorous. That’s where trust is built. That’s where growth holds together instead of breaking apart.
Let others chase perfection and hype. We’ll be over here shipping code that actually works, backed by humans who know when to step in.
That’s the bet. And so far, it’s paying off.
See you on Friday,
-kevin
Recent Social Posts
Recent Podcasts
2810 N Church St #87205, Wilmington, DE 19802
Unsubscribe · Preferences
