Abundance Mindset
You must unlearn what you have learned!
For more than three decades, I’ve built companies, products, and teams under a single, unquestioned assumption:
Everything is constrained.
Engineering time is limited.
Budgets are limited.
People are limited.
Attention is limited.
So I learned to prioritize.
And our corporate processes were based around these constraints.
Endless debates on feature prioritization.
Hard calls on fixing bugs versus shipping on time.
Choosing which leads to call or which competitors to analyze.
This mental model of constraints has shaped everything about how modern businesses operate.
And sometime around December 2025, it quietly became wrong.
Not incrementally wrong.
Categorically wrong.
The Scarcity Era (a.k.a. Everything I Was Taught)
Scarcity thinking was rational for a very long time. It simply reflected the reality of the world. In software, engineers were scarce and expensive. And even the best engineers could only code or design so quickly. Ditto with business analysts, salespeople, marketers, etc. Any work requiring intelligence required humans, and that meant scarcity.
Scarcity wasn’t a mindset.
It was physics.
The Inflection Point: “Why Not Do Everything?”
In late 2025, AI crossed an invisible but profound threshold. Interestingly, there wasn’t really any momentous “oh wow” release like the original ChatGPT release in 2022. Instead, it was a series of incremental updates to a wide variety of AI technologies—Opus 4.5, GPT 5.2, Cursor plan mode, and so on.
Individually, each update was relatively unremarkable—a few extra points on testing benchmarks, a new ‘plan mode’ feature, etc. Collectively, however, we reached a point where AI tools could run for a long time in parallel, tackling many tasks at once.
As one of the top AI engineers, Andrej Karpathy, has said:
“I rapidly went from about 80% manual+autocomplete coding and 20% agents in November to 80% agent coding and 20% edits+touchups in December.
i.e. I really am mostly programming in English now.”
Suddenly:
Listening to and analyzing every customer interaction is trivial.
Analyzing every support ticket is cheap.
Testing every business idea against every major strategy framework (Jobs to Be Done, Blue Ocean, Porter’s Five Forces, etc. costs pennies and can be done in seconds.
Writing, fixing, refactoring, translating, and validating software is now parallelizable. Why not have AI tools start fixing every bug the moment it's reported?
Monitoring the competitive landscape is now cheap. Just watch every product release, every press release, every job posting, every patent application for every one of your competitors.
The old question —
“What should we prioritize?”
Is now replaced with a much stranger one:
“Why not do all of it?”
That question sounds incredibly naïve until you experience it firsthand.
As Yoda famously said in The Empire Strikes Back, “You must unlearn what you have learned”.
Abundance in the Real World
1. Every Customer, Fully Understood
We used to sample customer feedback. Now we can exhaustively understand it.
Every sales call.
Every email.
Every support ticket.
Tools like https://bagel.ai can integrate across a variety of data sources to provide these insights. Gong.io records and analyzes all of your sales calls.
2. “Fix All the Bugs” Is No Longer a Joke
Windows famously shipped with hundreds of thousands of known bugs. Not because Microsoft didn’t care — but because prioritization was mandatory.
Today?
AI doesn’t care which bug is “more important.”
It works on all of them simultaneously. Software bug-tracking tools like Linear can integrate directly with AI programming tools like Codex—simply assign bugs to the AI, and it’ll start working on them! https://linear.app/integrations/codex
3. Lead generation
I can’t tell you how many go-to-market plans I’ve seen (and created myself!) that were purely theoretical: “We’re going to sell to mid-sized businesses in the healthcare industry.”
OK, that’s a nice start, but which companies? What unmet need are we solving? Who specifically in those companies has that need? Who would make the purchase decision?
In the past, we had to resort to these broad categorizations. Tools like targeted advertising allowed us to get a bit more precision, but a more thorough analysis was simply impractical.
No longer—instead of these theoretical discussions, it will soon be possible to have incredibly precise go-to-market plans: here is a list of 100 possible customers that need our product, here are the key individuals at those companies, and so forth.
Branding and brand awareness (particularly for novel products and services) are still incredibly important, of course. Even here, with AI content generation tools, it’s possible to create vastly more content than was possible previously. Why not create content for all of the scenarios your product or service solves? That can help with awareness marketing and also with follow-on support! I still find it frustrating that, for many of the products I use, the documentation and online help are often incredibly bare. Will this product really solve my specific problem?
4. Results meetings, not planning meetings
Can 100% of all bugs be fixed with automated AI tools? Probably not—at least not yet! But that doesn’t matter. Imagine now running a bug triage meeting where, instead of debating the theoretical impacts of fixing a bug (or adding a feature or changing a legal contract or whatever), the discussion is about “hey, our AI has it working and here are the results.”
This same idea can apply to many different kinds of meetings. A few weeks ago, for example, a good friend currently teaching at a university called with an interesting technical question. His class was debating this topic, and he asked if I could hop on a call with them to discuss it.
I said “sure”—but rather than show up to the call and pontificate, I used v0.app to build a working prototype of the solution. I showed up with answers and proof, not speculation. Most remarkably, it only took a few minutes of my time—I had an idea of what the answer might be, so I simply asked v0 to give it a try.
What if every corporate meeting ran that way? There is basically no reason anymore for opinion and theory meetings. Why not show up with results and analysis? Often, that only takes seconds now—at least to get started.
Attention: the one true scarcity
With all of these incredible AI advances, there is still one thing that will forever be scarce: human attention. Each of us is bombarded with constant demands for our attention: emails, texts, kids, colleagues, social media, you name it.
Simply generating 10,000 social media posts for your company doesn’t solve the problem of breaking through all the noise.
Ditto for features. Just because you can add a feature to a product doesn’t mean you should! Microsoft Teams, for example, is the Cheesecake Factory of software. It has a bit of everything! It’s hard to think of a feature Microsoft did not add! The result is a confusing and cumbersome mess.
In a world in which content and software are plentiful and cheap, real value will be created by those who can solve the attention problem.
Make my life simpler, please!
AI technologies are not likely, at least anytime soon, to substitute for the human experience: Empathy. Taste. Artistry. Judgment. Emotion. Companies that tackle these dimensions will excel in a world of abundant intelligence.
Final Thought
Scarcity trained us to think small.
Abundance forces us to think honestly.
The hardest part isn’t learning new tools.
It’s unlearning the reflex to ask:
“What can’t we do?”
And replacing it with:
“Why aren’t we already doing this?”
Yoda was right.
You must unlearn what you have learned.