Orthodox Jewish neighborhoods can get very expensive.
Why?
No driving allowed on Shabbat (Saturdays).
So walking distance and proximity to a synagogue becomes very important and turns into a mini market w/ limited supply
where the footprint is hard to expand.
btw if you're making a lot of money from the OpenAI, Anthropic, or SpaceX
ut you consider yourself a largely uninteresting person (e.g. you're thinking about using the money to buy a bigger place/second place and angel invest)
consider giving the money to me - I will do way more fun an interesting things with it
thank you
The first fundamental truth is that FDE is not a role, it is a product strategy.
It has to originate from the fount of the product org.
You judge an FDE org by the fruits they bear! What new product was discovered, what Thielan secrets were mined and refined into weapons grade product?
(but hey, what do I know about FDE 🤷♂️)
Mets won a series against the best team in ball where:
- Sean Manaea threw a QS
- AJ Ewing had a 3-hit game
- Freddy got off the schnide
- Bichette had 3 multi-hit games
- Bullpen pitched 12 IP w/ 4 ER
and Lindor comes back in like a week? I unfortunately am buying back in.
In all seriousness… I grew up a huge fan of the New Jersey Nets. When the Nets decided to abandon us, they spent the next 7 years insulting the State and all of us fans. So many of us here in NJ who were Nets fans have migrated over to the Knicks for one reason or another. There are basically no Nets fans left in NJ and the State is firmly behind the Knicks. This was an incredible run and cements that.
"This requires a new architectural approach where every business is able to build agentic systems that improve over time, while still retaining control over their IP."
Every AI company will read this and convince themselves this is their approach while still selling expensive SaaS
We should be happy that we can still tell robot text ("slop") apart from human text and that labs don't care about optimizing this away; not sure we would prefer a world where they are indistinguishable
The key to FDE is to not build from scratch, and when we do, we bring it back so we can all dream bigger.
A recipe we follow @percepta
[ ] Find a customer problem
[ ] Design a high level solution
[ ] Find gaps in the solution design that might be solved by an industry tool. If there is one, use that tool.
[ ] Find gaps in the solution that might be solved with a tool in our platform (or with minor investments in that tool). If there is one, make those investments and use that tool
[ ] Build the solution. On the way, build harnesses and tools to make building the solution easier and faster.
[ ] Once you’ve built the solution, reflect on what you built and what could have made it easier. Was there a gap you wish had a solution? Was there a key insight you had? Did you build a tool to make it easier? Let’s go fold that into the platform
[ ] Spread the gospel of your new tool / pattern / approach
[ ] Find out about tools other teams are building and use them to Dream Bigger about how you can Win with the Customer
Thats the whole FDE loop
An ex-consultant @percepta asked me this week why my engineers hadn’t written code in weeks. Do we really need engineers on site? Can’t the PMs do the shadowing and then tell the engineers what to build?
No.
It may have always been true, but especially in the age of AI, engineers need to see the thing with their own eyes to know what to build.
The problem on paper was simple: nurse scheduling was still being done manually. Hours of work every week, totally automatable. It seems simple enough, just a simple matching algorithm the same one old systems have been solving for decades. The executives told us there’s nothing here to improve. The nurse managers doing the work told us otherwise.
And so, in went the engineers, eyes and ears wide open.
The obvious constraints surfaced fast. Everyone works 3 shifts. No back-to-back. No more than 1 extra shift. We scoped and shipped the very basic version in a month.
Then the real needs came out. Every nurse had personal limits that shifted week to week, sometimes requiring a full reshuffle on short notice. Donna works Monday and Thursday and every 3rd Sunday except in July. Matthew likes to work two days in a row, but Jan isn’t allowed to. Per-diem nurses get scheduled the next week. Float pool gets looped in early but only can schedule on the next week and then has to re-reschedule the following week as adjustments came in. MedSurg A1 is different from MedSurge B3 and PCU D4 is different from 4H.
None of this is written down anywhere. It’s stored in the manager’s head, along with 1000s other things, like the needs of the actual patients in their care. And we only learned these things because we had spent three days, watching, talking, texting, eating breakfast lunch and dinner with nurses and managers.
Two weeks later, we shipped a working MVP. The engineers understood the problem intimately. They talk directly with the users, understand the problems, explain tradeoffs they’re going to need to make and come to the right answer.
And today, the nurse managers spend less than 15 minutes on scheduling. That’s more time to coach nurses, more time to care for parents, and really more time to make sure they don’t burn out.
Engineers and Operators working together. Thats AI Transformation.
I reminded my colleague that the gap between what AI could automate and what actually gets automated is only solved by seeing it with our own eyes. Builders who learn to close that gap are going to be the most valuable people in this industry.