Building better context end-to-end. Investor and tech exec. Reformed pool guy.
New York, USA
Joined March 2014
- Tweets3.5K
- Following1.9K
- Followers1.9K
- Likes6.1K
Pinned Tweet
FACT: Businesses thrive on process.
Also FACT: Leaders don't invest enough in process documentation.
Want to know how a 150-person pool company operates?
Reply with "Workflow" and I'll DM you the process map for an 8-digit business.
Kirk Newcombe retweeted
human-in-the-loop
└ human-on-the-loop
└ human-near-the-loop
└ human-cc'd-on-the-loop
└ human-legally-liable-for-the-loop
He’s right. Gonna be a while. Enablement is the gap.
STOP TURNING YOUR TEAM INTO SHITTY AI DEVELOPERs.
- be a law firm
- want team to use AI more
- pay for Claude for everyone
- start getting random slop, need to teach them better
- start pulling them out of billable to teach
- need to have better client data security, workout of right folders with a ZDR model
- more time off billables, lawyers can barely manage their output in outlook now you have them a nuke
- start passing out API keys like chiclets , hope it doesn’t bite you in the ass?
- new tool comes out, big announcement we are switching to Grokbot
- more training, bad outputs, leaky everything
- have to teach team to manage keys, auth, DBs (remember these are LAWYERS IN THEIR 50s)
- send team your new brittle vibe coded tool. You are really excited , they are at the promise of it but it’s buggy and now you are an IT guy.
- more tools more spend
- less time billing growing and obsessing about your customer (only thing that matters)
- rinse and repeat
Ad nauseum
IT TOOK THIRTY YEARS FOR INCREASED PRODUCTIVITY WHEN ELECTRICITY WAS INVENTED.
This shit will not happen without you being very smart.
In the last 90 days we’ve FINALLY used more AI THE RIGHT WAY than the last three years combined.
… and it’s not one magical agent, one magical hire, one magical agency promising the world, or one magic local data lake.
GET YOUR LAWYERS BACK TO LAWYERING
Stay tuned
The first thing I think of when I read “I know 30 people doing AI implementation“ is that every single one of those consultants is experimenting with your company.
The playbook hasn’t been written.
Kirk Newcombe retweeted
Codebases are going to start looking like optimized topology 3d prints
When AI is better at math than humans it also becomes better at algorithm & codebase design
At that point the system and delivering quality software doesn’t require human understanding to exist in the actual code layer.
Only needs to be understood by humans at the high level of passing testing, verification, vibe checks and being understandable when queried via AI in the rare case one need to dive deeper.
The piracy case may be valid but isn’t SaaS just an amalgamation of patterns? Are patterns IP?
Free markets nurture real competition. Saas has been on a generational run raising prices via dark patterns. Consumers are entitled to competition too lest we forget.
AI Cloning should be illegal. And I think it might be. It feels self-evident, but we may need to wait for the courts to be sure.
This morning, I took a full Windows application from Microsoft, had AI reverse-engineer it using NSA's Ghidra into a set of atomicizable actions, then translated those actions backward into Rust logic, then compiled the Rust into a brand new application.
And it did it. In one shot. So the process is:
C Code -> Compiler -> Linker -> Exe
then
Exe -> OPS -> Rust -> Compiler -> Linker -> Exe
So you feed it an application, and it produces a Rust version of that same application with 100% of the same functionality and appearance and "none" of the original code - just the logic behind that code.
I don't consider this coding.
It's not even vibe coding.
It is piracy.
This might be a hot take, but cloning an application when the application itself is used as an input does NOT, in my mind, yield an independent work.
I'm even dubious about letting the AI base its work on the application. I specifically did NOT do so for TMOG (tmog.org) because I didn't want it just copying. So I wrote a huge spec instead (which I guess could just be copying via indirection, but I'll leave that aside for now).
This time, I chose an in-the-box Windows application as a proof of concept, but there's no reason this couldn't be applied to Photoshop.
Or heck, have the AI port Final Cut from the Mac to Windows. That's a bigger ask, of course, but we're not far from it.
And so I propose we ruthlessly distinguish any process that uses an application binary itself.
Prompt -> Code -> Binary == OK
Binary -> Prompt -> Code => Binary == Clone
My trauma from college engineering days.
So true, I still remember at uni a Math professor gave us as an exercise an unsolved problem that can be stated in a language that a first year student would understand and that sounded simple. He did not tell us that the problem was unsolved, he told us we had two weeks to bring him the solution and that this was required to pass the exam. After two weeks the whole class was depressed, the prof came to class and said "well did you enjoy your past two weeks?", radio silence, he proceeded to explain that if we didn't we would not become good at Math
Increasingly, I think the distribution will trump elegance and more so when you’re solving a real problem that your customers just don’t understand. The word “taxonomy” simply isn’t in the zeitgeist yet.
Graphlit went offline yesterday.
Five years. 50+ integrations, a temporal knowledge graph, durable agents, paying customers in production since 2023.
The product worked. The business didn't. Both are true.
Agents are still only as good as their context. That problem isn't going anywhere.
The tech and IP are still available as an asset purchase, for now.
Thank you to everyone who built on it.
Reassuring interaction with ChatGPT this morning, where it reminded me unprompted that “both can be true“ in regards to competing perspectives on a contentious topic.
This is one way to do it. Or put better guardrails in your repos that defend the design patterns you’ll endorse.
Can't say I have a lot of love for 5.6 Terra
Hopefully, 6 Sol fixes the basics that it was missing too
One of the hardest things for business owners is placing a value on preparation in seasonal businesses.
How do you put a price on three weeks of training for pool technicians before the season starts? How can you possibly measure the impact during the height of the season?
Kirk Newcombe retweeted
Can’t make this up…the MacBook is using its webcam to look at its screen in a mirror to improve AMD Radeon chip support in Omarchy.
Kirk Newcombe retweeted
>math professor spends a year on one of the hardest unsolved problems in math
>"Navier-Stokes"
>his drafts for the whole project went through codex sessions
>openai had his logs in codex
>suddenly rumor leaks AI solved it
>he emails openai to ask what's going on
>openai calls within days
>"funny story, our model also solved it"
>using his exact approach
>asks openai: did you train on my sessions?
>no answer
