Hi, I'm Lawrence.
I'm a Payment Systems Engineer from Zimbabwe.
I'm documenting my journey learning AI Automation while continuing to build expertise in payments.
Here you'll find:
- Payment systems
- AI automation experiments
- Engineering lessons
- What I'm building in public
Follow along.
@HLPClips People hear this and think it means grind harder.
Dorian trained 4 days a week. Sessions under an hour. One all-out set.
"Use it like fire" was never about more hours. It was about how hot you burn in the time you're there.
@hubermanlab People hear this and think it means grind harder.
Dorian trained 4 days a week. Sessions under an hour. One all-out set.
"Use it like fire" was never about more hours. It was about how hot you burn in the time you're there.
Everyone's going to focus on the 85%.
The real line is: "not yet usable for professional biology research."
Anthropic just publicly admitted their best bio model is still walled off from the people who'd get the most out of it.
"Trusted access pathways" is the thing to actually watch.
We’re updating Claude Fable 5’s biology safeguards to reduce false positives. In our testing, this update reduced biology-related fallbacks by about 85% across our product surfaces.
Fable can now assist on a wider range of everyday health and educational questions.
We believe the biggest positive impacts of AI will be in biology and medicine, and we’re committed to putting frontier intelligence safely into the hands of as many researchers as possible. Fable will continue to fallback to Opus 5 for requests we consider dual-use—including virology, toxicology, and molecular design—so it isn't yet usable for professional biology research and drug development.
We're committed to closing that gap through trusted access pathways for frontier biology capabilities.
https://t.co/zbjxPhc909
Here's the part nobody mentions:
Japan has the most AI-friendly copyright law on earth. Article 30-4, train on almost anything, legally, since 2018.
The exact fight US labs are still burning millions on in court, Japan settled before ChatGPT existed.
And they still didn't ship a frontier model.
So it was never a data problem. It was an org chart problem.
They had a cheat code but they never used it!
@HLPClips@JamesClear@hubermanlab We don't have habits. We are them.
Your body = what you ate for 5 years
Your bank account = what you bought for 5 years
Your skill = what you practiced for 5 years
Your thinking = what you consumed for 5 years
None of it happened today.
All of it is showing up today.
@AIatMeta You guys always introduce new products that don't match what your competitors are offering in this AI space.
Plus when it comes to privacy 0/10
I'll stick to Brother Claude.
Pure per-seat SaaS pricing dropped from 21% to 15% in twelve months.
Everyone read that as "outcome pricing won."
It didn't.
Hybrid won. Base fee + metered AI credits. Now 41% of SaaS, up from 27%.
Intercom's Fin is the proof. $100M+ ARR at $0.99 per resolved conversation.
And they still charge per seat underneath it.
Nobody killed the subscription. They stapled a meter to it.
If your product has AI in it, are you metering, or still selling chairs?
@tcmundondo The four questions nobody wants to raise on day one:
– who owns what percentage
– who decides when you two disagree
– what happens if someone stops working
– how someone exits, and at what price
Every one of these gets raised eventually. You just don't get to pick the timing.
Here's what I think @ConnorTalksAI
The wiring isn't the hard part. The confirmation step is.
Live is latency-bound, speak/pause/interrupt decisions multiple times a second. Delegation is the opposite shape: long-running, stateful, connector-authed, approval gate before anything writes.
Two surfaces, one product name.
And approving by voice when you can't see what it's about to touch is how you ship something you didn't intend.
Voice captures intent. Screen approves. Notification when it lands.