I've learned a lot in the last couple of weeks about Customer Discover to identify new niches that we can sell to. Because who doesn't like a good beachhead?? 🏖️ Here's some lessons learned as I go through this process: > Don't open the call with a demo. It primes the brain

Memory with LLMs is what makes things sticky. And stickiness reduces churn. I have Claude, GPT, Gemini, Grok … but when all said and done always revert to GPT for research and writing and CC for code, simply because I don’t what (or can’t) transfer memory. Take away, make

If you are at risk of burnout Port to your 80 year old self Was it really that hard? If yes, good. You didn’t quit, you got good sleep, ate well, exercised. You spent money on experiences, spent time with your loved ones. You were always learning esp when you failed. You

Can Cursor be considered a startup? - Parent company Anysphere barely 3 yrs old - 1 billion ARR (as of Nov '25) 🤯

Days 65-73 on X: 🏖️ Took a week off and mostly disconnected. Batteries recharged. ✅ Today started mobile app version of our solution. React Native + Supabase. Let's close out the year with a bang!

Promoting merch from the CLI with /stickers is kind of genius.

Feels like all these LLM models are like different Linux distributions. All work well. Just depends on how you have them configured to get the most out of them.

For B2B DM campaigns > email campaigns. > Email reply rates 2% (on a good day, even less “interested” replies) > DM conversion rates 10% or more > Compiling a list with good emails is expensive (or can be). > DMs in many cases don’t require verified emails, thus cheaper to

Days 74/75 of x: ✅ 3-6 organic registrations a day. Mostly from LI and X DM campaigns. ✅ Moving along with mobile app with React Native + Expo Go + Supabase Feel so much more productive after my break. Sometimes it's good to step away for a bit.

Days 74-80 on x: ✅ 0 organic registrations 3 months ago. 12 on average now. SEO / AEO starting to show results. ✅ More outreach on LI and X (mostly DMs) to get feedback. It's been really interesting and fun to connect to so many people from around the world. Perspectives and

New Claude Code with Opus 4.5 as default. > Faster, more thorough. > Especially plan mode. Tip: create epics and sub-issues in GitHub to keep track of progress. > PRs are more granular and higher quality now that the issues have better structure.

Days 81 - 83 on : ⚡️ Built a mobile version of the all in two weeks (with CC and now use Opus 4.5 mostly). Insane. 🍀 New version of backend in TyoeScript (ported from Python) is now private beta. These new coding models are so good it’s ridiculous. IMHO the biggest benefits

Even on a Holiday weekend you should post something on X. The algo gives bonus points for this!

Can’t say enough good things about Opus 4.5. This flow has been working great: > Plan epic issues and add them to GitHub with CC”s integrated access to the GitHub CLI. > Review epics and adjust. > Create sub issues for each epic. > If making an MVP write tests last, verify

I wonder if in 6/12 months companies will realize how difficult and time consuming it is to maintain all this code? Reminds me of the "low code" push from the 2010's. This is different of course but the stuff still has to be hosted, secured, backed up, monitored ... Doubt that

Goals for December: - 1.5k followers on X - 24 demos secheduled from LinkedIn - 24 discovery (feedback) calls Tough since it's really a 3 week month, but hey why not?

Days 84-87 on x: ✅ Migrated e2e tests for mobile app from Maestro to Detox (use React Native). Maestro proved to have some gaps. ✅ Added more file MIME types for attachments. Building an app is like building a house. The foundation, frame, architecture etc happens so fast.

Enter Lux. Scored 83.6 percent success rate on Online-Mind2Web vs 61.3 and 56.3 percent for OpenAI and Claude, respectively. Question is can they scale? Is it too late? Or early days enough for more specialized models? agiopen.org/blog

Ready for Apple TestFlight soon! Developed with React Native + Supabase backend. First mobile app (hopefully of many?).