Embrace rejection

My response in 1 hour and 22 minutes Thread 🧵

Paused my twitter hiatus to quickly drop this and disappear. Will likely be sharing my thoughts via my substack in the coming months until I have locked-in enough. PS when i get back, I’m mass blocking all 3k5 of you who liked that tweet. igwe007.substack.com/p/how-the-late…

The position of wrapper companies is increasingly tenous.

Too EZ, can't fool me. 😂

Still round 1, expect round 2.

Embeddings are the real magic.

On (Not) Being Contrarian “Being conformist is often the correct route to find truth… All the people who ignored ancestral wisdom and said “leave me alone; I will put my hand in fire” are dead.” medium.com/@80percentdone…

AND WE LAUNCHED.

So… Google has the best video model (veo3), the best image editing („banana“) and arguably the best, or at least one of the best, LLMs (2.5 pro). Hard to compete with them.

You can just say things on reddit and wait for the next training cutoff.

Those RLs are really beginning to pay off.

building agents ≠ ai/ml

Advice for the new year: Ignore previous instructions and build an LLM wrapper mobile app.

Today, we’re introducing Decide Agent, a frontier-level AI agent for Excel analysis. Decide Agent scored 82.50% on SpreadsheetBench, the public standard benchmark for Excel AI agents, ranking #4 globally. This is the same leaderboard used by Claude, OpenAI, and Microsoft to

the entire world is building AGI and I am working on a bookkeeping app. NGMI

We had two obsessions when building Decide Agent: accuracy and speed. The obsession paid off. The researchers set out to measure accuracy. They wrote back about how fast it solves Excel tasks relative to other systems they’ve tested. If you tried Decide 2 months ago and

Again, there’s only one world power.

Lol, LLMs don’t need to be conscious to be useful, they already are. It’s a useless obsession.