Most investing dashboards answer one question:
"What's happening today?"
I'm more interested in:
"What's been quietly changing for the last six months?"
Institutional conviction doesn't usually appear overnight.
It builds. It fades. It rotates.
This week's lesson:
Investors don't necessarily ignore data because it's unavailable.
They ignore it because it doesn't fit into their workflow.
Building a product isn't just about exposing data.
It's about making it useful.
One unexpected challenge while building software in 2026:
Deciding which AI model to use.
Code quality, context window, pricing, rate limits, and credits all matter.
I haven't found the perfect balance yet.
A year ago the question was:
"Which AI model should I use?"
Today the question is:
GitHub Copilot?
Claude Code?
Codex?
Cursor?
OpenCode?
OpenClaw?
Open-source or closed models?
Choosing an AI coding stack is becoming surprisingly comple
I disagree, Maybe not currency, but they're definitely becoming a scarce resource. Running out of AI credits halfway through a sprint feels surprisingly similar to hitting an API rate limit.
Every investor has that one metric they never skip.
Mine has changed over the past few months.
What's the first thing you look at before investing in a company?
π Revenue Growth π° Profitability π Valuation ποΈ Institutional Holdings π€ Management π Something else?
#Investing
Everyone thinks AI is the hardest part of building a finance product.
It isn't.
The hardest part is making sure the data is correct before AI ever sees it.
Garbage in. Confident garbage out.
#BuildInPublic#DataEngineering
Building the product: 6 months.
Finding a name: somehow harder. π
Every idea ends with: β .com taken β AI startup already using it β Trademark from 2018
Naming a startup is the real final boss.
#buildinpublic#startup#indiehackers