[Not loaded yet]
So finally, after all this, we come to: do LLMs actually ease the impl...
actually, no. I started answering that and then deleted it. LLMs are irrelevant to the point I'm trying to make here, which is this:
"What if instead of buying software companies build their own software," is not in any way a dumb question. Tens of thousands of people ask it for themselves and their colleagues every single day.
And most of the time it's a fucking disaster. So yeah, if someone can make it easier, we'll take it.
Awesome thread 😂I'm currently building simple custom enterprise software and there is one more blocker:
big non-tech companies are aweful at SaaS. They are aweful at building it, at maintining it, at selling it, at buying it, at using it.
They just "can't SaaS", really.
Big Civil Eng. company (where I'm at) 1 year ago: "We need a team of geeks making our software because digital transformation".
Same company today: "We're disbanding every team after all. Maintain… ? We'll buy externally". Pays 5x the price, has to rewrite specs. Specs are aweful. Saas crashes.
Feb 5, 2026 07:59But big boss has a revolutionnary plan: a friend of their cousin has a "turnkey" solution for only a bajillion €/month (they have share in that company? Well why not?).
Except turnkey cost a 2nd bajillion to customize, a 3rd to train people and a 4th to migrate data and everybody hates it anyways.
Yet, your typical engineering firm has dozens and dozens of custom Excel sheets that are excellent SaaS in disguise. Basic CRUD and spreadsheet like software that is super easy to ship by small team (even more with LLMs), with people who can actually use it should you listen to what they want.
BUT : "strategic reorg for the age of AI" (i.e. Microsoft sales have sold a huge copilot package) and you f' things up all over again.
The question is not "buy or make", AI or not. It's: "can you SaaS at all"?