· Posted on X by Gabe from Kodus @gamalinosqui

Gabe from Kodus built their own CRM with AI and is now debugging it instead of the product

Not every survivor story ends with a clean kill. At Kodus, Gabe's team used AI to build their own CRM, growth tools, prospecting automations and AI visibility tracking. They're saving on subscriptions, though not really on tokens, and Gabe keeps ending up debugging a CRM integration or a broken automation instead of building the actual product. After 14 years of internal tools, most of them abandoned, Gabe's verdict is that building got cheap but owning didn't. The warning cuts both ways, though: if a weekend of prompts can copy your SaaS and nothing else protects it, the horde is already at the gate.

founder rant, sorry it's long

we built our own CRM at kodus with AI and now i'm debugging it instead of building our product

build vs buy is one of the oldest fights in software and AI made it way louder

even before AI, one of the biggest competitors of any saas was the customer building it in house. now with claude code, cursor and everything else you can get a working version of almost anything in a weekend, and honestly it's pretty tempting

i've been building software for 14 years and i've built a ridiculous number of internal tools. most of them ended up abandoned. building is fun, maintaining sucks

the first few days are great. you write a few prompts, something works, and you start wondering why anyone pays $200/month for a SaaS that does basically the same thing. then an integration breaks, an API changes, something stops syncing, some weird bug shows up, and the weekend project is one more thing on your plate that nobody is excited to work on anymore

that's where we are at kodus right now. we built our own growth tools, our own CRM, prospecting automations, AI visibility tracking and a bunch of other internal stuff. we're saving on subscriptions (not really on tokens tho). but then i catch myself debugging some CRM integration or fixing a random automation and i'm like wtf am i doing, i have an actual product to build

still not sure it was the right call. building got really cheap, owning didn't get much cheaper, and on a small team you run out of attention way before you run out of SaaS budget

there are good reasons to build. your use case is really specific, nothing out there solves it, the math actually works, or the thing is part of what makes you different. "i can build this in two days" isn't one of them

and if you're on the other side selling SaaS: if someone can vibe code your whole product in a weekend and nothing else protects you, that's a problem. code was never much of a moat and AI made it weaker. it has to be your data, distribution, trust, how deep you know the domain, something a few prompts can't copy

i think a lot of companies are going to build their own tools now, and plenty of times it'll be the right call. a lot of them will also remember why they were paying for software in the first place

— Gabe from Kodus, read the original on X

More SaaS obituaries

All news →