← All episodes 
Episode
AI Killed Software. Now Every Company Is Going to Build Its Own
For 30 years we just bought software off the shelf. Now an employee with Claude Code can build in an afternoon what used to cost months and a yearly license. Caio and Marina dig into why the biggest competitor to the next SaaS isn't another SaaS — and what that changes for your product team, your operations, and your stack of subscriptions.
In this episode
01 The Hook: The Competitor Nobody Saw Coming
- Caio opens with the provocation: the biggest competitor to the next generation of SaaS isn't another SaaS — it's an employee with Claude Code open on their screen.
- Marina pushes back in disbelief and sets the stage: we spent 30 years buying software off the shelf, signing license after license. Why would that change now?
- Caio teases the thesis without giving it all away: what used to be the blocker was the cost of building. That cost collapsed. And when building gets cheap, buying stops being the obvious move.
- Marina drops the anchor question of the episode: so does every company become a software factory now? Caio reins it in: hold on, there's nuance — and that's exactly what we're going to unpack.
02 Off-the-Shelf Software Became a Commodity
- Caio explains: most SaaS tools solve a generic problem for a thousand companies at once. The price you pay is that it never fits your process 100% — you end up adapting your operation to the software.
- A concrete example Caio lived: instead of buying yet another CRM full of fields nobody uses, he built a lean internal CRM with exactly the fields and automations of the company's real pipeline.
- Marina pushes: but good SaaS has integrations, security, support, years of development. Isn't this just a souped-up spreadsheet?
- The turn: Caio agrees that for critical, complex stuff SaaS still wins, but argues that 70% of what companies pay for is simple-task software — and that chunk is what's becoming a commodity you can build in-house.
03 The Cost of Building Collapsed (and What That Unlocks)
- Caio shows the before and after: building an internal tool used to require a developer, weeks of backlog, a priority queue. Today someone who understands the business describes what they want and AI writes the code.
- Practical example: an automation that grabs leads from a form, qualifies them with AI, routes them to the right salesperson, and pings the chat. That used to be a project — now it's an afternoon with an agent.
- Marina raises the real fear: what happens when that code breaks and the person who built it has already left the company? Doesn't it become a duct-tape time bomb?
- Caio answers without romanticizing: yes, governance becomes the new bottleneck. Building isn't the hard part anymore — it's maintaining, documenting, and deciding what deserves to exist. The risk moved, it didn't disappear.
04 Trading Licenses for Agents: What Changes in Practice
- Caio describes the concrete move: a company looks at its list of subscriptions, finds three tools that cost a fortune and use 10% of their potential, and replaces them with a small internal system that does only what it needs.
- Marina asks for the math: is it really worth it? Caio gives the honest comparison — a license is a recurring cost that only goes up, an internal system is a one-time build cost plus maintenance, and at scale that flips the game.
- Balance point: Caio is clear it's not about replacing everything. It's choosing where you want control and where it doesn't matter. Nobody's going to rebuild their own payment system or email.
- Marina ties it up: so the question stops being 'which tool do I buy' and becomes 'do I buy this or build it.' Caio confirms — that's the new decision every leader is going to have to make.
05 The New Role of the Product Team and Who Survives as SaaS
- Caio repositions the product team: it stops being whoever waits in the dev queue and becomes whoever orchestrates agents to build and test fast. The profile worth its weight in gold is someone who understands the business and knows how to direct AI.
- Marina asks what's left for SaaS then. Caio separates it out: survival goes to whoever has proprietary data, a network between companies, regulatory complexity, or a problem so hard it's not worth building yourself.
- The turn: Caio points out that the SaaS that was just a pretty interface on top of a database is exactly the most threatened — because AI can build the interface and the database on its own.
- Marina brings the tension to the operations debate: and what about headcount? Caio answers carefully — the most valuable professional of 2027 isn't the one who can code, nor the one who only knows the business, it's the bridge between the two.
06 Wrap-Up: How Your Company Prepares Right Now
- Step one, concrete: Caio suggests auditing your subscription list this month and flagging which tools are generic and underused. That's your queue of candidates to build in-house.
- Step two: start small and from real pain — an automation or micro-system that solves a bottleneck the team already complains about every day, not some massive moonshot project.
- Step three: invest in the right people. Caio insists the tool is useless if nobody in the house knows how to translate a business problem into a prompt for AI. That's the role that doesn't even have a name on the org chart yet.
- Marina closes by challenging the listener: look at your screen full of software tabs right now and ask how many of those you'd still be paying for two years from now. Caio wraps with no magic promise — this isn't the future, you can start this week.