Software is cheaper and easier than ever to make, which is shifting the line between buy and build. Your tooling stack is unique to your business, and owning it fully is another way to defend your unique context.
Says Jeff Huber in his “12 factor companies” post. This is my Part 3 response to this post.
This is something I’ve been saying for two years. On October 1, 2024, I sent an email to our investors that said, “SaaS is dead. Happy October.”
Since then, I’ve been championing the idea of companies building their own internal tools. I’m not saying you need to build SaaS apps and sell them to other companies. I’m saying you can build the software your own company needs.
We’ve done this at Gentic. We used to have a $25,000-a-year HubSpot bill. Now that bill is zero because we built our own internal tools. The companies I work with are capable of doing the same. It also reduces their dependence on SaaS vendors that put customer data in a data jail.
But building your tools does not mean building everything.
Jeff’s next point is to buy your infrastructure, and I agree. We should pay for AWS and the servers where our agents and software live. We can write code and use agentic engineering to build internal tools, but infrastructure is not our core competency. I have more than 25 years of software experience, and I still do not feel comfortable building infrastructure. That is something we should pay for.
He also says to hire agents before people. We’ve heard Tobias Lutke from Shopify make a similar argument: before putting in a request to hire, prove that the work cannot be done by AI and agents. I think that is a good rule of thumb.
Another principle is to maximize internal information openness. Mother is the Gentic Computer that lives in Slack. It does not allow direct messaging and can only operate in public channels, so the knowledge the company gains can be shared with everybody.
Finally, design a compounding learning machine. This is why we are spending so much time building a proper company brain. It is not just a collection of markdown files. It is a database.
In our case, Gentic DB is a data lakehouse based on DuckLake, with vector data, structured data, and files. As agents work and save what they produce there, it becomes a compounding learning machine.
Owning the data is directly tied to that compounding learning machine. You cannot build one without also owning your data.
Bora Celik
Founder, Gentic
--
Reply to this email or DM me on LinkedIn to discuss your brand's agentic transformation via gentic.co
