scharf blog

on tech and other things


Metasoftware: software that writes itself

The past 10 years of software have been dominated by the unbundling of various tools into vertical SaaS. The next ten years will be a rebundling via what I call metasoftware. Put simply, metasoftware is software that writes itself. I don't mean co-pilots that help you write code, I mean tools that write, run, bundle and deploy software that end users can interact with.

We’ve already seen low and no code tools like Retool, Airtable and simple spreadsheets serve as an alternative for myriad SaaS applications. Eventually, however, the cost and complexity of maintaining the low code tool usually outweigh the benefit. Imagine, though, that the cost was near zero. Want to add role-based permissions? Just tell the software how you want it to behave and it will update itself to meet your needs.

Let me share a concrete example of metasoftware. We built an AI data scientist called Seeker, similar to Python Data Analyst from ChatGPT with additional features like a Snowflake connection and custom agents specialized in data science. Off the shelf tools like Amplitude might cover 80% of use cases, but for complex joins, data enrichment, or even simple logic, bespoke SQL or Python code is required. Using Seeker, even non-technical users are able to answer complex questions about their data in plain english.

0:00
/1:29

Seeker is trivial compared to what’s to come. Products like Vercel v0 enable people to build generative UIs but stateful backends and databases are still in their nascent phases. In the next five years end users will come to expect bespoke, long-living customization of their favorite software. This is going to have deep-reaching consequences for all types of SaaS companies.

  1. A threat to vertical SaaS

The first implication is a threat to vertical SaaS. Of course, we will still use software, namely we will use metasoftware. But in-house bespoke solutions that plug in to a system of record will replace the myriad vertical SaaS tools that companies purchase today. Every piece of software will need to fight for its life as companies compare to metasoftware solutions. No company is going to pay thousands of dollars a seat for software that they could build themself internally for a fraction of the cost. This means that the vertical SaaS companies that do survive will need to compete much more fiercely on price and will need to leverage other moats like network effects and scale economies to stay relevant. If your margin is my opportunity, 80% SaaS margins are leaving a ton of opportunity on the table.

  1. The growing importance of the system of record

Second, the products and tools with end user data will become even more valuable. These are systems of record that hold information about users that they can leverage into building features. Of course, system of record applications have always been important, but their importance will only continue to grow as it becomes trivial to create bespoke applications that leverage the valuable user data that lives in these tools.

  1. Huge opportunity in vertical application companies

While vertical SaaS is at risk, there’s never been a better time to leverage software to solve end problems. Metasoftware will allow many more companies., even non-technical ones, to reap the benefits of custom software. The best workflow automation tools will come directly from the company doing the workflow today, or from the fast acting tech companies that move closer to the end user.

I could be wrong on the implications but I’m highly confident in the possibility of metasoftware and excited by the opportunity, both as a user and a builder. If you’re building cool metasoftware, I’d love to give it a try.