Many startups and large tech companies have attempted to create artificial intelligence to code software. Now, another new player is stepping out of the shadows and into the ring, with a mission to solve the many problems that will arise when humans and all these AIs write code together.

Tessl is building what it describes as a “native AI” platform that developers and their teams can use to create and maintain software, and opened a waiting list Thursday for those who want to try it.

We say “under construction” for good reason: Tessl’s product hasn’t launched yet and is expected to be ready early next year. But the London-based startup is now revealing a little more about what it’s doing with great financial fanfare: Tessl quietly raised $125 million in a seed round and a Series A round, both announced for the first time today. The latest round is led by Index Ventures, with participation from Accel, GV and Boldstart. GV (aka Google Ventures) and Boldstart co-led the funding round.

TechCrunch has confirmed with multiple sources that Tessl’s post-money valuation is north of $500 million, more precisely $750 million.

As you might have guessed, one of the reasons a company with no customers or shipped product gets this kind of attention from top-tier venture capital firms is the person building it.

The CEO and founder of Tessl is Guy Podjarny, something of a developer whisperer. His last startup was Snyk, a cybersecurity company that was last valued (in 2022) to $7.4 billion. Before that, he was CTO of Akamai, a role he took on after Akamai acquired his first startup, Blaze, which focused on speeding up website load times.

“Podjarny is incredibly visionary and thoughtful about his company,” said Carlos Gonzalez-Cadenas, the Index partner who led the investment. “He’s very, very good at (understanding) developer communities and building developer-oriented companies.”

Podjarny said in an interview that the concept for Tessl grew out of his experience at Snyk.

Snyk focuses on detecting (and fixing) security vulnerabilities in code, and Podjarny observed that a similar problem is becoming more pressing when it comes to code and software interoperability in general, especially due to of the rapid expansion of code written automatically by AI.

“What effect does AI bring to software development? » he remembers asking himself. The answer was: speed it up, but also create a lot more automatically. The process of maintaining and sending updates to this code would add complexity and risk of system failure. This ends up having many negative consequences (security, availability, cost, efficiency) for organizations. “The more that image formed in my mind, the more I knew I was going to build this,” he said.

The startup’s name, Tessl, refers to “tessellation,” Podjarny said, because it aims to ensure that software and the code behind it fit together seamlessly, rather than existing in a messy jumble and which overlap.

Podjarny was wary about the types of applications or code he envisioned being built or maintained on Tessl. But it looks like it will start small.

“We don’t yet share the full strategy on what that means,” he said of target applications or use cases. “I would say we don’t start with games. We start with relatively simple software that allows us to create an end-to-end system that is more manageable for LLMs to generate and more manageable for humans to specify. And we will evolve from there.

The basic idea behind the startup goes something like this: Developers and their teams (which include product managers and other people who don’t code) can provide specifications to Tessl in the form of natural language or code. Tessl will then write code corresponding to these specifications.

Teams can test this code in a sandbox, where issues can be reported and resolved, and continue to modify the specifications as necessary. After that, Tessl can be automated to maintain this code to this specification. So if something else might break because of the new code, Tessl will run a fix to identify and fix it.

It seems that Tessl is not designed as a walled garden. Tessl will initially aim to support Java, Javascript and Python, and will add more languages ​​over time, Podjarny added. And Podjarny said he’s talking with others who have built or are building AI coding assistants, with the idea that the work of these other platforms will also be maintainable using by Tessl.

This means that although, in theory, it will compete with solutions like Anysphere Cursor, By the pool, GitHub co-pilot, Magic, Codeium, Increase, OpenAI, IBM and many more, it could potentially also work with whatever a team uses. Note: Many startups on this list are also seeing their valuations climb into the single digits, another reason why Tessl’s valuation doesn’t seem that lopsided here. Could this be yet another bubble or will some of them manage to justify the numbers they achieve on paper?

One of the reasons why investors like the idea and support it is this scalability. Code maintenance is something that has “a lot of signals” as being important right now, Gonzalez-Cadenas said. “But he is building a registration system here,” he added. “Once you do that, there are lots of opportunities available to you.”