Henrik Torstensson's Weblog

My good friend James Pember has been writing an interesting Substack on AI, software development and business for about a year now. His latest post A roadmap for building the future of SaaS is great read. The article touches some of my recent thinking, and I like his ending to the article:

“…the bar for what we will consider “premium” software is going to be drastically raised, but the opportunity on the other side may be bigger than any other time in software history.”

This is critical for any company looking to raise capital. A startup has always needed a 10x product to succeed, but today’s x is yesterday’s 5x. This means today’s premium software needs to be at least 50x in 2021 terms.

Gustav Söderström interviewed by David Senra. A very good listen covering organizing and running technology companies, building products and AI.

The lesson that there’s not one best way to organize a company, and every way of organizing has advantages and likely disadvantages is something that should be repeated often. But as important is the part of the lesson that says founders and CEOs should organize their companies in a way that fit them personally.

I’ve been interviewed about Spotify and the Stockholm ecosystem by Chris Stokel-Walker in his Fact Company article How the Spotify mafia took over Sweden’s tech scene: “You got a really good mix of very ambitious, very good, mostly Swedish engineers and product people with a commercial acceleration which would have taken much longer” [without international commercial talent].

There is a growing narrative in venture right now: software is being commoditized, AI changes everything, and the real value is shifting from bits to atoms. I have written the article The Deep Tech Illusion with my thoughts on the subject on my Substack.

Alliance VC has lead the pre-seed round in Endform. The company’s official announcement, Alliance VC’s Linkedin, Tech.eu and Breakit (Swedish).

Quoting our Linkedin post:

AI is changing how software gets built. Engineers can now run multiple coding agents in parallel, generating more code per engineer. But the code still needs to be tested.

That makes testing one of the clearest bottlenecks in modern software development. More code means more tests, and manual QA does not scale. Endform addresses this challenge. It runs Playwright end-to-end tests massively in parallel, giving every test its own machine. Lovable, for example, runs over 400,000 Playwright tests per week on Endform and cut their suite runtime by more than 80%.

AI is currently breaking the historical relationship between engineering headcount and software output and software is becoming agentic. That has fundamental impact on both startup and investing strategy. I’ve written a longer article on my Substack, including how one can think about the strategic positioning of different types of software.

Lovable has announced that CapitalG and Menlo Ventures, with participation from several others, are leading the company’s $330 million Series B round.

A very impressive fundraising to close of a very impressive year.

Lovable is raising capital at $6.6 billion valuation, according to CNBC. Not yet announced, but are in the same neighborhood as rumors earlier in the fall.

Jack Altman interviews Sequoia’s new co-stewards Pat Grady and Alfred Lin. A very good episode on Sequoia (including decision making), venture investing and company building.

Also on Spotify.

Forbes ha published its annual ranking of European venture capitalists. The only Nordic-based investor in the top 25 is Johan Brenner at Creandum.