Legible EventsThe business case for software factoriesDec 1
← Writing

The Software Factory

Engineering has always been the bottleneck. Agents are turning that scarcity into abundance.

Lawrence Coburn
3 MIN

So I’m chatting with Ada in Slack, describing a feature for Ambient that I’ve wanted for a while, but have never been able to justify pushing to the top of the roadmap. The feature is the ability to import transcripts and meeting recordings from other systems (like my iPhone!) so that they can all be centralized. A good idea (says me), but not immediately needle-moving. The sort of feature request that might historically die on the vine.

Ada asks a few clarifying questions, chops the spec into a few tickets, and the next morning the feature is fully tested and in dev.

What’s interesting about this story is that Ada is an AI teammate, running locally on my cofounder’s machine, with a deep understanding of our code base and access to most of our systems of record.

This is an example of what might be called a “software factory.” If you’re not familiar with the expression “software factory,” you should probably take a couple minutes and start to get your head around it.

The term software factory dates back to 1968, and refers to a way of developing software that is less craft, and more production line. Ideas, requirements, and bug reports go in, and tested, production-ready software comes out.

With coding agents (and every other kind of agent) surging, software factories may now be in reach for any company with the skills to design them.

Why is this exciting? For anybody who has ever worked at a software company, the biggest bottleneck has always been engineering. There are never enough engineers to build all the features, fix all the bugs, and improve the platform.

It’s the classic tale of scarce resources. Product is balancing all sorts of ideas about where the product could go, but with limited throughput, they have to make very hard choices.

Sales reps are trying to get deals over the line, and sometimes a deal hinges on a feature the platform doesn’t have. The big renewal is at risk because you just haven’t been able to get around to fixing that annoying bug.

Some would argue the focus that comes out of this scarcity is a net positive, but there are many sad stories of sales and business opportunities that slipped away because of not enough engineering throughput.

The software factory flips this scarcity on its head.

String together coding agents, design agents, testing agents, and support agents that are constantly collecting feature requests from customers, and the pieces of a true software factory are starting to come online.

So what’s the business case for making your company a software factory?

For tech companies, it’s velocity, it’s sales win rate, it’s retention.

For services companies, it’s diversifying your revenue and improving your margins.

And for everybody else, it’s replacing the off the shelf tools you pay for.

When I was running a company at scale, pre AI, the most difficult recurring meeting of the week was our roadmap review meeting. We did our best to juggle feature requests and bug fixes from current customers, prospects, vision stuff, and infrastructure. It was brutal. We never had enough resources to do all the things we wanted to, that we NEEDED to do. And we had 60 amazing engineers!

Moving from scarcity to abundance sounds good to me. Focus and vision will still matter of course, but your throughput goes to the moon.

A word of caution: if your company can become a software factory, so can everybody else’s. Like everything with AI, market advantages are fleeting, so you best move quickly before the rest of the world wakes up.

Want to talk about software factories? Drop us a line, we’d love to chat.

Lawrence Coburn
Founding Partner & Head of GTM at Legible. Leads executive sessions and coaching.