Software has traditionally been built around a familiar model: identify a market, define a product, assemble a development team, build the application, and sell it to many customers.

But what happens when AI dramatically lowers the cost of creating software?

Could we move toward a world where software is built specifically for one customer but produced through a repeatable, scalable process?

In this episode of Tech Council, Duncan Mapes talks with returning guest Jason McMunn about Oasis and the concept of an AI software factory.

Rather than thinking about AI as simply another developer tool, they explore what happens when AI becomes part of an end-to-end system for producing software. The discussion covers everything from customer interviews and personas to journeys, AI model selection, infrastructure automation, cost management, deployment, and observability.

One of the most interesting ideas is the concept of SaaS for a customer of one. Traditionally, software economics depend on spreading development costs across many customers. But AI could change that equation by making highly customized software economically viable.

That creates a very different set of questions.

How do you standardize the parts of development that can be standardized? Where does human judgment remain essential? When should a company use a powerful commercial AI model versus an open-source alternative? And how do you build infrastructure that can support an expanding number of AI-powered applications without allowing costs to spiral?

The conversation also looks beyond development itself. An AI software factory touches the entire software lifecycle, from understanding what a customer actually needs to operating the finished system in production.

For engineering leaders, technology executives, entrepreneurs, and anyone trying to understand where AI is taking software development, this episode offers a practical look at a model that could fundamentally change the economics of building software.

Top Takeaways:

• Build for a customer of one, not a market segment
• The real moat is not code, it is context
• Small companies can be run with automation sooner than enterprises can
• The product is the workflow around the model, not the model itself
• Vendor lock-in is becoming optional
• AI is compressing decision cycles more than it is replacing labor
• A lot of AI disruption is really organizational slimming

Connect with our Guest:

Jason McMunn – https://www.linkedin.com/in/mcmunn 

Connect with us:

Duncan Mapes

DevGrid.io

DevGrid on LinkedIn

DevGrid on X