The Minimum Viable Unit of Saleable Software
Short summary
The author breaks down how large language models lower software development costs but don’t eliminate human-driven feedback loops and ongoing maintenance expenses. By comparing real-world SaaS prices (Jira at $400/month vs. Salesforce at $500/seat) to engineer-hour costs, he defines a “zone of viability” where buying remains cheaper than LLM-powered rebuilding. He frames his own project River against this threshold to gauge its business potential today.
- Rebuilding cheap SaaS with LLMs doesn't pay off: replacing $400/month Jira takes over three years to break even at $96/hour engineer costs, even with minimal maintenance.
- Expensive per-seat SaaS like Salesforce ($25,000/month for 50 seats) crosses into "build" territory since that budget covers 1.5 full-time engineers.
- The "zone of viability" for buy-vs-build depends on both price and novelty/difficulty of re-implementation, not price alone.
- River (Go/Postgres job queue, $125/month Pro tier for up to 20 devs) is positioned to stay on the "buy" side because its design and performance edge make LLM replication costly despite feature copyability.
Full summary
When companies compare buying software to building it with LLMs, the old rule was “only build what’s core.” Now tools like GPT and Claude can crank out code faster, but they don’t eliminate human costs. The author points to a LinkedIn story: a team dropped $400/month on Jira in favor of an internal tracker generated and maintained via Claude. At $96/hour for an engineer earning $200k/year, even cutting maintenance to two hours a month, it still takes over three years to match what they’d pay Atlassian. That math makes rebuilding low-cost tools a losing bet.
On the other hand, expensive SaaS like Salesforce at $500 per seat per month (50 seats = $25,000/month) flips the script. You could hire the equivalent of 1.5 full-time engineers for that money. Building a replacement CRM becomes feasible. The author calls this span the “zone of viability”: software priced low enough to favor buying (Jira) and high enough to favor building (Salesforce). Where a product sits in that zone depends on novelty (how hard it is to re-implement) and price.
He’s betting his side project River lands on the buy-not-build side. River is an open-source Go/Postgres job queue. Basic features are free; advanced workflows, concurrency controls and billing APIs are behind a Pro tier. At $125/month for teams up to 20 developers, it sits in the zone where ongoing license fees undercut LLM-driven rebuild costs. The author acknowledges LLMs could copy River’s features, but argues the design and performance edge makes replicated quality expensive to reach.