AI is changing enterprise software development. The shift reaches beyond faster coding, changing how teams organize work, measure contribution, and decide what software should deliver.
Tenex offers a clear example. The company started as an AI-powered development shop and has grown to 75 people, with customers split between Fortune 2000 companies and mid-market businesses. Alex Lieberman, Tenex’s cofounder, and CJ Hess discussed the company’s approach to building an AI-native software development lifecycle, or SDLC.
At Tenex, AI engineers are paid for the work they deliver rather than the hours they spend at a keyboard. Lieberman described the idea as “Paying for output when the job keeps changing.” The phrase captures a basic problem with traditional software work: when AI changes how the job gets done, time spent typing becomes a poor measure of value.
From Coding Hours to Business Outcomes
Most conversations about AI-assisted software development still focus on productivity. Luis Aburto, Founder and CEO of Scio, put the familiar question plainly: “Most conversations about AI-assisted software development focus on productivity: ‘How much faster can developers work?’”
Aburto’s argument shifts the target. AI is lowering the cost of producing code, but the value moves toward outcomes such as increasing revenue, lowering cost, reducing risk, or improving reliability. Faster production matters only when the resulting software creates one of those outcomes.
That distinction matters for enterprise customers. A lower cost for code does not automatically create a better business result, and a faster development process does not answer whether the work solves the right problem. AI changes the economics of production, but companies still need to judge the result by what it accomplishes.
Tenex’s operating model reflects that change. The company uses a shared set of skills, project documents, and a CLI called MetaHarness around its work. Plans and decisions are checked into Git alongside the code, keeping the reasoning behind a project near the software itself instead of separating decisions from implementation.
When AI Agents Become the Development Team
Tenex also tested this approach through its “Kill My SaaS” episode, which covered 72 submissions to replace speaker-management software. One example involved replacing a tool that cost $43,000 a year with AI agents — an unusually direct way to frame what enterprise software can become when code production costs fall.
The contest produced a revealing split. The team that won with human judges came 19th on the AI judge’s leaderboard. That result does not settle which evaluation method is better, but it shows that human and AI judges can rank the same work in very different ways.
The episode also turns an abstract claim into a concrete business question: what happens when an existing software cost of $43,000 a year becomes a target for AI agents? The answer depends on outcomes, not on how many lines of code an engineer or agent produces. The software must still reduce cost, lower risk, improve reliability, or create another measurable result.
Tenex’s approach connects those questions across the full SDLC. Shared skills and project documents give the work a common structure, MetaHarness provides a CLI around it, and Git stores plans and decisions alongside code. The result is an AI-native process built around deliverables and recorded decisions rather than keyboard time alone.
That model also explains why incentive design matters. If the work keeps changing, paying for hours can reward activity even when the outcome misses the business need. Paying for delivered work points attention toward what the software achieves — which is where AI’s economic promise has to survive.
The broader lesson is direct: AI-assisted development is not only a productivity story. Tenex’s 75-person operation, its mix of Fortune 2000 and mid-market customers, and its experiments with AI agents show how software companies are testing a different relationship between people, code, and business value.
Based on
