Tenex details agent-assisted SDLC workflow and compensation model for enterprise AI teams

Tenex, an applied AI consultancy, describes its agent-assisted software development lifecycle (SDLC) and a compensation model that pays engineers for delivered outcomes rather than hours worked. The firm says the system has scaled to 75 employees and is used with Fortune 2000 clients as well as mid-market businesses.

The company’s MetaHarness system is an agent-readable project record stored in Git, which includes skills, project documents, and a CLI. It captures not only what was built but also the rationale behind architectural decisions, enabling AI agents to understand the context of the work. Tenex argues that this structure makes the "surrounding work"—plans, acceptance criteria, decisions, and evidence—legible to both humans and agents.

Tenex retains story points as part of its engineering compensation framework but emphasizes that the biggest challenge is rewarding work that doesn’t fit neatly into tickets, such as understanding customer problems, designing architectures, or defining what to build. The firm’s AI strategists, incentivized by customer retention and growth, help balance output-focused rewards with quality and client value.

The company’s shift from hourly billing to pay-for-output was designed to align incentives as AI makes engineers faster. Tenex reports that feared failure modes—such as engineers inflating estimates or shipping low-quality code to accumulate points—have not been the primary issue. Instead, the harder problem is valuing non-ticket work that drives long-term outcomes.

Tenex’s engagements increasingly focus on broader AI transformation for enterprises, moving beyond pure development work. The firm’s public materials describe MetaHarness as a project record that integrates with Linear, GitHub, and client updates, providing a shared artifact for human and agent collaboration.

Tenex co-founder Alex Lieberman and Director of Engineering CJ Hess discussed these practices in a podcast interview, noting that the company’s approach is designed to make engineering work legible to AI agents while maintaining accountability for customer outcomes.