Davis Palmie co-founded Lumetric with Ross Cefalu and joined Factory through its acquisition of the company in May 2026. His work has moved from personalized audio to financial analysis and autonomous software development, exploring how AI systems can turn information into useful deliverables and carry tasks across tools.
From enterprise engineering to personalized audio
Palmie’s earlier roles included software consulting in Slalom’s ARC Labs and senior software engineering on Visa’s Next Generation Services team. At Slalom, he worked on prototypes and helped build the innovation lab’s engineering team. His personal site describes these roles and his subsequent work as a founder.
Palmie and Cefalu met as freshmen at Georgia Tech and later co-founded PocketPod, with Palmie serving as CTO. Their company joined Y Combinator’s Winter 2024 batch. PocketPod combined language models, information processing, and text-to-speech to create personalized podcasts: daily news or deeper explorations of subjects chosen by the listener, delivered through listening platforms such as Spotify. Generating and distributing the audio let listeners receive tailored information without a human producer creating every episode.
Completing professional workflows
Their subsequent product, Lumetric, offered an AI coworker for deal teams. Its deliverables included financial models, presentations, reports, and cleaned datasets; concrete tasks included assembling comparable-company tables, reconciling rent rolls against lease documents, and preparing investment-committee memos. These workflows extended the founders’ work from generating listening material to handling context across tools and producing structured business deliverables.
Factory’s acquisition brought Palmie and Cefalu’s team into autonomous software engineering, with their team assigned to help shape the next Factory Desktop App. In their joint acquisition announcement, the founders connected their work on contextual, cross-tool AI systems with Factory’s longer-horizon agents and desktop application.
Palmie also co-authored a 2024 study of AI-assisted neuromuscular examinations. The collaborative research proposed combining language processing and computer vision to divide telemedicine recordings into examination tasks and support quantitative scoring for myasthenia gravis. An essential distinction was between a doctor explaining an exercise and a patient performing it: scoring needed to use the active testing interval. The proposed workflow retained clinician review of the automated analysis.
From coding agents to a software factory
In his software-factory framework, Palmie argues that faster code generation leaves bottlenecks in review, debugging, documentation, and testing. A software factory connects incoming signals—such as bug reports and user feedback—to the work needed to reach production. Engineers increasingly govern agents while continuing to set direction.
His approach emphasizes several practical choices:
Model and deployment control: Systems should support different models and deployment under an organization’s control. Integration across the software development lifecycle, with agents available where teams already work, makes autonomy useful beyond a single coding interface.
Incremental autonomy: Begin with a narrow workflow, such as incident triage, then expand through governance and validation. This gives teams a bounded starting point for introducing autonomous work.
Outcome metrics: Palmie rejects token consumption as a productivity proxy. Signal-to-production time, human interventions, and cost per pull request measure how agent activity translates into engineering results.
Davis Palmie explains how agents can connect incident triage, planning, code, tests and deployment—and why the path to autonomy starts with shared context, narrow responsibilities and human judgment.
A software factory connects incident signals, planning, documentation, implementation, review, testing, deployment and monitoring; generating code is one step in that journey.
Start with a narrow, verifiable responsibility such as turning a Sentry alert and traces into a diagnosis for human review. Establish ownership and monitoring before connecting steps into a loop.
Model choice should balance task effectiveness, speed and cost, while deployment fits organizational policies and shared context connects the development lifecycle.
Documented builds, tests, reproducible environments, modularity, observability and security checks improve working conditions for both agents and engineers.
Humans retain architecture, strategy and validation responsibilities. Measure signal-to-production time, interventions, repair time, code shelf life and cost per PR rather than rewarding token consumption.
▶ Watch ↗