Plexe

Plexe is an applied AI lab that works alongside product and engineering teams to design, build and deploy production ML models and AI agents, then hands over ownership of the code, models and prompts. It addresses the gap between AI prototypes and systems that run reliably in production, with deployment either inside a customer's own VPC or fully managed.

The underlying platform provides three main modules: Builder, which turns a plain-language problem description into an evaluated model; Agents, which composes agents across a customer's tools and data with guardrails and policy controls; and Runtime, which handles deployment, monitoring, A/B testing, drift detection and retraining. Additional capabilities include an AI agent runtime, predictive ML serving, RAG and vector store, document understanding, workflow and DAG engine, conversational chat UI, eval framework, integration layer, and observability and monitoring. Deployment options are self-hosted in a customer's environment or fully managed.

Plexe is aimed at teams shipping customer-facing AI capabilities inside their own products. Pricing is scoped per engagement rather than sold as seat-based software, with engagement archetypes including a scoped pilot, a full custom build, and an embedded team arrangement.