MIT Professor’s Startup G5 Labs Raises $14M to Launch an AI Agent Coordination Platform

On September 15, G5 Labs, a startup founded by MIT computer science professor Tim Kraska, emerged from stealth with $14 million in seed funding and launched its first product, the G5 platform. The platform aims to translate enterprise requirements into a structured semantic graph using a system ontology, coordinating collaboration among AI agents and making natural language the new source code.

G5 is described as a secure, cloud-based web platform that converts business requirements, architectural decisions, policies and other human intent into a system ontology—a structured semantic graph that informs both humans and AI agents of the system's intentions and goals. A bidirectional compiler can uplift existing source code into the semantic layer and drive implementations back down into conventional languages and frameworks.

COO David Chang said the ontology captures data models, business rules, workflows and company-wide policies. Each ontology node is written in natural language and linked to the code that implements it, making every line of code traceable to a requirement. The system constantly audits the ontology for consistency and the code for adherence.

Kraska stressed that this is not better prompting but a new abstraction layer, where natural-language intent becomes compilable, mergeable, diffable and governable source code—similar to how high-level languages once replaced assembly. The platform sits above models and coding harnesses, coordinating tools such as Claude Code or Codex while remaining model-agnostic and able to select different models for different tasks.

G5 Labs' early customers are engaged in software modernization and replacing third-party SaaS applications, with roughly 90% of the organizations in financial services. Pricing has not been made public, and the company is deploying the platform in several regulated industries.