Enterprise software veterans have become all too familiar with the gap between flashy artificial intelligence demos and their performance in real-world production environments. The reality is that a lot of agentic AI deployments leave lots to be desired, and that’s the problem a new startup called Tribal AI Inc. is trying to solve after raising $10 million in funding today.

The seed funding round was led by Team8 and saw participation from DYDX Capital and a group of unnamed enterprise software founders behind multiple exits in the Salesforce ecosystem, who participated as angels.

Tribal is making the case that the biggest problem with failed enterprise AI agents is that they lack the organizational knowledge, impact analysis and governance required to operate safely at scale. While AI coding tools have mainly been successful, Tribal says agents have not had much success automating other kinds of enterprise work given the lack of information about how enterprises operate.

The startup’s solution is “metadata-native” AI agents that are powered by its proprietary Metadata Fabric framework. This serves the agents by ingesting and mapping the full metadata layer of each organization’s chosen enterprise system of record. It gobbles up all of the important context around objects, automations, permissions, dependencies and business rules, and uses this knowledge as a foundation for automating business workflows and application development.

The platform was originally designed to work with Salesforce Inc.’s customer relationship management software, ensuring that Agentforce agents operate safely in customer’s environments by giving them access to full organizational context. Every output is tested to ensure it abides by the company’s rules.

Tribal says its main advantage is that its founders have a wealth of expertise working with enterprise software. Chief Executive Yoav Kolodner (pictured, center) served as a vice president of engineering at Salesforce for years, while Chief Operating Officer Yakir Daniel (left) built two startups that were later acquired by NetApp Inc. and Huawei Technologies Co. Ltd. A third co-founder is Chief Technology Officer Lior Sidi (right), who led the AI teams at Wix.com Inc. and saw firsthand the importance of context for production workloads.

The co-founders say they all experienced the same thing: Building AI prototypes is easy, but shipping systems that can be trusted to work correctly in live enterprise environments is an order of magnitude more difficult. “The shift to headless enterprise infrastructure is real and accelerating,” said Kolodner. “But infrastructure without organizational context is just plumbing. This raise lets us go deeper on the product and build the intelligence that will make enterprise AI tick.”

Although it’s still building its platform, the startup’s ambitions now extend beyond Salesforce, and it hopes to bring its metadata-native agents to enterprise platforms such as ServiceNow, SAP, NetSuite and Workday.

Ori Barzilay of Team8 said Tribal’s raise is timely because enterprises are under pressure to move on from AI experiments and start showing real and repeatable impact from those investments, before their competitors get there first. “The Tribal team brings deep and rare experience building on the trusted platforms of global enterprises and is uniquely positioned to help organizations bring AI safely to fulfill core business needs,” he said.

Image: Tribal AI