Although AI vendors tout their enterprise products as turnkey solutions, it is unlikely that AI agents will be put into production anytime soon. Unless you make an effort to train your model based on the details of your business, you’re unlikely to understand, for example, how your company defines revenue or who is allowed to see what files. This is one reason why we see AI companies bringing in engineers to help integrate their AI products into their customers’ systems.
New York-based startup Jedify is challenging this very gap. The company says its platform connects to enterprise knowledge sources via APIs to build a “context graph” about the business that AI agents can use to work better. These sources include databases, data warehouses and lakes, SaaS apps, or BI tools, as well as reports, documents, code bases, and even unstructured sources like Slack channels and meeting recordings.
To build it, Jedify has raised $24 million in a Series A funding round led by Norwest, TechCrunch has learned exclusively. The round included participation from new investor Oceans Ventures as well as previous backers S Capital VC and Cerca Partners. Data giant Snowflake has also joined as a strategic investor, integrating its technology with AI products such as Cortex AI Services, Semantic Views, and CoWork.
Jedify’s pitch is that for an AI agent to be useful within an enterprise, it needs access to relationships between entities, data, permissions, domain knowledge, workflows, operational assumptions, and enterprise-specific terminology. According to the company, this context allows AI agents to focus their attention on information relevant to a specific task, rather than searching everything a company has.
Co-founder and CEO Assaf Henkin (pictured above, far right) cited compliance firm Kiteworks as an example of how customers are using Jedify. Kiteworks connected Snowflake, Tableau, Notion, and internal playbooks with documents and screenshots to Jedify to build agent tools for various customer workflows.
“They wanted to give merchants and account teams a sophisticated app. You can think of it as both a dashboard application and a real-time conversation application. When you enter a conversation with a customer, Jedify builds all the information the customer needs to know on the fly. And during the conversation, you can proactively reveal very specific details in real time,” Henkin said.
Henkin argues that Jedify’s context graph is multidimensional, unlike the semantic layers, metadata catalogs, and knowledge graphs that enterprises already use, because it captures the relationships between entities, data, people, permissions, and customers. It is also model agnostic and updates in real time as information enters and exits connected systems.
“If you want to really make your agent solution autonomous and drive decision-making across CRM data, Zendesk tickets, and perhaps telemetry data coming in in real-time, then a context graph is much better at what it does than a semantic layer,” he said.
Permissions are clearly a hurdle here. For example, it wouldn’t be appropriate for an agent to give an intern access to the CFO’s revenue forecasts. Henkin said the company’s platform addresses this issue by inheriting permissions from identity systems, file systems, SaaS tools, and databases, including row-level, column-level, and table-level access rules, and allows customers to create additional groups to define who and what agents and workflows have access to. It also provides observability and governance tools that allow customers to verify that their AI agents are working as intended.
Jedify currently targets medium and large enterprise customers with mature data stacks and multiple databases or data warehouses. Henkin said the company has 10 to 20 early customers, including The Weather Company, and is seeing interest from data-heavy sectors such as gaming, industrials and consumer packaged goods.
Snowflake’s investment and partnership are notable as large data platforms are also looking to build similar capabilities. But Henkin argues that Jedify complements those efforts, since much of a company’s data and most of its organizational knowledge is typically not stored in a single cloud provider.
“[Large data companies]will say, ‘Oh yeah, bring it all.’ But the reality is that companies have multiple databases, warehouses, data solutions (…) The point is, not all the data is in those environments, most of the knowledge is not there, so it’s actually a bit of a disadvantage that companies have,” he said.
Henkin also pointed out that for companies looking to do this in-house, training an AI model to build an equivalent context layer can be cost-prohibitive, especially as companies scrutinize and police the use of AI tokens.
And rapid advances in AI model development have implications for the company’s broader bets. As models become more capable and interchangeable, the unique context that helps those models function better within an enterprise can become a valuable and durable moat.
Startups will use the new cash for product development, hiring, and go-to-market activities. This brings the company’s total funding to approximately $33 million.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
