Autonomous context graphs get Jedi powers

Jedify, an AI context graph startup, has raised $24 million in Series A funding to develop technology that gives AI agents contextual understanding of enterprise data relationships. The New York-based company, founded in 2023 by CEO Assaf Henkin, CTO Adi Elimelech, and CPO Erik Shani, will use the capital to accelerate product development and expand its workforce. Jedify's platform creates live context graphs that help AI agents navigate fragmented enterprise data across multiple systems, enabling more accurate and business-ready responses.

Autonomous context graphs get Jedi powers AI context graph startup Jedify has raised $24 million in an A-round to further develop its bringing of Jedi powers to autonomous context graphs for AI agents. Jedify, based in New York, was founded in 2023 by CEO Assaf Henkin, CTO Adi Elimelech, and Chief Product Officer Erik Shani. It raised an $8.5 million seed round that year. The trio wanted to bring context to AI agents so that they could understand relationships between the entities they were dealing with. The funding will be used to accelerate product development, expand go-to-market and recruit more staff. CEO Henkin said: “In order for an agentic workflow to really work well for an enterprise at scale, it needs a very deep understanding of that business. Enterprise data is fragmented across systems, definitions, permissions, and workflows. Jedify turns that fragmented knowledge into a live context graph that agents can use to produce accurate, cost-efficient, business-ready answers." Jedify says that, while AI agents and models can generate fluent answers, they cannot determine things like which definition of revenue to use, which customer record is current, or which operational assumptions matter unless that context is available at runtime. Enterprises are sitting on vast, complex data spread across dozens of SaaS tools, data warehouses, CRMs, financial systems and unstructured sources like documents, Slack and meeting recordings. Manually pulling that data together into reliable, AI-ready context is slow, costly, and typically has to be rebuilt from scratch for every new agent or workflow. The process needs to be automated and the semantic context, the relationships between things an agent deals with, the entities in its little universe, are best captured using graph technology https://www.blocksandfiles.com/storage-management/2022/08/18/graph-database/1595670? gl=1 1xy2ala ga MzkxNDQyMTIwLjE3NzcwMzc0NTc. ga NSDTXHMMN0 czE3ODEwNzk5ODkkbzk5JGcxJHQxNzgxMDgwMDA3JGo2MCRsMCRoMA.. . CTO Adi Elimelech blogs https://jedify.com/context-graph-for-ai-agents/ : ”AI agents operate differently from human analysts. They execute autonomously, chain decisions across multiple steps, and need to know not just what a metric means today, but who owns it, how it has been defined historically, which source to trust when two sources disagree, and what counts as an anomaly worth acting on. A semantic layer gives an agent a definition. A context graph gives it the decision-making infrastructure to act on that definition correctly.” She says:”A semantic layer sits between your warehouse and your BI tool. It stores metric definitions, dimension hierarchies, and join logic so that every analyst querying “monthly recurring revenue” MRR gets the same number regardless of which tool they’re using.” But “An AI agent querying your warehouse for a RevOps forecast doesn’t just need to know what MRR means. It needs to know: Which of your three MRR tables is canonical for this quarter, given that finance reconciled Q3 manually in a Snowflake view that sits outside your semantic layer That churn events from one product line are tagged differently from another, because two engineering teams made incompatible schema decisions in 2022 That the VP of Sales considers any account under $10K ARR a long-tail account — a business rule that lives in a Slack thread from 18 months ago That the anomaly it’s seeing in last week’s data is a known billing cycle artifact, not a signal worth escalating.” “None of this is a metric definition. It’s institutional knowledge — the kind a senior analyst carries in their head and deploys silently every time they build a report. Semantic layers were never designed to store it. They store what a metric is. Context graphs store what an agent needs to know to use that metric correctly.” These graphed relationships represent intrinsic organizational knowledge and this data shouldn’t be handed over to AI Agent or foundational model suppliers. Jedify points out that enterprises that hand over such data to the same vendors selling them tokens face inherently misaligned incentives. Those vendors benefit from the least efficient, most token-intensive solutions, creating a clear conflict of interest. And when the company building your context layer is also the one charging you per token to use it, the economics rarely favor the customer. Jedify was built as an independent, model-agnostic context layer that provides agents with the relevant business meaning they need at runtime, without locking enterprises into a single model vendor. Its software autonomously builds a customer-specific context graph, powered by its patent-pending Semantic Fusion technology, on top of an enterprise's existing data and knowledge infrastructure. By connecting structured operational data from data warehouses, CRMs, financial systems and BI tools with unstructured knowledge documents, playbooks, Slack, meeting recordings , Jedify creates a continuously updated live , AI-ready semantic model that deeply understands how a business actually works. The context graph captures metric definitions, entity relationships, lineage, permissions, business rules, operational assumptions and domain-specific terminology, giving AI agents the runtime context needed to generate more accurate, grounded responses. The A-round funding was led by Norwest, with a strategic investment from Snowflake Ventures and participation from existing investors S Capital VC and Cerca Partners, as well as new investors Oceans Ventures. Assaf Harel, a partner at Norwest who has joined Jedify’s board, said: "Jedify is solving a foundational problem by autonomously fusing structured and unstructured data into a context graph that gets smarter with every interaction. Its compounding value and model-agnostic approach give enterprises flexibility rather than lock-in, which is exactly the kind of durable infrastructure layer we look for.” Bootnote We think that graph technology will become widespread within the AI agent world. Here are four examples of activity in this space; Israel-based Illumex https://www.blocksandfiles.com/ai-ml/2024/06/28/illumex-secures-13m-to-combat-chatbot-hallucinations/1612871 , a startup developing Generative Semantic Fabric graphs, for LLMs, was acquired by Nvidia in February this year for between $60 and $75 million. CEO and founder Inna Tokarev Sela is now a Director at Nvidia. Neo4J https://www.blocksandfiles.com/ai-ml/2025/10/03/neo4j-bids-to-take-graph-technology-into-ais-mainstream/1613209 is developing graph technology for AI agents as is Memgraph https://www.blocksandfiles.com/ai-ml/2025/02/15/storage-news-roundup-15-february-2025/1586841? gl=1 1r3xrj3 ga MzkxNDQyMTIwLjE3NzcwMzc0NTc. ga NSDTXHMMN0 czE3ODEwNzk5ODkkbzk5JGcxJHQxNzgxMDgwNDYwJGo2MCRsMCRoMA.. . Data protector Druva https://www.blocksandfiles.com/data-management/2026/03/02/druva-uses-graph-relationships-to-mine-metadata/4092745 has added data graph technology to organize backup metadata for use by its Deep Analysis Agents https://www.blocksandfiles.com/security/2026/02/25/druva-adds-agentic-memory-to-speed-forensic-compliance-probes/4091863 .