Jedify

**Product and the concrete problem it solves.** Jedify is aimed at a bottleneck that appears after an enterprise has already bought the models and connected the data warehouse: an AI agent can retrieve a great deal of information and still produce an unreliable answer because it does not understand how the business defines its entities, metrics, permissions, exceptions, and workflows. Corporate knowledge is split among warehouses, CRM and ERP systems, BI tools, documents, playbooks, tickets, spreadsheets, and collaboration systems. A table can tell an agent that an invoice exists, but not necessarily which legal entity owns it, which bridge table determines its status, which business definition controls the metric, or who is authorized to see it. Jedify's product, which it calls an autonomous context graph, is designed to supply that missing layer for teams building enterprise AI applications and agents. Its practical promise is grounded answers and more dependable actions without forcing the customer to rebuild its existing data estate or accept a new, model-specific application stack.

**Core technology and how it works.** Jedify's central technical construct is Semantic Fusion, described by the company as a context graph that unifies structured data, unstructured knowledge, and company-specific business context. The graph is intended to represent not just tables and metadata but entities, relationships, definitions, permissions, operational assumptions, and the history of how data is used. The product can ingest information from data systems and organizational knowledge sources, reconcile the meaning between them, and expose the resulting context to AI agents and conversational analytics. Jedify says the graph is built from the customer's data, query history, and knowledge and becomes more accurate through queries and corrections. In product terms, this is a grounding and orchestration layer: an agent can navigate relationships and business meaning before generating an answer, instead of sending an underspecified retrieval request to a general-purpose model. The company also presents the platform as model-agnostic, which matters to buyers that want to change models or clouds without rebuilding their context layer. The public evidence does not establish proprietary model training, independent hallucination benchmarks, or a formal standard for graph quality, so the technical thesis remains promising infrastructure rather than a proven category monopoly.

**Market, customers, and go-to-market.** Jedify sells into data leaders, analytics teams, AI application builders, and enterprises that have discovered that pilots do not become production systems merely by adding a larger language model. Its initial wedge is likely organizations with complex data relationships and enough internal demand to justify a shared context layer. The company's own material positions Semantic Fusion for autonomous workflows and conversational analytics, while its customer examples show a practical path: Kiteworks connected Snowflake, Tableau, Notion, internal playbooks, documents, and screenshots to build agentic tools for specific workflows. Jedify's public site also names or quotes teams associated with Exodigo and OpenWeb. The commercial motion combines a platform sale with technical implementation, because context quality depends on connecting and reconciling the customer's own systems. That creates expansion potential across departments but also a services and integration burden. The company's U.S. presence and Israeli R&D footprint give it access to American enterprise buyers while keeping product development connected to Israel's data and AI talent base. There is no public customer-count, recurring-revenue, retention, or contract-value disclosure.

**Traction, funding, and third-party validation.** Jedify emerged from stealth with a $24 million Series A announced in June 2026, led by Norwest Venture Partners with strategic participation from Snowflake and participation from existing backers S Capital VC and Cerca Partners, plus Oceans Ventures. CTech reported that the round brought total funding to just over $33 million, following an $8.5 million Seed round in September 2023, and put headcount at 35. Snowflake's investment and integration of Jedify's technology with Snowflake products such as Cortex AI, Semantic Views, and CoWork are important ecosystem signals because Snowflake has both the data-plane distribution and the technical incentive to test context infrastructure. S Capital describes Semantic Fusion as patent-pending technology that contextualizes structured and unstructured data for autonomous data agents and conversational analytics. Jedify's own website carries attributed customer statements about business understanding and data-model accuracy, including a statement from an Exodigo data executive. These are useful reference signals but remain company- or investor-mediated. The public record does not disclose patent numbers, independent product evaluations, audited financials, or a customer deployment count, and the $33 million total should be treated as reported financing rather than a guarantee of commercial maturity.

**Founders and team background.** Jedify was founded in 2023 by Assaf Henkin, Adi Elimelech, and Erik Shani. The founders say they had worked together for more than 15 years at the intersection of data and AI, and their prior operating history gives the company a coherent founder-market fit. Public profiles identify Henkin as co-founder and CEO, Elimelech as co-founder and CTO, and Shani as co-founder and CPO or CBO depending on the source. The trio previously worked at Kontera and Amobee, where they dealt with data-heavy products and business intelligence, and CTech reports the company as Israeli-founded. An Israeli company-registry record identifies Jedify Ltd as an active private Israeli company incorporated in August 2023 with a Tel Aviv address, while the U.S. operating entity is listed in Brooklyn. The team therefore brings more than a generic AI-app background: it has firsthand experience with data fragmentation, analytics products, and enterprise systems. The caveat is that the public record gives limited information about the wider engineering organization, research credentials, security leadership, or experience operating in regulated and classified environments. Three experienced founders and a 35-person team reduce product execution risk, but they do not by themselves prove that context-graph quality scales across customers.

**Competitive dynamics.** Jedify competes at the intersection of data integration, semantic layers, enterprise search, knowledge graphs, and agent infrastructure. Palantir Foundry and AIP offer ontology-driven data integration and operational applications with strong government and enterprise relationships. Databricks and Snowflake can embed catalog, governance, semantic, and agent features directly beside the data plane; Snowflake is also a Jedify investor and partner, which creates distribution opportunity alongside strategic dependence. Glean approaches the problem through enterprise search and an organizational knowledge graph, while Stardog and similar graph vendors provide knowledge-graph and semantic reasoning infrastructure. Data catalog and observability vendors such as Atlan and Monte Carlo can own adjacent metadata and trust workflows, and internal data-platform teams can build bespoke semantic layers. Jedify's claimed edge is the combination of structured and unstructured context, business semantics, query-history learning, model agnosticism, and an agent-ready execution surface. The crucial test is not whether the graph can be built once, but whether it stays accurate as schemas, permissions, definitions, and workflows change. Integrations and graph representations are reproducible; a durable edge would require faster deployment, higher answer accuracy, lower token and analyst cost, and stronger governance evidence than the alternatives.

**Defense, security, and resilience dual-use relevance.** Jedify is credibly dual-use at the infrastructure and resilience layer, although no public source reviewed here establishes a defense or government customer. Defense organizations, intelligence teams, emergency operators, and critical-infrastructure owners face the same context problem as commercial enterprises, but with higher consequences: mission data is distributed across sensor feeds, operational databases, reports, maps, maintenance systems, access-control domains, and human procedures. A context graph that preserves relationships, provenance, permissions, and business or mission definitions could help an authorized AI agent answer questions across those systems without treating every document or row as equivalent. Potential scenarios include logistics and sustainment analysis, infrastructure incident triage, supply-chain risk investigation, cyber-operations knowledge management, and secure decision support in disconnected or sovereign environments. The Israeli-founded team and the product's governance emphasis make the adjacency credible, but the company is not a defense prime and has disclosed no classified deployment, security accreditation, air-gapped installation, export-control posture, or fielded military capability. Its strategic relevance should therefore be scored as enabling infrastructure for secure, context-grounded AI rather than as evidence of operational defense performance.

**Growth stage, trajectory, and key diligence risks.** Jedify is best classified as mid-stage: it has an active Israeli entity, a public product, a 35-person team, an $8.5 million Seed followed by a $24 million Series A, named ecosystem and customer references, and a strategic investment from Snowflake. It remains early relative to the scale required for foundational enterprise infrastructure because public sources do not establish recurring revenue, retention, gross margins, customer concentration, deployment volume, or independent performance benchmarks. The trajectory is attractive if context graphs become a required control layer for reliable agents and Jedify can become the model-agnostic system that enterprises trust across warehouses, applications, and clouds. Diligence should focus on: (1) measured improvement in answer accuracy, task completion, latency, and token cost against native warehouse assistants and enterprise-search tools; (2) permission enforcement and provenance when agents traverse mixed-sensitivity data; (3) integration maintenance and time-to-value across real customer environments; (4) dependency on Snowflake for distribution or product access; (5) the defensibility and scope of the patent-pending Semantic Fusion claims; (6) the risk that Databricks, Snowflake, Microsoft, Palantir, or model vendors bundle equivalent context features; and (7) whether the business can scale as software rather than becoming a high-touch data-integration consultancy. Progression toward mature status would require repeatable production deployments, disclosed commercial metrics, durable retention, and evidence that the graph improves real agent outcomes rather than only demos.