Uvi

AI reasoning layer providing structured, decision-aware context for enterprise AI agents.

Cover Block

From the public record

Field	Value
Name	Uvi
Tagline	AI reasoning layer providing structured, decision-aware context for enterprise AI agents.
Headquarters	Tel Aviv, Israel [Blocks & Files, October 2026]
Founded	2023 [Uvi team page]
Stage	Seed [The Next Web, October 2026]
Business model	SaaS
Industry	Deeptech
Technology	AI / Machine Learning
Geography	Middle East / North Africa
Growth profile	Venture Scale
Founding team	Co-Founders (2): Gilad Salinger and Dror Wolmer [Uvi team page]
Funding label	Seed
Total disclosed funding	~$7,500,000 [The Next Web, October 2026]

The Short Version

PUBLIC Uvi builds an AI reasoning layer for enterprise agents, and it merits attention now because it emerged from stealth in October 2026 with a $7.5 million seed round around a problem that is becoming more visible in production AI deployments: agents often lack reliable company-specific context and decision history [The Next Web, October 2026] [SiliconANGLE, October 2026] [VentureBeat, October 2026]. Founded in 2023 in Tel Aviv by Gilad Salinger and Dror Wolmer, Uvi is positioning itself as infrastructure for enterprises that want agents to do more than retrieve snippets, according to company materials and launch coverage [uvi.io] [Blocks & Files, October 2026].

The product claim is straightforward but technically ambitious: Uvi says it connects code, logs, tickets, documentation, and business systems, then organizes relationships and prior decisions into a graph that agents can query across models including Claude, Gemini, and GPT [uvi.io] [The Next Web, October 2026]. The differentiation, if it holds up in customer environments, rests less on model novelty and more on whether this structured context layer materially improves agent accuracy, actionability, and unit economics in complex engineering and operational workflows [The Next Web, October 2026] [SiliconANGLE, October 2026].

The founding team has some relevant continuity. Uvi identifies Salinger as co-founder and CEO and Wolmer as co-founder and CTO, while public profiles tie both to Naboo, another enterprise AI context-layer effort, which suggests prior exposure to the same organizational knowledge problem they are now addressing at Uvi [uvi.io] [linkedin.com] [Forbes, March 2026].

On capitalization, the company has disclosed a single $7.5 million seed round led by CT Innovations and Cardumen Capital, with participation from Mana VC, 91 Ventures, and Gaingels VC, and the business model is SaaS aimed at enterprise deployments [The Next Web, October 2026] [SiliconANGLE, October 2026]. Public signals of early go-to-market buildout are modest but directionally useful: Uvi's careers pages show U.S.-focused openings in enterprise sales and customer success, consistent with its stated plan to expand its U.S. team after the seed financing [uvi.io] [The Next Web, October 2026].

Over the next 12 to 18 months, the main question is whether Uvi can convert an attractive architecture story into repeatable production adoption with evidence that goes beyond company-reported benchmarks and cost claims, including whether named customer relationships mature into durable references and whether the reasoning layer becomes a system of record inside agent workflows rather than an experimental add-on [The Next Web, October 2026] [TokenPost, October 2026] [uvi.io].

Single-source, plausible -- This section combines independent coverage on funding, launch timing, and investor syndicate with company-sourced product and team claims.

Taxonomy Snapshot

Axis	Value
Stage	Seed
Business Model	SaaS
Industry / Vertical	Deeptech
Technology Type	AI / Machine Learning
Geography	Middle East / North Africa
Growth Profile	Venture Scale
Founding Team	Co-Founders (2)
Funding	Seed, total disclosed ~$7,500,000

The Company in Brief

PUBLIC

The public record on Uvi is still thin, but the basic company shape is clear. Uvi describes itself as a Tel Aviv-based startup founded in 2023 by Gilad Salinger and Dror Wolmer, with Salinger serving as CEO and Wolmer as CTO [uvi.io]. The company frames its core product as infrastructure for enterprise AI agents, built to give those systems structured, decision-aware context drawn from internal company systems rather than relying on document retrieval alone [uvi.io].

The clearest milestone on the timeline is its October 2026 emergence from stealth alongside a $7.5 million seed round led by CT Innovations, with Cardumen Capital also identified as a lead in coverage of the announcement [The Next Web, October 2026]. That financing was reported to include participation from Mana VC, 91 Ventures, and Gaingels VC [The Next Web, October 2026]. Uvi's own hiring pages also suggest an early push into the U.S. market, with open roles in enterprise sales and customer success based in the U.S. at the time of capture [uvi.io].

No public state filing or legal-entity detail was confirmed in the supplied source set, so the overview rests mainly on the company website and the seed-announcement coverage. That is enough to establish founding year, headquarters, leadership, and the first disclosed financing event, but not enough to confirm corporate structure or earlier operating history [uvi.io] [The Next Web, October 2026].

Single-source, plausible -- Founded year, headquarters, founders, and roles are supported by the company website; the seed milestone is supported by one independent public report and referenced in company-adjacent materials.

What They Have Built

MIXED

The product is pitched as infrastructure for a problem that has become more visible as enterprise agents move past simple retrieval. Uvi says its platform gives AI agents structured, decision-aware context drawn from code, logs, tickets, documentation, and business systems, then organizes the relationships and decision history into a graph that agents can query [PUBLIC] [uvi.io] [The Next Web, October 2026]. In the company’s framing, the goal is not only to surface facts but to preserve why a prior decision was made, a distinction that matters in engineering and operational workflows where context can be spread across multiple internal tools [PUBLIC] [The Next Web, October 2026] [uvi.io].

Uvi describes this layer as an "AI reasoning layer" for engineering and agent workflows [PUBLIC] [uvi.io]. Press coverage says the system supports agents built on Claude, Gemini, and GPT [PUBLIC] [The Next Web, October 2026]. The technical architecture beyond that remains mostly company-described rather than independently verified, although the team page attributes the underlying "Decision Graph" and live ETL infrastructure to co-founder Dror Wolmer [PUBLIC] [uvi.io].

The performance case is directionally interesting, but the evidence base is still thin. Uvi has publicly claimed that its structured business context can reduce enterprise-agent answer costs by eight times and bring answer costs to 10 to 16 cents [PUBLIC] [SiliconANGLE, October 2026] [TokenPost, October 2026]. It has also reported 88% correctness on multi-source questions versus 75% for alternatives [PUBLIC] [uvi.io]. Those figures help explain the company’s wedge, but they should be read as company claims until a third-party benchmark, verified customer case study, or public demo with reproducible methodology emerges.

Unconfirmed -- This section relies materially on company website language and company-attributed performance claims in press, with limited independent technical verification.

Market Size and Demand

PUBLIC The market matters now because enterprise interest in AI agents has moved from experimentation to production pressure, and that shift exposes a basic constraint: agents fail when they lack durable, company-specific context rather than raw model capability alone [VentureBeat, October 2026] [The Next Web, October 2026].

The available public record does not support a direct TAM, SAM, or SOM for Uvi itself, so the cleaner way to frame demand is through analogous markets that sit immediately upstream of its product. Uvi positions itself as an "AI reasoning layer" that organizes code, logs, tickets, documentation, and business-system data into a queryable graph for agents [uvi.io] [The Next Web, October 2026]. That places it between enterprise search, retrieval and knowledge infrastructure, observability, and agent orchestration, rather than in a single well-bounded software category [VentureBeat, October 2026].

Public reporting around agentic AI in 2026 is directionally consistent on one point: retrieval alone is proving insufficient once enterprises expect agents to take actions across multiple systems. VentureBeat's October 2026 coverage argues that the next production bottleneck is not generating language, but compiling trustworthy context across fragmented enterprise systems so agents can reason over state, decisions, and dependencies [VentureBeat, October 2026] [VentureBeat, October 2026]. That framing lines up with Uvi's own thesis, but the category remains early, which means budgets may still be allocated under adjacent line items such as AI infrastructure, developer tooling, knowledge management, or automation platforms rather than a dedicated "reasoning layer" budget [The Next Web, October 2026] [uvi.io].

Market frame	What it suggests for Uvi	Source
Agentic AI infrastructure (analogous market)	Buyers may spend to improve agent reliability, evaluation, and orchestration before they spend on a new context layer	[VentureBeat, October 2026]
Enterprise knowledge and retrieval infrastructure (analogous market)	Uvi can be read as a higher-order context system built on top of fragmented internal knowledge sources	[The Next Web, October 2026] [uvi.io]
Developer and operations tooling (adjacent market)	The initial wedge appears strongest in engineering-heavy environments where code, logs, and tickets already shape decision flow	[uvi.io] [The Next Web, October 2026]

The table points to a practical market reality: Uvi may be selling into an acknowledged problem before a settled category exists. That can help with urgency if failures are visible, but it usually makes budget ownership and vendor comparison harder in the first sales cycles.

Demand drivers are easier to observe than market size. One is the increase in multi-system agent workflows, where an answer depends on joining technical telemetry, documentation, historical tickets, and business process context instead of retrieving a single document [VentureBeat, October 2026] [The Next Web, October 2026]. Another is the rising cost of repeated large-model inference, which gives economic weight to any product claiming better context efficiency, although Uvi's cost and performance metrics remain company-reported and should be treated cautiously [SiliconANGLE, October 2026] [TokenPost, October 2026] [uvi.io]. A third driver is enterprise discomfort with confident but incorrect outputs, a theme repeated across October 2026 agent-evaluation coverage and one that supports spend on systems intended to preserve decision history and reasoning traceability [VentureBeat, October 2026] [VentureBeat, October 2026].

Adjacent and substitute markets cut both ways. Enterprises can address the same problem with improved retrieval-augmented generation stacks, internal knowledge graphs, vector databases, observability platforms, or agent frameworks that add their own memory and planning layers [VentureBeat, October 2026]. That means Uvi is not competing only for new budget, it is also competing against the idea that existing infrastructure can be extended far enough to solve context quality without a separate platform [VentureBeat, October 2026] [uvi.io].

Macro and regulatory forces are present, but the evidence in the current source set is still broad rather than Uvi-specific. As enterprises push AI agents into customer, financial, and operational workflows, there is more pressure for traceability, internal controls, and repeatable evaluation, especially when answers are assembled from many systems and can trigger downstream actions [VentureBeat, October 2026] [The Next Web, October 2026]. For a product built around preserving relationships and decision rationale, that backdrop is favorable in principle, but public sources do not yet show whether compliance or governance is a primary buying trigger for Uvi's early customers.

Single-source, plausible -- Market context is grounded in named public reporting, but no third-party market sizing specific to Uvi's category was confirmed, and several category claims rely on analogous markets and interpretation from VentureBeat coverage and company positioning.

Who Else Is Fighting for This

MIXED Uvi is positioning itself one layer below the application agent and one layer above raw enterprise data, competing less on model ownership than on whether it can become the system that structures company-specific context before an agent acts [The Next Web, October 2026] [uvi.io] [VentureBeat, October 2026].

Company	Positioning	Stage / Funding	Notable Differentiator	Source
Uvi	AI reasoning layer for enterprise agents, organizing code, logs, tickets, documentation, and business systems into a decision-aware graph	Seed, $7.5M disclosed	Claimed focus on decision history and structured reasoning context rather than basic retrieval	[The Next Web, October 2026] [uvi.io]
Zep	Memory and context infrastructure for AI agents	Not established in the provided public sources	Known in the category for agent memory positioning, but specific differentiators were not verified in the provided research	[Structured facts]
Mem0	Memory layer for AI applications and agents	Not established in the provided public sources	Known in the category for persistent memory positioning, but specific differentiators were not verified in the provided research	[Structured facts]
Modern Relay	Enterprise context and relay layer for AI systems	Not established in the provided public sources	Competes around enterprise context orchestration, though architecture details were not verified in the provided research	[Structured facts]

The field is starting to split into three practical groups. Second are reasoning-layer or orchestration products, the lane Uvi is claiming through its "Decision Graph" language and its emphasis on why a decision was made, not only where a document sits [uvi.io] [The Next Web, October 2026]. Third are adjacent substitutes, including in-house RAG stacks and broader agent frameworks, which may not market themselves as direct competitors but can absorb this budget if enterprise teams believe retrieval plus workflow logic is sufficient [VentureBeat, October 2026].

Uvi's present edge, on the public record, is conceptual clarity more than proven scale. The company is making a specific argument that enterprise agents fail because context is fragmented across code, logs, tickets, documentation, and business systems, and that a graph of relationships and decision history is a better substrate than simple retrieval for production agent behavior [uvi.io] [The Next Web, October 2026] [VentureBeat, October 2026]. That edge could become durable if the graph improves with each new system connected and each decision path captured, because those relationships are company-specific and not easily replicated by a generic model layer. It is still perishable today because the performance claims remain company-reported, and because larger platform vendors can move into the same layer if customers treat structured context as a feature rather than a standalone control point [SiliconANGLE, October 2026] [TokenPost, October 2026] [VentureBeat, October 2026].

The sharpest exposure is that memory-layer competitors may be easier to adopt and easier to explain. Uvi is also exposed to the build-versus-buy instinct inside strong engineering organizations, especially the type already operating internal data pipelines, observability tooling, and documentation systems. Publicly surfaced customer names such as Lemonade, monday.com, Melio, Global-e, and Fiverr suggest the company is at least speaking to technically sophisticated buyers, but the available sources do not establish whether those are scaled production deployments, which matters when judging whether Uvi has already crossed the adoption gap from interesting infrastructure to embedded system of record [The Next Web, October 2026] [uvi.io].

The most plausible 18-month scenario is that the category consolidates around whichever layer can show measurable agent quality gains with the least implementation burden. Uvi is the most plausible winner if enterprises decide that decision-aware context, not just retrieval memory, is the missing production layer and if its claimed cost and accuracy advantages hold up in third-party deployments [SiliconANGLE, October 2026] [TokenPost, October 2026] [uvi.io]. The reverse is also possible: if customers optimize first for speed of integration, then the lighter memory vendors would have a cleaner path and Uvi's heavier architecture would face a longer sales cycle.

Opportunity

Upside Case

PUBLIC The prize here is not a point tool for prompt quality, it is a chance to become a control layer that enterprise AI agents rely on before they are trusted to take actions inside production workflows.

The clearest upside is that Uvi becomes default infrastructure for decision-aware enterprise context, especially in engineering-heavy environments where agents need to reason across code, logs, tickets, documentation, and business systems rather than retrieve isolated documents [The Next Web, October 2026] [uvi.io] [Blocks & Files, October 2026]. That outcome is reachable, not merely aspirational, because the company has already emerged from stealth with a disclosed $7.5 million seed round led by CT Innovations and Cardumen Capital, plus participation from Mana VC, 91 Ventures, and Gaingels VC [The Next Web, October 2026] [SiliconANGLE, October 2026]. Just as important, public reporting and Uvi's own sales materials point to early customer references including Lemonade, monday.com, Melio, Fiverr, and Global-e, although the public record does not establish whether each relationship is a paid production deployment or a pilot [The Next Web, October 2026] [uvi.io].

The underlying market logic is straightforward. As agent deployments move from answering questions to taking actions, the bottleneck shifts from model quality alone to context integrity, traceability, and decision history, a theme that wider enterprise AI coverage has begun to surface [VentureBeat, October 2026] [The Next Web, October 2026]. Uvi's pitch is that an agent should understand not only what information exists, but also which decisions were made and why, and that framing positions the product closer to operating infrastructure than to a narrow retrieval feature [The Next Web, October 2026] [uvi.io].

Scenario	What happens	Catalyst	Why it's plausible
Engineering system of record	Uvi becomes the standard context layer for AI agents operating across software delivery, incident response, and internal engineering support	A category shift toward production agent workflows that require cross-system reasoning rather than document-only retrieval [VentureBeat, October 2026]	Uvi already describes integrations across code, logs, tickets, documentation, and business systems, which is the right surface area for engineering and operations use cases [uvi.io] [The Next Web, October 2026]
Reference-account expansion	A handful of recognizable customers turn into broader enterprise adoption across adjacent teams and geographies	One or more named accounts expand usage from a narrow workflow into company-wide agent deployments [The Next Web, October 2026]	Public materials already cite customers such as Lemonade, monday.com, Melio, Fiverr, and Global-e, which suggests at least an initial beachhead in companies with meaningful internal software complexity [The Next Web, October 2026] [uvi.io]
Multi-model orchestration layer	Uvi wins as a model-agnostic context substrate used beneath Claude-, Gemini-, and GPT-based agents	Enterprises standardize on mixed-model environments and need a persistent reasoning layer above the model tier [The Next Web, October 2026]	Uvi is publicly described as supporting agents built on Claude, Gemini, and GPT, which aligns with buyer demand to avoid coupling context architecture to a single model vendor [The Next Web, October 2026]

The compounding mechanism, if this works, is partly technical and partly organizational. Every additional deployment can deepen Uvi's internal graph of relationships, decision history, and system interdependencies inside a customer environment, which should make the product more useful as more workflows are routed through it [uvi.io] [The Next Web, October 2026]. That is not a network effect in the consumer sense, but it can still create strong switching costs: once an enterprise has mapped how operational decisions connect across its systems, replacing that layer becomes harder than swapping an interface or a single model endpoint.

There is also an early go-to-market flywheel visible in the public record. Uvi's two surfaced U.S. roles, Enterprise Sales and Head of Customer Success, line up with the company's stated intent to use seed proceeds for U.S. hiring and sales expansion, suggesting management sees repeatable commercialization, not pure research, as the next gating factor [uvi.io] [The Next Web, October 2026]. If reference customers produce credible internal ROI around answer quality or cost, even where those metrics are still company-reported, that can tighten sales cycles for the next wave of design partners [SiliconANGLE, October 2026] [TokenPost, October 2026].

The upper bound is still difficult to anchor because there is no confirmed market-size dataset in the current source set, and no public valuation was disclosed with the seed round [The Next Web, October 2026]. Even so, a reasonable upside frame is that if Uvi becomes a foundational control layer for enterprise agents across large R&D organizations, the outcome could resemble infrastructure companies that sit persistently in the application stack rather than workflow-specific copilots, which would support venture-scale value creation far beyond a feature sale price (scenario, not a forecast) [VentureBeat, October 2026] [The Next Web, October 2026]. The burden of proof remains on adoption depth and production durability, but the shape of the opportunity is real: if enterprises conclude that agent performance depends on decision-aware context as much as on model choice, the category could become core infrastructure rather than optional tooling [VentureBeat, October 2026] [Blocks & Files, October 2026].

Single-source, plausible -- This section relies on two independently reported public sources for the funding round and product framing, plus company materials for product detail and customer references that are not fully independently verified [The Next Web, October 2026] [SiliconANGLE, October 2026] [uvi.io].

Sources