Reco Company Spotlight: Mapping Enterprise AI Risk

Reco is a New York cybersecurity company building control infrastructure for enterprise AI agents. Co-founders Ofer Klein, Gal Nakash, and Dr. Tal Shapira started the company in 2020, before agent sprawl became a boardroom phrase and every software vendor discovered a copilot.

The company now sits at a specific pressure point. AI agents do not remain inside a prompt window. They inherit permissions, connect applications, reach corporate data, trigger workflows, and create paths across systems that were never designed to be treated as one security boundary. Reco maps those relationships so security teams can see what an agent owns, what it can reach, and what changes when that access becomes risky.

That mission has attracted $140M in disclosed funding, including a $55M financing announced in September 2026. The money matters because Reco is racing a market that moves every time another enterprise application adds an autonomous feature.

About Reco

Reco was founded by Ofer Klein, Gal Nakash, and Dr. Tal Shapira. Klein serves as CEO, Nakash as CPO, and Shapira as CTO. The company operates across New York, Tel Aviv, and Moldova, combining enterprise go-to-market work with product and security research.

The founding team brings a mix of commercial, technical, and academic experience. Reco's official biography describes Klein as a former Israeli pilot and serial entrepreneur who built US go-to-market teams for SaaS companies. Shapira's academic biography documents research in deep learning, computer networks, and cybersecurity, along with prior leadership in cybersecurity and algorithms. Nakash leads product as the company moves from SaaS posture management toward a broader agent-security platform.

The company's stated job is not to stop AI adoption. It is to make that adoption governable. That distinction matters. Security teams lose credibility when their only answer to a new technology is no, while business teams create exposure when every yes arrives without ownership, inventory, or policy.

The Reco Graph Maps the Chain Reaction

The Reco Platform combines agent discovery, identity governance, posture management, data exposure management, threat detection, and response. Its core is the Reco Graph, a continuously updated knowledge graph connecting identities, permissions, applications, integrations, activity, workflows, and data movement.

Most security inventories flatten those objects into separate lists. Reco's thesis is that the relationship between them is where the real risk appears. An agent may look acceptable by itself. The same agent becomes more dangerous when it can read a shared drive, invoke a workflow in another application, and act through a service account nobody actively owns.

Reco currently markets coverage for more than 280 applications and agents and more than 1,000 detection controls. Reco Factory is designed to add new integrations in days as customers introduce tools the existing catalog does not yet cover. Those figures are company-reported, but they show the operating challenge: an agent-security product has to expand almost as quickly as the software environment it monitors.

Why Agent Sprawl Became a Security Category

Enterprise AI adoption has shifted the unit of risk from a single application toward the chain of identity, permission, action, and data behind it. Prompt filters can inspect what enters or leaves a model, but they do not automatically explain which Salesforce records an agent can change, which files it can retrieve, or which downstream workflow receives its output.

Reco's 2026 agent-security research reported 414 unsanctioned AI tools per 1,000 employees at small and midsize organizations, with 4 in 5 AI tools operating without IT oversight. The same company study found that half of 500 published agent tools could execute shell commands on a host. These are Reco's own research findings, not independent market measurements, but they capture the governance gap the company is selling against.

The strategic question is no longer whether employees will use AI. It is whether enterprises can establish ownership, least privilege, monitoring, and response before agent behavior spreads across the application layer.

Customers Turn Visibility Into Operations

Reco's public customer stories span healthcare, financial services, pharmaceuticals, technology, retail, energy, and legal services. The company names organizations including Tampa General Hospital, Exela Pharma Sciences, Banco BS2, BigID, Wellstar Health, Belk, SolarEdge, Cresta, and CSK Legal.

The common problem is operational rather than theoretical. A hospital may need to discover unassessed apps carrying sensitive data. A bank may need visibility during an acquisition. A security software company may want detection coverage without building every rule internally. In each case, inventory is only the first step. The useful outcome is a prioritized decision, an owner, and a remediation path.

Reco says it grew 400% during 2025 after 500% growth in 2024 and now works with multiple Fortune 500 companies. The company has not published the underlying revenue base or audited growth figures, so the percentages should be read as management-reported traction. The customer roster offers a more concrete signal that agent and SaaS governance has moved into regulated and operationally complex environments.

Funding Expands the Distribution Test

Reco announced a $30M Series B in February 2026, led by Zeev Ventures, after reporting $55M in prior funding. Seven months later, the company added $55M with strategic participation from AT&T Ventures and support from Forestay and Quadrille Capital, bringing total disclosed funding to $140M.

The September financing did not include a round label, valuation, security type, or investor allocation. Reco said the capital would expand sales, partnerships, channels, and customer support. That use of proceeds shifts the next test from product coverage alone toward distribution and repeatable customer outcomes.

AT&T is both a customer and an investor. That relationship gives Reco an enterprise operator feeding real requirements into the product, plus a strategic backer with a large technology ecosystem. It does not guarantee commercial expansion, but it is stronger evidence than a logo added after a financing announcement.

Hiring Shows Where Reco Has to Scale

Reco's careers page displayed current openings across enterprise sales, finance, and product at the time of research. That mix aligns with the company's disclosed plan to build commercial reach while continuing product development.

The hiring signal is not simply that a funded startup is adding people. Reco has to scale several systems together. Sales cannot outrun implementation. Integration coverage cannot outrun detection quality. Customer support cannot become the place where unclear product ownership goes to hide. Finance and operating discipline matter because $140M of disclosed capital creates expectations around durable category leadership, not just another year of impressive percentage growth.

Reco describes its culture through openness, shared wins, results, and putting people first. Those claims belong to the company. The more revealing signal is where it is hiring: close to customers, close to the product, and close to the machinery that turns funding into a company capable of supporting complex enterprises.

What Reco Signals for Enterprise AI Security

Enterprise AI is creating a new security layer between identity systems, SaaS applications, data platforms, and automated action. The winners in that layer will need more than a scanner. They will need context that remains current as agents add tools, inherit privileges, and create connections nobody planned in advance.

Reco's wager is that the graph becomes the control point. If the platform can map who owns an agent, what it can reach, what it actually does, and how that behavior deviates from policy, security teams can govern AI without pretending the business will slow down.

The $55M financing gives Reco more room to prove that thesis through enterprise distribution. The harder work begins after visibility. A map matters when it helps somebody make the right decision before a permission chain becomes an incident.