Endor Labs

• A code-platform vendor such as GitHub or GitLab, or a model host such as Hugging Face, could bundle open source AI model inventory and scoring, eroding the AI Model Discovery line Endor leads its marketing with before it becomes a paid differentiator.
• Endor's reachability and code-context-graph differentiators are reproducible by a funded rival, so sustained growth depends on execution and reference depth rather than a structural data or compliance barrier.
• Endor competes against consolidating AppSec suites and against the rivals independent work places in the same commercial SCA peer group, so a buyer standardizing on one vendor could displace Endor even where its reachability context runs deeper.
• Endor's disclosed traction figures, 30x ARR growth and weekly scan counts, are vendor-published and not independently audited, so a slowdown would be hard to detect from the public record until a later funding event.

Endor addresses the gap between how much code modern teams ship and how much application security can review, a gap that AI coding assistants widen. The buyer is the AppSec or engineering owner who must govern open source dependencies, code, and pipelines across many teams without drowning developers in non-actionable alerts.

The problem is evidenced beyond Endor's own framing. Gartner tracks software supply chain security as a named category, and press coverage ties the urgency to the surge of AI-generated code that outpaces manual review.

Reachability is Endor's answer to the noise. It determines whether a known vulnerability is actually invocable through an application's code paths, which reframes the buyer's problem from counting CVEs to fixing the ones that matter. [s2, s5, s1]

Endor's AURI platform is built on a code context graph that maps how code, dependencies, container images, and services connect. On top of that graph it runs full-stack reachability, AI SAST, secrets detection, and agentic remediation, and the vendor reports large noise reductions such as Cursor's 97.5% in an attributed case study.

The AI capabilities split into two parts. AI Model Discovery inventories open source AI models from Hugging Face and scores them on security, activity, popularity, and operational integrity, and an Agent Kit installs Endor security agents inside AI coding assistants.

The platform applies agentic AI reasoning and deterministic program analysis together, which is the same automation direction its AppSec peers describe. The reachability graph and the open source dataset behind it are the technical core a buyer evaluates.

Outside documentation now describes the mechanism. Microsoft's Defender for Cloud documentation states that Endor Labs reachability findings feed natively into its cloud security posture surfaces, including Cloud Security Explorer and Attack Path analysis, that the GitHub app clones and scans every repository in an organization on a 24-hour cycle, and that an unreachable verdict removes a finding from remediation duty under compliance standards such as FedRAMP. Academic work from Paderborn University, SAP, and Fraunhofer IEM names Endor Labs SCA among the popular commercial scanners while documenting a gap in dependency scanning generally, since dependencies that have been recompiled, rebundled, or repackaged evade approaches that read project metadata. [s2, s3, s4, s12, s13]

Endor sits in a crowded application security field. The peer group the reviewed sources establish is the commercial SCA set named by the 2026 Paderborn, SAP, and Fraunhofer IEM study: Snyk, Mend, Black Duck, and SonarQube. The reviewed sources do not compare Apiiro, Cycode, or Legit Security with Endor, so no ranking against them is drawn.

The competitive risk runs in two directions. Consolidating code-security suites pull buyers toward a single vendor, and code platforms or model hosts could move into the open source model inventory space Endor's AI line occupies.

Endor's defensible edge is the enterprise install base and the reachability context accumulated in customer pipelines. A larger vendor could add the AI-model-discovery feature Endor now markets, which is the most absorbable part of the offering.

Independent researchers place Endor Labs in a defined peer group and read it less favorably than the company does. A 2026 study from Paderborn University, SAP, and Fraunhofer IEM lists Endor Labs SCA alongside Snyk, Mend, Black Duck, and SonarQube, calls that group metadata-based, and reports that nearly half of the 1,808 most popular open source Java Maven projects on GitHub carry at least one modified, hidden dependency with a known vulnerability that metadata-based scanning misses. Endor's own materials do not use the metadata-based label, so the characterization runs against the code-level reachability positioning the company leads with. [s2, s4, s6, s14]

Endor's traction is anchored by named enterprise references. Press named OpenAI, Rubrik, Snowflake, and Dropbox as customers, the homepage customer logo wall also names Atlassian, Glean, Robinhood, and Zebra, and attributed case studies such as Cursor carry quantified results.

Endor pairs that customer evidence with disclosed growth and a displayed analyst placement. It reports 30x ARR growth since its 2023 Series A and over a million scans each week, and it displays a Gartner Visionary placement in software supply chain security on its site.

The motion sells the platform to enterprise AppSec and engineering teams, and the AI line gives the sales team a current reason to re-engage accounts as AI coding spreads across the development organization.

Analyst coverage reaches past the placement Endor displays on its own site. Omdia published an On the Radar profile of the company in May 2024, written by chief analyst Rik Turner, describing Endor Labs as a governance platform for selecting, securing, and maintaining open source software. The report body sits behind an Omdia subscription, so the reviewed evidence here is its public abstract. [s5, s6, s10, s11]

Endor Labs was founded in 2021 in Palo Alto by Varun Badhwar and Dimitri Stiliadis and came out of stealth in October 2022. Both founders are serial security entrepreneurs who led Prisma Cloud through hypergrowth at Palo Alto Networks.

The background pairs application-security domain depth with proven company-building, since both founders are serial entrepreneurs who led Prisma Cloud through hypergrowth at Palo Alto Networks.

Investor backing adds a signal. DFJ Growth led the $93 million Series B with Salesforce Ventures and existing backers Lightspeed, Coatue, and Dell Technologies Capital, established investors whose participation signals conviction in the team.

Outside recognition arrived early. Dark Reading covered Endor Labs in July 2023 as one of four finalists in the Black Hat USA startup competition and reported Varun Badhwar's claim that a third of the research and development team have earned doctorates, which puts a research-depth marker on the engineering bench beyond the founders. [s7, s6, s15]

Endor maintains a trust center at trust.endorlabs.com, which is live but sits behind a Cloudflare human-verification challenge that blocks automated retrieval, so its full document list could not be read in the reviewed sources.

Endor's own blog reports a clean SOC 2 Type I audit result, the commercial baseline expected of a vendor that reads source code and dependencies. The accessible reviewed sources confirm only SOC 2 Type I, and because the trust center was not retrievable, any SOC 2 Type II or ISO status could not be verified from this evidence.

This attestation eases enterprise procurement without creating a barrier a funded rival could not also clear. No accessible cited source verifies a federal authorization or a mandatory sector-specific certification. [s9, s8]

Add analyzed competitors to compare them side by side with Endor Labs.

Endor sells to application security and engineering teams at companies whose developers ship code, including AI-generated code, faster than security can review it. The platform targets the alert-noise problem that reachability is built to cut, separating invocable vulnerabilities from the long tail that scanners flag.

The buyer is the AppSec or product-security owner, not an individual developer. Endor's named references span AI labs and large enterprises, including OpenAI, Rubrik, Atlassian, Dropbox, and Snowflake, which places the segment at the demanding, high-stakes end of the market.

The AI line addresses the same buyer from a newer angle. AI Model Discovery frames open source AI models as assets the AppSec owner must inventory and score, which extends the existing relationship rather than opening a separate market.

Endor's AURI platform is built on a code context graph that maps how code, dependencies, container images, and services connect. On top of it run full-stack reachability, AI SAST, secrets detection, and agentic remediation, and the vendor reports large noise cuts such as Cursor's 97.5% in an attributed case study.

The AI capabilities split into two parts. AI Model Discovery inventories open source AI models from Hugging Face and scores them on security, activity, popularity, and operational integrity, and an Agent Kit installs Endor security agents inside AI coding assistants.

Endor combines agentic AI reasoning with deterministic program analysis, the same automation direction its AppSec peers describe. The reachability graph is the technical core, and the cited materials emphasize open source packages and code relationships, so the reviewed public sources do not verify a protected non-public corpus behind it.

Endor's motion sells the platform to enterprise AppSec and engineering teams, and press named OpenAI, Rubrik, Snowflake, and Dropbox as customers. The homepage customer logo wall also names Atlassian, Glean, Robinhood, and Zebra, and attributed case studies such as Cursor carry quantified results.

Disclosed growth reinforces the enterprise pitch. Endor reports 30x ARR growth since its 2023 Series A and over a million scans each week, and it displays a Gartner Visionary placement in software supply chain security on its site.

The AI line gives the sales team a current reason to re-engage existing accounts as AI coding spreads, layering model discovery and AI SAST onto the reachability platform a buyer already runs.

Endor's pricing page names the packaging without naming a number. Six products are sold separately, Code, Open Source, AI Coding Agent Governance, Package Firewall, Patches, and SBOM Hub, each behind a request for pricing, and buyers who want the whole platform are pointed to a sales conversation about bundling.

The pricing metric is disclosed. Endor states that pricing is seat-based and varies by SKU, with seats defined by contributing developers. Editions run from a free Developer tier to paid Core and Pro tiers.

What the public record withholds is the price level. No figures appear for any product or edition, so buyers reach a number through a quote. The absence is a disclosure gap in the price level rather than in the pricing model.

The catalog records a SaaS deployment for Endor Labs AI Model Discovery. The platform integrates with the code, dependencies, containers, and CI/CD it analyzes, and with AI coding agents, so the integration surface spans the development toolchain rather than a single console.

Endor positions AURI as an independent enforcement layer that integrates with AI coding agents through hooks, skills, MCP, or a CLI. That places part of the delivery surface at the developer's AI assistant rather than only in a central console.

The operational burden sits with the customer's AppSec and engineering teams, who tune policy and act on findings. This is software the buyer runs, not a managed service that accepts accountability for outcomes.

Endor maintains a trust center at trust.endorlabs.com, which is live but sits behind a Cloudflare human-verification challenge that blocks automated retrieval, so its full document list could not be read in the reviewed sources.

Endor's own blog reports a clean SOC 2 Type I audit result, the commercial baseline expected of a vendor that reads source code and dependencies. The accessible reviewed sources confirm only SOC 2 Type I, and because the trust center was not retrievable, any SOC 2 Type II or ISO status could not be verified from this evidence.

This attestation eases enterprise procurement without creating a barrier a funded rival could not also clear. No accessible cited source verifies a federal authorization or a mandatory sector-specific certification.

Endor positions AURI as an integrated but independent security layer that gives AI coding agents security context wherever they work, enforcing policy across every agent rather than letting an agent verify its own output.

The integration surface spans repositories, CI/CD, containers, and AI coding assistants through hooks, skills, MCP, and a CLI. That reach into the development and agent toolchain is where the platform earns its place in a buyer's stack.

The ecosystem position cuts both ways. The same toolchain breadth that embeds Endor also puts it on ground that code platforms and model hosts could contest, a competitive risk the reviewed sources do not settle either way.

Endor Labs was founded in 2021 in Palo Alto by Varun Badhwar and Dimitri Stiliadis and came out of stealth in October 2022. Both founders led Prisma Cloud through hypergrowth at Palo Alto Networks.

The founding background pairs application-security domain depth with proven company-building, since both founders are serial entrepreneurs who led Prisma Cloud through hypergrowth at Palo Alto Networks.

DFJ Growth led the $93 million Series B with Salesforce Ventures and existing backers Lightspeed, Coatue, and Dell Technologies Capital. That backing from established investors signals conviction in the team and the category.