Mend.io

• Mend.io's own about page states that three of five AI security vendors are now inside big platforms, so the AI line it prices separately at up to 300 dollars per developer a year competes against capabilities a buyer may be offered inside a platform it already owns.
• The customer count and Fortune 100 share a buyer is most likely to hear are Mend.io's own 2022 statement relayed through press, so a buyer weighing the company's reach today has no cited source that confirms its current scale.
• Mend.io publishes its open source vulnerability database on its own site and describes it as an aggregation of the NVD, security advisories, and open source project issue trackers.
• The detection-quality figures Mend.io reports, including 38 percent better precision and 48 percent better recall than benchmark competitors, rest on its own pages with no third-party benchmark in the cited record, so an independent evaluation reporting weaker results would undercut the positioning.
• Mend AI inspects prompts and responses in production, deployed either inside the application or as a standalone proxy, so a security review of that line turns on data-handling terms the fetched pages do not state.
• Renovate is published under an open licence and Mend.io offers a free self-hosted tier, so the paid Renovate Enterprise line rests on the scale and workflow features around the tool rather than on access to it.

Mend.io sells security for the whole span of what a development team ships, from the open source packages it pulls in to the custom code it writes to the AI models and agents now embedded in its applications. The audience is the security and development teams inside an enterprise. Mend's homepage argues that AI components, open source dependencies, and production agents have introduced risk that traditional developer security tools were not built to find, test, or govern.

The pain is tied to named regulation rather than to generalities. Mend's homepage names the EU AI Act, Executive Order 14028, and the Cyber Resilience Act as requiring verifiable technical evidence that software and AI systems have been inventoried, tested, and secured. Mend AI scans code to surface shadow AI, the models and libraries developers adopted without security or governance review.

Independent evidence supports the demand rather than the size of the harm. Enterprise Technology Research's March 2026 survey of 517 security leaders found that protection for large language models and generative AI had overtaken cloud security as the top cybersecurity budget priority, and a Thales survey relayed by Help Net Security put dedicated AI-security budgets at 30 percent of respondents, up from 20 percent the year before. Those measure what buyers are funding, not how often the risk Mend describes actually bites. [s1, s2, s17, s18]

Mend.io runs three priced lines rather than a single point product. Mend SCA finds open source risk using what the company calls a unique reachability analysis, showing whether code interacts with vulnerable functions in both direct and transitive dependencies, then ranks findings with CVSS 4.0 severity and EPSS exploitability data and exports SBOMs in SPDX and CycloneDX with VEX data. Mend SAST scans proprietary code across 30-plus languages including Java, Python, Go, and C/C++, detects hardcoded credentials, API keys, tokens, and certificates, and, Mend says, can run on-premises so source code never leaves the customer's perimeter.

Mend AI extends the platform to the AI layer of an application. It maintains a real-time inventory of every model, framework, and AI agent in that application, including hard-to-detect shadow AI and the tools and MCP servers agents connect to, hardens system prompts, and runs prebuilt red teaming scenarios against conversational AI for prompt injection, context leakage, data exfiltration, jailbreaks, hallucinations, and bias. It measures a program against the OWASP Top 10 for LLM Applications, the NIST AI Risk Management Framework, ISO/IEC 42001, and the EU AI Act.

The runtime layer now ships. Mend says Mend AI inspects every input before it reaches the model and every output before it reaches the user, catching prompt injection, jailbreak attempts, sensitive data exposure, and harmful content in real time, and announced in July 2026 that those guardrails are deployable in the application or through a standalone proxy or interface. Detection-quality figures stay Mend's own: the 38 percent better precision and 48 percent better recall it reports for static analysis carry no third-party benchmark in the fetched pages. [s3, s4, s2, s5, s23]

Mend.io competes on two fronts at once and says so in print. It publishes head-to-head pages against established application security vendors, telling readers that Snyk is racing to become an AI security company while Mend.io already is one, and that Checkmarx focuses on remediating code faster inside the editor while Mend.io also secures the models, agents, and system prompts generating that code. On the AI side it runs the same play against Noma Security, arguing that Noma brings AI asset discovery and posture management while Mend AI adds supply chain security, prompt hardening, and red teaming in one product.

The stated differentiator is one governed system across both layers. Mend's about page puts it as the dangerous gaps living at the seams between layers, and says that is exactly where Mend.io sits. The company also argues that AI tools let developers ship faster than ever, and that doubling output without security governance grows the attack surface faster.

Independence is the other selling point. Mend's about page observes that three of five AI security vendors are now inside big platforms and states that Mend.io is not beholden to any model provider. [s24, s25, s26, s7]

Named enterprise references speak on record. Andrei Ungureanu, a security architect for WTW's ICT business, says his team's remediation time has gone down considerably, and in a testimonial on the Mend AI page he puts the reduction at at least 80 percent. Chris Wallace, a senior security architect, credits Mend with letting Vonage align speed, security, and compliance in a competitive market. Mend.io's own July 2026 announcement quotes Alen Pešikan, a principal software engineer at Span Software and AI Solutions, on Mend.io helping his team face the risks AI brings in generated code and in AI components.

The channel motion is documented but dated. CRN reported at the 2022 rebrand that Mend worked closely with systems integrators and planned to strengthen and deepen its channel ties, quoting the company on strong relationships with GuidePoint Security and Defy Security and on a move toward a full channel-led business model over two years. No newer cited source reports where that move landed.

The headline scale figures are the company's own and four years old. CRN and Help Net Security both relayed Mend's May 2022 statement of more than 1,000 customers including about a quarter of the Fortune 100, alongside revenue up 800 percent over three years and enterprise net retention of 127 percent in 2021. So a buyer weighing Mend.io today has the named references and the published prices, and no cited source that updates the customer count. [s13, s2, s23, s9, s10]

Mend.io's founders had built and sold a security company before this one. Globes reported that CA was acquiring Eurekify, an Israeli identity and role management developer that Ron Rymon founded in 2002 and that Azi Cohen ran as chief executive, and KuppingerCole's profile of Rymon records the two of them managing Eurekify up to that acquisition. TechCrunch reported that Sass, Cohen, Rymon, and Roni Einav went on to start WhiteSource in 2011, and Israel's companies registry lists that entity as active from its 09/05/2011 incorporation.

The company changed chief executive in December 2025. Mend.io announced that Azi Cohen, co-founder and president, had been appointed chief executive officer with immediate effect, and that co-founder Rami Sass would move to general manager for Mend AI. The current about page lists Cohen as co-founder and chief executive and Gil Regev as general manager of Mend AI, and does not list Sass.

Press has followed the product line as it widened. TechCrunch notes that WhiteSource was among the early companies to coin the term software composition analysis, and SecurityWeek recorded the move into static analysis. [s27, s28, s14, s12, s22, s7]

Mend.io publishes the assurance package an enterprise security review expects. Its trust page states an integrated information security management system incorporating controls for ISO 27001, ISO 27017, and ISO 27701, reviewed externally every year, and a SOC 2 Type II assessment evaluated by Schellman with the report available through the account team. It also states GDPR commitments for a company headquartered in Israel with operations across the EU, the US, and worldwide.

Those are credentials a funded competitor can obtain. No federal authorization or government mandate attached to the product appears in the reviewed sources, so nothing in that record blocks a replacement on compliance grounds and the attestations read as enterprise entry cost.

The newest layer is where a review will concentrate. Mend AI inspects prompts and responses in production, and Mend.io states that those guardrails can be deployed inside the application or as a standalone proxy or interface, so a security team will ask where model traffic goes and on what terms. The one adversarial record among the cited sources is NVD's CVE-2020-5304, a log injection in the legacy WhiteSource Application Vulnerability Management dashboard fixed in version 20.4.1. [s6, s2, s23, s11]

Add analyzed competitors to compare them side by side with Mend.io.

Mend.io sells to the security and development teams inside an enterprise, reaching developers where they already work. The about page frames the pain as the dangerous gaps living at the seams between layers, and argues that AI tools let developers ship faster than ever while doubling output without security governance grows the attack surface faster. The AI line reaches that same buyer through a platform it already owns, and Mend also sells it standalone.

The base is enterprise and, where it is named, carries its own governance obligations. The named reference is WTW's Insurance Consulting and Technology business, whose legal teams treated copyleft licences as an area of particular concern. The wider scale claim, more than 1,000 customers including about a quarter of the Fortune 100, is the company's own May 2022 statement relayed by CRN and Help Net Security, with no newer cited source updating it.

Published prices widen who can evaluate Mend.io without widening who can buy it. The pricing page lists ceilings of up to 1,000 dollars per developer per year for Mend AppSec, up to 300 for Mend AI, and up to 250 for Mend Renovate Enterprise, so a team can size the cost before a conversation. Each of the three tiers still closes on a demo request, so the purchase itself runs through sales.

Mend.io's technical claim rests on deciding what actually matters in a large codebase. Mend SCA employs what the company calls a unique reachability analysis, showing whether code interacts with vulnerable functions in both direct and transitive dependencies, then ranks findings with CVSS 4.0 severity and EPSS exploitability data. Mend SAST covers 30-plus languages spanning web, mobile, server-side, and infrastructure-as-code, detects hardcoded credentials, API keys, tokens, and certificates, and can fail the build before exposed secrets reach production.

The AI line applies the same shape one layer up. Mend AI keeps a real-time inventory of every model, framework, and agent in an application, including shadow AI and the tools and MCP servers agents connect to, runs prebuilt red teaming for prompt injection, context leakage, data exfiltration, jailbreaks, hallucinations, and bias, and measures the program against the OWASP Top 10 for LLM Applications, the NIST AI Risk Management Framework, ISO/IEC 42001, and the EU AI Act.

Runtime protection is the newest piece and the one with the least outside evidence. Mend says the product inspects every input before it reaches the model and every output before it reaches the user, and announced in July 2026 that the guardrails deploy in the application or through a standalone proxy or interface. The efficacy figures around the platform stay Mend's own, including the 38 percent better precision and 48 percent better recall it reports for static analysis, with no third-party benchmark in the fetched record.

The motion is enterprise sales with a channel layer on top. CRN reported at the 2022 rebrand that Mend worked closely with systems integrators, quoted the company on strong relationships with GuidePoint Security and Defy Security, and recorded a stated plan to move to a full channel-led business model over two years. Where that plan landed is not in any cited source.

Product marketing does a lot of the qualifying work. Mend.io publishes per-developer price ceilings and head-to-head pages against named rivals on both the application security and AI fronts, which lets a buyer self-assess fit and cost before contact. The conversion step is still a demo request on every pricing tier.

The named references now reach the AI line as well as the older business. A security architect for WTW's ICT business credits at least an 80 percent cut in remediation time and a senior security architect names Vonage, both on code security. Mend.io's own July 2026 announcement quotes Alen Pešikan, a principal software engineer at Span Software and AI Solutions, saying Mend.io has helped his team face the risks AI brings in generated code and in AI components, so the AI line now has a named reference in the fetched record rather than only the code-security portfolio.

Mend.io prices per contributing developer and publishes the rate. The pricing page lists Mend AppSec at up to 1,000 dollars per developer per year, Mend AI at up to 300, and Mend Renovate Enterprise at up to 250, so cost grows with engineering headcount and with the modules a team turns on. Mend AI can be bought as an add-on to Mend AppSec or as a standalone product.

Mend defines the unit tightly. A contributing developer is any employee or contractor who uses the web application or writes, develops, or modifies scanned code, and the page states that the same individual is not counted more than once even when acting in two roles. Mend answers that there are no additional fees per gigabyte, arguing that managers have better visibility into headcount growth than into the size of their software or lines of code.

Publishing the number cuts both ways for Mend.io. A team can size the platform from the rate card and the counting rule before it talks to anyone, and the same published unit is available to anyone building a competing quote. In-app runtime guardrails sit inside the Mend AI tier rather than the base platform, so the newest capability is a separate line on a buyer's quote.

Mend.io is delivered where developers already work. Mend SCA feeds vulnerability information with reachability analysis into AI code assistants for remediation directly in the workflow, and secrets detection in Mend SAST triggers automated policy violations that can fail a build before exposed credentials reach production. Both paths put the control inside the pipeline the team already runs.

One line is deliberately not cloud-only. Mend says Mend SAST performs scanning on-premises or inside the customer's own environment so proprietary source code never leaves the perimeter, while findings, dashboards, and policy management run in the Mend cloud. Mend positions that split for regulated and high-IP industries, and it means the platform is not uniformly operated by Mend.

The AI line adds operational surface a buyer has to accept. Mend says Mend AI inspects every input before it reaches the model and every output before it reaches the user, and that these runtime guardrails deploy in the application or as a standalone proxy or interface. Which of those a customer picks decides where its prompts and responses are handled, and the fetched pages state no data-handling terms for either, so a buyer should ask for them.

Mend.io publishes the assurance package an enterprise review expects. The trust page states an integrated information security management system incorporating controls for ISO 27001, ISO 27017, and ISO 27701, reviewed externally every year, and a SOC 2 Type II assessment evaluated by Schellman with the report available through the account team. GDPR commitments are stated for a company headquartered in Israel with operations across the EU, the US, and worldwide.

Compliance is also sold as a product feature. Mend positions inventory, findings, test results, and remediation status in one governed workflow that it says gives security and compliance teams the structured, auditable record regulators require, and Mend AI measures a program against the OWASP Top 10 for LLM Applications, the NIST AI Risk Management Framework, ISO/IEC 42001, and the EU AI Act. The characterization of what regulators require is Mend's own.

No federal authorization or government mandate attached to the product appears in the reviewed sources, so the attestations Mend.io holds read as ordinary enterprise entry cost rather than something that blocks a replacement. The assurance question that will take real work is the AI line, because reviewing a control that inspects live model traffic turns on data-handling terms the fetched pages do not state.

Mend.io is built to be one vendor across code and AI rather than a point tool. The platform covers open source dependencies, custom code, secrets, dependency updates, and AI components under a single per-developer billing relationship, and the company's argument for that shape is that the dangerous gaps live at the seams between layers.

Outward, the platform reaches the developer ecosystem through open source. Renovate, the dependency updater behind Mend Renovate, is published on GitHub under the AGPL-3.0 licence with more than 22,000 stars and about 3,300 forks, supports over 90 package managers, and is available self-hosted at no cost. Mend describes it as the trusted standard for automated dependency updates running on millions of repositories, which is the company's own characterization.

Inward, the same breadth concentrates a customer's application security operations on one supplier. Standardizing open source, custom-code, and AI-component security on Mend.io puts accumulated findings, policy, and audit evidence in one place, which is the consolidation Mend sells and also the concentration a buyer wary of a single supplier weighs against it.

Mend.io's founders had built and sold a security company before this one. Globes reported CA acquiring Eurekify, an Israeli identity and role management developer that Ron Rymon founded in 2002 and that Azi Cohen ran as chief executive, and KuppingerCole's profile of Rymon records the two of them managing Eurekify up to that acquisition. TechCrunch reported that Rami Sass, Cohen, Rymon, and Roni Einav went on to start WhiteSource in 2011, and Israel's companies registry lists that entity as active from its 09/05/2011 incorporation.

Leadership turned over in December 2025. Mend.io announced that Azi Cohen, co-founder and president, had been appointed chief executive officer with immediate effect and that co-founder Rami Sass would move to general manager for Mend AI. The current about page lists Cohen as co-founder and chief executive and Gil Regev as general manager of Mend AI, and does not list Sass.

The Series D investor list is on record. Calcalist reported that the 75 million dollar round was led by Pitango Growth with existing investors M12, Susquehanna Growth Equity, and 83North participating.