Salt Security

• Edge, cloud, and gateway platforms could bundle API and agent security into infrastructure buyers already own, collapsing the standalone budget line Salt sells against. Akamai's Noname acquisition and Harness's merger with Traceable show the absorption is underway.
• The agentic repositioning could outrun the budget. If buyers do not fund agentic security as its own line by 2027, the March 2026 platform competes in crowded API security carrying extra vocabulary.
• The traction record is own-voice on scale. Company releases carry the revenue and customer-scale claims, and no independently reported revenue figure or funding event newer than February 2022 appears in the fetched record, so a stall would stay invisible until late.
• The 2022 valuation could overhang any exit or raise. Noname sold to Akamai for less than half its 2021 peak valuation, and a comparable repricing of Salt's $1.4 billion mark would narrow its options.
• AI-native security startups could anchor agent security at the model and prompt layer. If buyers budget agent security alongside their LLM tooling rather than their API program, Salt's argument that every agent action is an API call loses its procurement force.

Salt Security sells to large enterprises whose APIs have multiplied beyond what security teams can inventory or govern. The company's pages name the concrete failures it hunts, exposed internal APIs, authentication and authorization weaknesses, hardcoded tokens and credentials, and risky configurations that nobody notices until an attacker does.

In 2026 the company widened the problem to AI agents. Its March 2026 platform announcement argues that agents create risk through what they do across MCP servers and APIs rather than through what models say, and Gartner research relayed by Help Net Security projects that by 2028 most API consumption will come from AI agents rather than human developers.

Outside corroboration now reaches past a single Gartner line. KuppingerCole's 2025 review of the market states that you cannot secure AI without securing APIs, the same framing Salt leads with, which moves the problem statement from vendor assertion toward analyst consensus. The quantified pain still runs through Salt's own survey research, such as the 92 percent of organizations it calls unprepared, so the scale of the problem stays vendor-produced even as its shape is independently shared. [s6, s7, s13, s25, s20]

The Salt Agentic Security Platform covers three phases on one graph, discovery of every API, agent, and MCP server, posture analysis of the misconfigurations and excessive permissions among them, and runtime detection that flags abuse as it happens. The platform page names API discovery and posture, while the homepage and the free assessment spell out discovery of shadow agents and undocumented MCP servers. The runtime page describes Salt Collect observing real traffic, and partner documentation describes behavioral baselines spanning hundreds of attributes for each user or entity.

The 2026 additions extend the same machinery to AI assets. The platform pairs posture management with detection and response across LLM connections, MCP servers, and APIs, Siemens describes using it to scale AI across its software business, and Salt Code carries policy enforcement into AI coding assistants including Claude, Cursor, GitHub Copilot, Windsurf, Codex, and Gemini CLI. A free assessment maps an organization's agentic attack surface starting from a provided domain, with no internal access required.

External validation now runs through two analyst sources rather than one. KuppingerCole evaluated the platform for its 2025 Leadership Compass and, in the placement the company announced, named Salt an Overall Leader, and Gartner Peer Insights carries a 4.6 rating across 56 customer ratings. The site still offers no public documentation portal and routes access through a demo request rather than a self-service trial, so the visible capability depth is real but analyst-mediated. [s2, s1, s7, s5, s9, s19, s25, s21, s24]

Salt Security competes on two fronts, consolidated platforms and repositioning specialists. Akamai bought Noname in 2024 for $450 million to fold API security into its edge business, Harness completed a merger with Traceable in March 2025, and independent rivals Cequence and Wallarm have rebranded around securing AI agents and workloads, the same turn Salt made in March 2026.

Against the consolidators, Salt argues architecture. Its comparison page positions Akamai's offering as API security bolted onto a WAF and its own platform as built for the agentic attack surface, vendor framing that nonetheless names the right incumbent. The CrowdStrike relationship gives Salt an ally whose ecosystem reaches buyers its own sales force would not.

Analyst placement validates the position while shortening its shelf life. KuppingerCole's evaluation means buyers can shortlist Salt without coaching, and every platform vendor weighing a Noname-style acquisition reads the same market the analysts cover. [s15, s29, s30, s23, s10, s16, s25]

Named customers anchor the traction story across regulated and industrial sectors. A Siemens security officer endorses the agentic platform on the record, and the customers page logo wall plus matching case-study pages name accounts including Standard Bank Group, SoFi, Cathay Bank, Hyundai, Stryker, Luxottica, and Carrefour.

Independent corroboration of adoption now exists beyond the vendor's logo wall. Gartner Peer Insights carries a 4.6 rating across 56 customer ratings, with reviewers describing deployments, which is outside evidence the customer count alone does not provide. The partner program lists technology integrations with CrowdStrike, Wiz, Google Cloud, Azure, AWS, Kong, and HCL AppScan alongside channel partners M3corp and Iris Networks, and SecurityWeek reported a strategic CrowdStrike investment in 2022 that deepened that alliance beyond co-marketing.

The independent record of revenue still ends with the 2022 funding coverage. SC Media verified the $140 million round and $1.4 billion valuation in February 2022, while the company's own releases and surveys carry every later customer-count and growth claim. [s18, s31, s24, s32, s16, s12]

Co-founders Roey Eliyahu and Michael Nicosia still run Salt Security as CEO and COO, and the published bench includes a chief marketing officer, chief financial officer, and executive VP of technologies. The about page describes the founders as cybersecurity entrepreneurs who started Salt in 2018.

Salt Labs carries the team's public technical credibility, and that record is now corroborated outside the company. The Hacker News covered its Expo OAuth disclosure, which NVD rated critical as CVE-2023-28131 at 9.6, and TechTarget covered its ChatGPT plugin research, a sustained and independently documented stream of in-domain API vulnerability work. The recurring State of API Security report adds to the same research voice.

That independently documented research record is a stronger team-credibility signal than founder pedigree alone, and it is what the score depends on. [s3, s8, s26, s28, s27]

Salt Security announced completing a SOC 2 audit on its own blog, tying the audit to the controls behind its cloud service. That post is dated January 2019 and is not surfaced from the current footer, so the attestation reads as old and buried. The company surfaces no current trust portal or vulnerability disclosure policy in its public pages as of June 2026, and newer assurances such as ISO 27001 or FedRAMP are neither displayed nor independently inspectable, a notable gap for a vendor whose collectors watch customers' API traffic.

Compliance positioning and the customer roster stand in for current attestations. The posture product sells audit readiness, a dedicated EU AI Act page targets incoming AI regulation, and banks and a credit union among named accounts suggest Salt clears regulated-buyer diligence, though the public pages do not document those reviews. [s11, s6, s31]

Add analyzed competitors to compare them side by side with Salt Security.

Salt Security sells globally to large enterprises, and the named customer set shows where the platform lands. The customers page and its case-study roster name financial institutions such as Standard Bank Group, Cathay Bank, SoFi, and Solaris, alongside industrial and consumer names including Siemens, Hyundai, Stryker, Luxottica, and Carrefour. The spread suggests the company segments by API exposure and regulatory pressure rather than by vertical.

The newest public proof of demand is current, which matters for a vendor founded in 2018 rather than a recent entrant. Siemens went on record in March 2026 about running the agentic platform, and the company repositioned its entire pitch after the current model generation shipped rather than leaving it where the API security market settled. Release activity in the past two quarters, a new platform in March and Salt Code in June, is category-level change rather than incremental polish.

The economic buyer the materials address is the security leader who already owns an API security line item, while the assessment-led entry point targets the CISO confronting unsanctioned AI adoption. Persona work below that level stays generic on public pages, and no pricing page exists to reveal segment tiers.

The claimed advantage is behavioral context accumulated from traffic rather than a wrapped frontier model. Salt's partner documentation describes mirroring API traffic into the platform, where models baseline typical behavior across hundreds of attributes per user or entity and flag deviations, and the customers page credits patented AI that correlates the activity of millions of users in parallel. The homepage frames the agentic extension as the Agentic Security Graph spanning every agent, MCP server, and API.

The agentic platform extends the same graph thinking to AI assets. Salt's homepage says discovery maps every agent, MCP server, and API, posture management analyzes those components for misconfigurations, excessive permissions, and exposed credentials, and runtime protection covers API activity generated by agents and MCP tool usage. Salt Code pushes policy enforcement into AI coding assistants, with Claude, Cursor, GitHub Copilot, Windsurf, Codex, and Gemini CLI named at launch.

External validation now runs through analyst coverage and user reviews. KuppingerCole listed Salt among the vendors it evaluated for its 2025 Leadership Compass and, in the placement the company announced, named it an Overall Leader, while Gartner Peer Insights carries a 4.6 rating across 56 customer ratings that Gartner presents as individual end-user opinions rather than its own analyst view. The site exposes no public documentation portal or sandbox and routes access through a demo request rather than a self-service trial, so an evaluator outside a sales motion sees marketing pages, the free assessment, analyst coverage, and Salt Labs research rather than testable product.

Salt Security runs a hired, channel-assisted enterprise motion, the stage-appropriate shape for a company of this maturity. A chief marketing officer and chief financial officer appear on the published leadership team, executives such as VP of Product Strategy Nick Rago front product launches, and CEO Roey Eliyahu handles analyst and press commentary. What the cited pages establish is a specialist executive bench rather than a founder-only front line, and no sales leadership role appears among the named executives.

Distribution spreads across technology and channel partners. The partner program pairs integrations with CrowdStrike, Wiz, Google Cloud, Azure, AWS, Kong, and HCL AppScan with channel partners M3corp and Iris Networks. The CrowdStrike relationship runs deeper than co-marketing, because SecurityWeek reported a strategic investment from CrowdStrike in 2022 alongside joint integrations.

Independent demand evidence now reaches past the vendor's own logo wall. Gartner Peer Insights carries a 4.6 rating across 56 customer ratings, with reviewers describing live deployments, outside corroboration the customer count alone does not provide. The assessment-led top of funnel asks only for a domain and promises findings without agents, tickets, or internal access, while demo access otherwise requires a request form.

Salt Security publishes no pricing, and the site has no pricing page at all, which signals large negotiated deals aimed at enterprise procurement. Public materials give an evaluator no unit to compare against the size of the problem, the pattern that fits incumbents in a budget line rather than challengers trying to reframe one.

An outside description of the model now exists where the vendor's own pages stay silent. Gartner Peer Insights describes the platform as subscription priced by the volume of API traffic and the features selected, which is an outside read on the metering even if the figures stay private. Whether agent and MCP coverage prices separately from API coverage is still undisclosed, a question the agentic expansion makes commercially significant for renewal conversations.

The assessment-led entry point is the one visible packaging signal. Salt offers discovery of shadow agents and exposed APIs up front, and the paid contract covers the posture and runtime layers that act on those findings, an inference from the page structure that the Gartner pricing note broadly supports.

For API runtime protection, the cited pages establish out-of-band analysis of mirrored API traffic, and they do not describe how that analysis is hosted. The partner pages describe mirroring API traffic from cloud networks into the platform, the runtime page describes Salt Collect analyzing real traffic, and integration pages cover AWS, Azure, GCP, Kong, HCL, GitHub, and CrowdStrike, the surfaces where enterprise API traffic and findings already live. Out-of-band collection keeps the product off the request flow, which lowers deployment risk relative to inline gateways.

Time-to-value claims center on discovery. The agentic assessment promises a mapped attack surface with no agents, no tickets, and no internal access, and the homepage makes an in-minutes claim for seeing the agentic graph. Operational collateral such as SLAs, uptime commitments, or a status page does not appear among the site's public pages.

Salt Security announced completing a SOC 2 audit on its own blog, where it ties the audit to the controls behind its cloud service, but that post is dated January 2019, so the dated attestation reads as old. The site's navigation links a Trust Center described as security and compliance information, and the cited record does not enumerate the current attestations it lists. A vulnerability disclosure policy does not appear among the fetched pages, a gap that stands out for a vendor whose collectors observe customers' API traffic.

The customer roster and the compliance positioning stand in for the attestations the fetched record can inspect. Banks and a credit union among named customers suggest passed procurement diligence, the posture product sells audit readiness, and a dedicated EU AI Act compliance page targets the regulation arriving around enterprise AI. Salt Labs gives the company an outward research voice that security buyers will read as maturity, though research output is not a compliance posture.

Salt positions as a platform, and the brand now says so explicitly. The Agentic Security Platform name spans discovery, posture, and runtime on one graph, Salt Code extends the same platform into developer tooling, and the integration surface across gateways, WAFs, clouds, EDR, and code repositories is how the graph gets its data. The public positioning presents these modules as operating around the same agent, MCP, and API graph.

Distribution assets exist that a copycat could not assemble by writing software. The partner network pairs cloud and security-platform integrations with channel partners, the CrowdStrike alliance includes a strategic investment and joint integrations, and KuppingerCole's leadership placement is analyst standing that takes years to earn. Buyers encounter Salt through clouds, channel partners, and a major EDR vendor's ecosystem rather than only through its own sales force. The moat holds better against startups than against edge platforms that already sit in the traffic flow.

Third parties building on Salt are not visible. The ecosystem runs inward, with partners feeding the graph, rather than outward through public APIs and an app marketplace, so network-effect platform economics remain ambition rather than evidence.

Co-founders Roey Eliyahu and Michael Nicosia still run the company they started, as CEO and COO. The about page describes founding by cybersecurity entrepreneurs in 2018, and the published bench adds a chief marketing officer, chief financial officer, and executive VP of technologies, the organizational shape of a company past its founder-only phase.

Salt Labs is the team's public technical face, and its output is now corroborated outside the company. The Hacker News covered its Expo OAuth disclosure, which NVD rated critical as CVE-2023-28131 at 9.6, and TechTarget covered its ChatGPT plugin research, a sustained and independently documented stream of in-domain API vulnerability work that the recurring State of API Security report reinforces. For buyer-facing credibility, the research organization does the work that individual founder pedigree does at younger companies.

That independently documented research record is the credibility signal the score depends on, stronger than founder pedigree alone.