Quilr

• QuilrAI has disclosed one seed round, from April 2025, and no later raise appears in the reviewed sources, so a rebuilt platform, five named executives, and an enterprise sales motion have arrived with no fresh capital on the public record.
• Its partners page lists large consulting firms, resellers, and managed-security providers whose relationships no reviewed source outside QuilrAI's own site corroborates, so a buyer counting on that channel reach may find less of it than the page suggests.
• The endpoint agent depends on hooking coding assistants at the operating-system level, so a change in how those tools expose their calls could narrow what QuilrAI sees.
• Cisco AI Defense already sits beside QuilrAI in the same peer-review market, so a platform vendor with an installed base competes for the same budget line.
• The reviewed sources disclose no exclusive dataset, so QuilrAI's detection value rests on classifiers and policy logic a funded rival can rebuild.
• Public evidence of scale is 23 peer ratings and two case studies, so if those accounts do not multiply the demand signal stays thin.

QuilrAI targets a problem enterprise security teams can state plainly. One employee instruction turns into dozens of autonomous tool calls across the company's systems, and QuilrAI argues that no existing control follows the whole chain: identity sees the agent authenticate, endpoint tools see a process run, data-loss tools see a file move, and network tools see encrypted traffic to an interface.

The problem is felt, and its size is not public. Reviewers on Gartner's peer-review site describe the same gap in visibility over AI use inside their own environments, which is buyer-side confirmation that it is real for them. The reviewed sources carry no independent measurement of how large that exposure is across the market QuilrAI sells into. [s1, s13]

QuilrAI runs one platform in two halves. At design time it scans endpoints, browsers, and tool chains to find every agent, model connection, and copilot, builds a bill of materials for each, and generates a policy agent it calls a Guardian. At runtime that Guardian decides each prompt and tool call, and a red-team agent keeps attacking the policy after the agent is live.

The traffic paths are documented in detail. The model gateway runs seven stages on the way out and again on the way back, covering identity, rate limits, personal and payment data detection, custom classifiers, versioned prompts, token compression, and routing. The tool gateway multiplexes many tool servers behind one address, brokers credentials so agents never hold them, and sorts each tool into read, write, or destructive. An endpoint daemon on macOS and Windows covers coding assistants that act locally.

The depth rests on QuilrAI's own material. It publishes a developer documentation portal and a public repository holding a policy tool for coding agents, carrying 29 stars, and the reviewed sources contain no third-party technical evaluation of how well the detection performs. [s2, s3, s4, s11, s12]

Gartner's peer-review site places Quilr in an AI Security and Anomaly Detection market and lists three alternatives beside it: Grip Security at 4.9 across 21 ratings, Cisco AI Defense at 4.7 across 18, and Prompt Security at 4.3 across 13. QuilrAI's own 23 ratings at 4.7 are the most in that group, on a base small enough that a handful of reviews would reorder it.

One rival argues the boundary in public. Aurascape's comparison page says enterprises typically evaluate the two as alternatives rather than running both, which is a competitor's characterization rather than an independent test.

The position QuilrAI defends is the whole chain rather than one surface. It covers employee use of outside AI, AI a customer builds, AI embedded in bought software, and coding assistants on developer machines, so the competitive case it has to answer is a buyer picking a narrower tool for whichever of those surfaces hurts most. [s13, s19, s2]

QuilrAI runs two published case studies. Its customer page carries Sentara Healthcare, where a named vice president of security describes rolling out AI without trading speed for health-data risk, and for Spotnana, where a named head of security describes governing coding agents across a developer team. The April 2025 launch release names further early adopters, including Protiviti and UiPath, for the earlier product.

The independent layer is Gartner's peer-review site, which carries 23 ratings averaging 4.7 out of 5, with reviews dated May and July 2026 from an engineer and a chief information security officer at finance companies. The May reviewer describes the browser extension and both gateways running in their own environment. The July entry is rated 3 out of 5, carries a critical tag, and its text sits behind the site's placeholder notice.

The channel story is broad on QuilrAI's site and thin away from it. Its partners page lists large consulting firms, resellers, and managed-security providers, while outside coverage carries the RNS Technology Services reseller agreement announced in April 2026 and the Hitachi Systems partnership announced at the April 2025 launch. QuilrAI discloses no revenue figure or customer count, and what the reviewed sources carry instead is an aggregator's comparables-based estimate with its own stated range and confidence. [s5, s13, s10, s14, s16]

Founder and chief executive Vidit Arora carries the team's domain credibility. His lead seed investor describes him as a founding team member at Securonix, a security analytics company, and QuilrAI's own page calls him a founding member there who helped build enterprise security products.

QuilrAI names four more executives: a chief technology officer credited as a founding product and engineering leader at Traceable AI, a chief marketing officer, a chief revenue officer credited with scaling revenue at three prior software companies, and a chief customer officer with prior roles at Securonix and Splunk.

Those descriptions come from QuilrAI's own page. Outside sources name Tanuj Gulati, a Securonix founder, and Sam Kassoumeh, a Security Scorecard co-founder, as angel investors rather than operators, and the reviewed sources show no prior exit, publication record, or independent recognition for any of the named executives. [s8, s15, s16]

QuilrAI's trust posture is one audited certification plus a set of stated mappings. It holds SOC 2 Type II, which it says is audited annually across security, availability, and confidentiality.

The rest is readiness rather than attestation. QuilrAI states HIPAA readiness with redaction of health data and a business associate agreement on request, a mapping of its Guardian governance to the NIST AI Risk Management Framework, and payment-card detection at the gateway. Each of those is QuilrAI's own statement in the reviewed sources.

For regulated buyers the deployment options may weigh more. On-premise and virtual-private-cloud installation keeps model weights, customer data, and inference results inside the customer's own environment, enforcement decisions are forwarded to the customer's logging system, and QuilrAI says third-party penetration testing runs quarterly. [s6, s7, s3]

Add analyzed competitors to compare them side by side with Quilr.

QuilrAI sells to two buyers inside the same large organization. Its own navigation splits between security teams, who want an inventory of the AI agents nobody registered, and engineers, who want one endpoint in front of many model providers and tool servers.

The buying organizations are large. The demo form asks for a license count starting at 100, the two published case studies are a health system and a travel software company, and the reviewers who rated the product on Gartner's peer-review site place themselves at finance companies with revenue between 500 million and 10 billion dollars.

Breadth is a deliberate choice and a cost. QuilrAI covers employee use of outside AI tools, AI a company builds itself, AI embedded in bought software, and coding assistants on developer laptops, so it can enter through whichever surface a buyer worries about first, and naming its ideal customer takes four clauses rather than one.

The distinctive move is one policy agent for each protected AI agent. QuilrAI calls it a Guardian, has it read the agent's purpose and map the data it touches before writing a rule, and then enforces least privilege on every tool call that agent makes. A separate red-team agent attacks that policy before and after the agent goes live, which turns hardening into a running process rather than a setup step.

The gateways carry the traffic. One change to a base address routes model calls through identity checks, rate limits, detection of personal and payment data, prompt-injection defense, versioned prompts, token compression, and provider routing, applied on the way out and on the way back. The tool gateway multiplexes many tool servers behind one address, brokers the credentials agents would otherwise hold, and sorts each tool into read, write, or destructive.

The endpoint agent reaches what no gateway sees. It installs as a system daemon on macOS and Windows and hooks the coding assistants that read files and run shell commands locally, before any network request exists. QuilrAI publishes a developer documentation portal and a small public repository holding a policy tool for coding agents, and no independent evaluation of detection quality appears in the reviewed sources.

QuilrAI runs a demo-led enterprise motion. The site offers no self-serve signup, and its calls to action end at a booked 30-minute session, and the form asks for the buyer's role, the AI surfaces they want covered, and a license count of at least 100.

Independent evidence of who is buying runs through a review site. Gartner's peer-review listing carries 23 ratings averaging 4.7 out of 5 in an AI Security and Anomaly Detection market, with reviews dated May and July 2026 from an engineer and a chief information security officer at finance companies. The May reviewer describes running the browser extension and both gateways in their own environment. The July entry is rated 3 out of 5, carries a critical tag, and its text sits behind the site's placeholder notice.

The channel is where QuilrAI spends the rest of its public credibility, and it is the part the reviewed sources confirm least. Its partners page lists large consulting firms, resellers, and managed-security providers. Two relationships appear away from QuilrAI's own site. RNS Technology Services, a Dubai provider, said in April 2026 that it would carry the platform to Middle East customers, and the April 2025 launch release described a partnership with Hitachi Systems.

QuilrAI publishes no price, and no rate card or tier list appears on the reviewed pages, so a buyer cannot see one before speaking to sales.

The form discloses a floor rather than a rate. It asks for a license count in bands starting at 100 to 500 and states a minimum of 100 licenses, so a buyer learns that QuilrAI sizes deployments in licenses and that small ones are out of scope. Neither the basis of the bill nor what a license costs appears in the reviewed sources.

QuilrAI ships software the customer integrates and operates, in cloud-hosted, virtual-private-cloud, or on-premise form. The gateways take one change to a base address in the calling application, the endpoint agent installs as a system daemon on developer machines, and policies are authored inside the product.

Deployment is not limited to a hosted service. QuilrAI supports on-premise and virtual-private-cloud installation, and it states that model weights, customer data, and inference results stay off its own infrastructure unless a customer picks the hosted mode, where requests are processed without retention afterwards.

The operating promise is a decision inside the request rather than a report after it. QuilrAI states that every prompt and tool call passes through its decision path under 50 milliseconds and that each decision is forwarded to the customer's own logging system.

SOC 2 Type II is the one audited certification in the reviewed sources. QuilrAI states that the audit runs annually and covers security, availability, and confidentiality.

Its other compliance claims are mappings and readiness rather than audits. The security page states HIPAA readiness with built-in redaction of health data, a mapping of its governance to the NIST AI Risk Management Framework, and payment-card detection at the gateway, and the compliance page offers a business associate agreement on request. Each of those is stated by QuilrAI without an outside attestation in the reviewed sources.

For a regulated buyer the deployment model may weigh more than the badge row. On-premise and virtual-private-cloud installation keeps data inside the customer's own boundary, and QuilrAI says it runs third-party penetration testing quarterly and patches critical findings within 24 hours.

QuilrAI's position is a dependency as much as a strength. It reads browsers, endpoints, model provider interfaces, and tool servers that other companies control, and its coverage claims rest on that access.

The dependency runs deepest on the endpoint. QuilrAI hooks coding assistants at the operating-system level on macOS and Windows to see file reads, shell commands, and tool calls that never reach a network gateway. A vendor that changes how those tools expose their calls could narrow what QuilrAI sees.

The ecosystem QuilrAI describes around itself is wider than the one the reviewed sources confirm. Its partners page names model providers, cloud platforms, security tools, consulting firms, resellers, and managed-service providers, while only the RNS reseller relationship, announced in April 2026, and a Hitachi Systems partnership announced at the April 2025 launch of the earlier product appear in coverage away from QuilrAI's site, so a buyer who cares about a named partner has to confirm it with that partner.

QuilrAI's domain credibility runs through its founder. Its lead seed investor describes Vidit Arora as a founding team member at Securonix, a security analytics company, and QuilrAI's own page calls him a founding member there.

The bench below him is named. QuilrAI lists a chief technology officer it credits as a founding product and engineering leader at Traceable AI, along with a chief marketing officer, a chief revenue officer, and a chief customer officer with prior roles at Securonix and Splunk. Every one of those descriptions comes from QuilrAI's own page.

Outside sources name backers rather than operators. The launch release names Tanuj Gulati, a Securonix founder, and Sam Kassoumeh, a Security Scorecard co-founder, as angel investors. No prior exit, publication record, or independent recognition for any of the named executives appears in the reviewed sources.