How much does Enclave cost?

No public pricing; recently out of stealth and enterprise sales led. Access is via the vendor directly.

What are the best alternatives to Enclave?

Based on documented capability coverage across the Agentic Index 14 point taxonomy, the closest alternatives to Enclave among security and SOC agents are Crogl, Ocean, Quantro Security, Airrived. Each is graded on the same public evidence, and no vendor pays for placement.

Can you bring your own model or route between providers with Enclave?

Full under the Agentic Index verification standard. Enclave publishes public evidence of bring your own model and bring your own key (BYOK), and of routing tasks between models or providers rather than being locked to one. Evidence basis: "Choose frontier models or open weights running on US inference providers. Use Enclave inference or bring your own keys." The model list names GPT-6 Astra, Luna and Sol and Claude Opus 5.5 and Claude Sonnet 5, OpenAI and Anthropic frontier. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Does Enclave have SOC 2, SSO and role based access control?

Partial under the Agentic Index verification standard. Enclave publishes some public evidence of security and identity controls, but narrower than the full set of SOC 2, single sign on (SSO) and role based access control (RBAC). Evidence basis: The site footer carries an AICPA SOC 2 badge, and Enclave lists a trust center at trust.enclave.ai. No SSO or role model is published on the site. Security material from unrelated companies named Enclave does not describe this product. Encl. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Does Enclave give you tracing and observability into agent runs?

Partial under the Agentic Index verification standard. Enclave publishes some public evidence of logging or observability, but narrower than full tracing and auditability across agent runs. Evidence basis: Enclave traces the evidence behind each finding and records verification results. After one fix was deployed, for example, it replayed the expired webhook request and recorded the HTTP 401 response. That evidence shows what the agents concl. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Back to vendors

Enclave

Q: Who is Enclave for?

engineering teams, security teams, AppSec, startups

Visit site

Security / SOC agentindependentVerified 2026-09-28

Entry priceContact sales; no public pricingFull pricing detail

Autonomous security agent team that scans code, pen tests live applications, reviews configurations, monitors CVEs, triages and prepares verified fixes, taking jobs in Slack or Teams and running on the customer's choice of frontier or open weight models.

Enclave sells what it calls an autonomous security team for the enterprise: a team of security agents that finishes scoped security jobs end to end across a customer's code, infrastructure and live applications. Its specialist agents cover code scanning, penetration testing, configuration review, network inventory, CVE monitoring, intrusion detection, triage and remediation; triage reproduces findings, and remediation prepares a fix, sends it for review and retests after deployment, for example by replaying an expired webhook request and recording the rejected response.

Teams assign scoped jobs and review evidence in Slack or Microsoft Teams, and the agents work from GitHub pull requests and commits, AWS Security Hub and scanners such as Wiz, Snyk, Semgrep, Dependabot and Socket. Customers choose frontier models or open weight models on US inference providers, through Enclave's inference or their own keys, and Enclave's forward deployed engineers build custom agents for a customer's environment and stay until the team runs them on its own.

Enclave came out of stealth in March 2026 with a six million dollar seed led by 8VC; its founders, CEO Tal Hoffman, CTO Dvir Segev and CPO Yanir Tsarimi, worked together in application security, and Hoffman and Tsarimi served in Israel's Unit 8200. Its trust center renders only in a browser, and the site documents no attestation, deployment option, API or run trace of the agents. It fits engineering and security teams that want an agent team to find, verify and fix exploitable issues across code, cloud and running applications with their own choice of model; a buyer who needs documented attestations or a self hosted option before a pilot will need to ask for them.

Vendor details

Canonical URL

https://enclave.ai

Funding status

Launched from stealth on 26 March 2026 with six million dollars backed by 8VC and angels including Patrick Collison, Aaron Levie, Diane Greene, Matt Huang, Jeremy Stoppelman and Marc Benioff (Enclave's launch post). Founded by Tal Hoffman (CEO), Dvir Segev (CTO) and Yanir Tsarimi (CPO), who built application security tooling together at Enso Security; Hoffman and Tsarimi served in Israel's Unit 8200.

Company status

independent

Use cases & customers

Primary use cases

AI generated code security reviewvulnerability discovery across code and cloudexploitability verificationattack surface inventory and remediation context

Target customers

engineering teamssecurity teamsAppSecstartups

Deployment options

SaaS

Integrations

Slack and Microsoft Teams for jobs and evidence; GitHub for pull request review, commit analysis and fixes; AWS Security Hub; scanners and security tools including Wiz, Snyk, Semgrep, Dependabot and Socket. No API, SDK or MCP server is documented.

In practice

A pull request changes Terraform in a way that would let every AWS account restore a production snapshot. Enclave's code review flags it before merge and lays out the evidence in Slack.

Wiz, Snyk and Dependabot raise more findings than the team can check. Enclave's Triage agent reproduces each one, and Remediation prepares a fix as a pull request and retests it after deployment.

A small security team wants scoped work done overnight. It assigns a job in Slack or Teams, and the agents keep working after people sign off for the day.

Sources & related URLs

Related / legacy domains

https://enclave.ai/blog/securitys-blindspot

Research sources

https://enclave.ai/ https://openclawai.io/blog/enclave-6m-seed-ai-code-security-flaws-8vc-2026 https://benzatine.com/news-room/startup-enclave-secures-6-million-to-uncover-ai-code-vulnerabilities

Agentic Index coverage score

7.0 / 14 capabilities · 50%

Integrations & Tool Calling	Full
Enclave takes jobs and reports evidence in Slack or Microsoft Teams, works from GitHub (pull request review and commit analysis), AWS Security Hub and scanners such as Wiz, Snyk, Semgrep, Dependabot and Socket, and prepares fixes as pull requests and tickets. In one example it flags a Terraform change in a pull request that would let every AWS account restore a production snapshot. SourceEnclave, enclave.airead 2026-10-06
Workflow Orchestration	Full
Enclave describes "an autonomous team of security agents that finishes scoped security jobs end to end". Specialist agents for code, infrastructure and live application testing run in parallel, Triage verifies and reproduces findings, and Remediation prepares fixes and retests them after deployment. The agents keep working when people sign off. SourceEnclave, enclave.airead 2026-10-06
Knowledge Grounding & RAG	Partial
The agents work from the customer's code, infrastructure, scanners and security tools, and a Network Inventory agent maps the attack surface. Enclave describes no maintained knowledge store the agents retrieve from. SourceEnclave, enclave.airead 2026-10-06
Human Oversight & Guardrails	Partial
Jobs are scoped by the customer in Slack or Teams, evidence is reviewed there, and "a draft fix goes to review" as a pull request, so the team decides what merges through its own code host. Enclave documents no approval step inside the product before an agent acts. SourceEnclave, enclave.airead 2026-10-06
Security, Identity & Governance	Partial
The site footer carries an AICPA SOC 2 badge, and Enclave lists a trust center at trust.enclave.ai. No SSO or role model is published on the site. Security material from unrelated companies named Enclave does not describe this product. SourceEnclave, enclave.ai and trust.enclave.airead 2026-10-06
Observability & Auditability	Partial
Enclave traces the evidence behind each finding and records verification results. After one fix was deployed, for example, it replayed the expired webhook request and recorded the HTTP 401 response. That evidence shows what the agents concluded, but no run trace of the agents' own steps and tool calls is documented. SourceEnclave, enclave.airead 2026-10-06
Memory & State Persistence	Not documented
Enclave describes no memory its agents keep from one job to the next, what it would hold or how long it would last. Each job starts from the scope a person sets in Slack or Teams. SourceEnclave, enclave.airead 2026-10-06
Deployment & Data Residency	Not documented
Open weight models run on US inference providers, and customers can use Enclave's own inference or bring their own keys. Beyond where inference runs, Enclave publishes no hosting region or customer environment for the platform itself, and it lists a trust center at trust.enclave.ai. SourceEnclave, enclave.ai; trust.enclave.airead 2026-10-06
Prebuilt Agents / Templates / Packs	Full
Eight named specialist agents have stated jobs, covering Code Scanning, Pen Testing, Config Review, Network Inventory, CVE Monitoring, Intrusion Detection, Triage and Remediation. Forward deployed engineers also "build custom agents for your environment". SourceEnclave, enclave.airead 2026-10-06
Triggers & Channel Coverage	Full
Pull request review and commit analysis on GitHub, CVE Monitoring and continuous penetration testing wake the agents on repository events and new disclosures, besides jobs assigned in Slack or Teams; the event wiring itself is not documented. SourceEnclave, enclave.airead 2026-10-06
Model Flexibility & Routing	Full
"Choose frontier models or open weights running on US inference providers. Use Enclave inference or bring your own keys." The model list names GPT-6 Astra, Luna and Sol and Claude Opus 5.5 and Claude Sonnet 5, OpenAI and Anthropic frontier models, beside open weight options, so customers choose among providers and can bring their own keys. SourceEnclave, enclave.airead 2026-10-06
APIs / SDKs / MCP Extensibility	Not documented
No API, SDK, MCP server or developer documentation is published. Custom agents are built by Enclave's forward deployed engineers, who embed with the customer's team and stay until the team runs the agents on its own. SourceEnclave, enclave.airead 2026-10-06
Testing, Debugging & Optimization	Not documented
Triage reproduction and retests after deployment check the customer's systems and fixes. Enclave's Hacking Arena posts benchmark frontier models on hacking tasks, reporting for example that GPT-6 Sol went four for four without faking a single flag. Customers get no harness, scored test cases or quality gate for testing Enclave's agents themselves. SourceEnclave, enclave.airead 2026-10-06
Browser / Computer-use	Not documented
Live application tests replay requests and record responses; no agent operating a browser or desktop is documented. SourceEnclave, enclave.airead 2026-10-06

The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded

Pricing

Contact sales; no public pricing

not disclosed

What is public

No prices. The site states the model options (frontier or open weights, Enclave inference or your own keys) and that forward deployed engineers build custom agents.

Cost watchouts

Inference, not stated by the vendor: cost may scale with the code, cloud and application surface under continuous testing; bringing your own keys moves model costs onto your provider bill.

Variable cost rationale

Inference, not stated by the vendor: likely scales with the attack surface in scope; the BYOK option means model usage can be billed by the customer's own provider rather than by Enclave.

Sales call required

Yes, required for paid access

Free / trial

No public free tier; access via the vendor

Lowest paid plan

Not public

Key ambiguities

No public pricing and no disclosed billing model for this recently unstealthed company.

Agentic Index verified 2026-09-28

Official pricing page

Related vendors

Alternatives to Enclave

The closest documented capability profiles to Enclave among security and SOC agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.

Crogl9.0 / 14Adds documented Security, Identity & Governance and Deployment & Data Residency
Ocean5.0 / 14Adds documented APIs, SDKs & MCP Extensibility
Quantro Security7.0 / 14Adds documented Security, Identity & Governance
Airrived8.5 / 14Adds documented Security, Identity & Governance and APIs, SDKs & MCP Extensibility, among others
Equixly4.5 / 14Adds documented Security, Identity & Governance
Parameter7.5 / 14Adds documented Security, Identity & Governance and Memory & State Persistence

Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded

Compare Enclave with other vendors →Compare all security and SOC agents →See where Enclave ranks →

Data confidence: medium

Head to head

Enclave vs Vulnetic