Rein Security
by Rein Security, Inc.
Runtime security for enterprise AI agents, deployed as a sidecar inside your own cloud
Rein Security is a runtime security platform for the AI agents enterprises build in-house. It runs as a sidecar next to production applications, records every prompt, tool call, API request and database query an agent makes, and blocks actions that deviate from a learned baseline. It is aimed at CISOs and AppSec teams in regulated industries.
Rein Security is a New York and Tel Aviv company founded in 2024 by CEO Matan Bar-Efrat and CTO Netanel Rubin. It came out of stealth on January 28, 2026 with an $8 million seed round led by Glilot Capital, pitched at first as production application security: a patent-pending, agentless sidecar that gives code-level visibility into running applications without proxies, sampling or eBPF, with a claimed performance impact under one millisecond. In June 2026 the company repositioned the same technology as a security platform for enterprise AI agents, and on October 8, 2026 it announced a $25 million Series A co-led by Glilot Capital and Sienna Venture Capital, bringing total funding to $35 million. The sidecar sits at the runtime where an agent acts and captures the prompt it received, the APIs and databases it touched, the tools it called, and lower-level actions such as network connections, binary executions and file system access. A Protect Mode learns a behavioural baseline and blocks deviations in real time without taking the agent down, and a Business Impact Analysis layer maps each action to its operational consequence for audit work under frameworks such as the EU AI Act. The same deployment also covers SCA, SAST and real-time API security, so a team can buy it for AppSec alone. Execution data stays inside the customer's environment. Rein's site lists Dun & Bradstreet, Lemonade, H&R Block, HiBob and Flex as customers, and the company reports 8x revenue growth and 5x customer growth since its January launch, with 31 employees. Pricing is not published.
A CISO or AppSec lead at a regulated enterprise (financial services, insurance, healthcare) whose engineering teams are shipping in-house agents into systems that move money or customer data.
A per-action audit trail of what each production agent actually did, plus real-time blocking of actions outside its learned behaviour, without routing data through a third-party gateway.
At a Glance
- Category
- Governance & Security
- Pricing
- Contact for pricing
- Target Market
- CISOs, AppSec Teams, CTOs, Security Engineers
- Deployment
- Self-hosted
- Founded
- 2024
- Headquarters
- New York, United States (also Tel Aviv, Israel)
- Team Size
- 11-50
Key Features
- ✓Runtime sidecar
Deploys once beside production applications and auto-discovers services, with no proxy, sampling or eBPF and a claimed sub-millisecond overhead.
- ✓Agent action visibility
Logs every prompt, tool invocation, API call, database query, network connection and file access an agent makes, with who triggered it.
- ✓Protect Mode
Learns a baseline of each agent's normal behaviour and blocks deviating actions in real time without shutting the agent down.
- ✓Business Impact Analysis
Translates each agent action into its operational consequence so auditors and risk teams can see what an incident actually affected.
- ✓Built-in detectors
Flags prompt injection, hallucination and misconfigured tools; Rein cites a blocked prompt injection hidden in a PDF at a customer.
- ✓Production AppSec
Adds SCA reachability, SAST and real-time API security in the same deployment, showing which vulnerable libraries and functions actually run.
- ✓Data sovereignty
Agent execution data stays inside the customer's own cloud environment, which matters for banks, insurers and healthcare providers.
Use Cases
- •Insurance claims and pricing agents
An insurer records the data and reasoning behind each automated pricing or claims decision, giving regulators a per-action audit trail.
- •Prompt injection containment
A malicious instruction hidden in an uploaded document tries to redirect an agent, and the runtime layer blocks the resulting out-of-baseline action.
- •Production agent inventory
A security team discovers which agents, MCP servers, libraries and APIs are live in production after a single sidecar deployment.
- •Vulnerability triage by reachability
AppSec narrows a long CVE backlog to the vulnerable functions that actually execute in production, cutting false-positive work for developers.
- •Retail shopping agent hardening
Rein's Agent Breakers research showed a compromise of a top-five US retailer's shopping agent, the class of attack its runtime controls target.
Ideal For
Best For
- ✓Securing custom-built agents that touch payments, claims, underwriting or customer records
- ✓Regulated teams that cannot send agent execution data to a vendor-hosted proxy or gateway
- ✓Producing audit evidence of agent actions for EU AI Act, SOX or HIPAA reviews
- ✓AppSec teams that want SCA reachability and API security from the same runtime deployment
- ✓Inventorying the agents, MCP servers, libraries and APIs actually running in production
Not Ideal For
- ✗Teams whose main risk is coding agents on developer laptops or in CI, which a sidecar beside production applications does not see (a gap Arcjet's comparison page points out and Rein does not document)
- ✗Buyers who want explicit, reviewable allow/deny rules from day one, since Protect Mode relies on a learned baseline that needs a learning period and tuning
- ✗Organisations that need published pricing or self-serve onboarding; everything goes through a sales demo
- ✗Companies mainly worried about employees using third-party SaaS AI tools, which Rein's own founders describe as a different problem from the in-house agents it targets
Deployment
Market Analysis
Pros
- ✓Named production customers in regulated sectors (Lemonade, Dun & Bradstreet, H&R Block, HiBob, Flex)
- ✓No gateway or proxy in the data path, and a claimed sub-millisecond overhead
- ✓Covers both agent security and conventional AppSec, so it can be justified on either budget
- ✓Lemonade's CISO credits it with reducing false alerts
Cons
- ✗Young product: out of stealth only in January 2026 and repositioned around agents in June 2026
- ✗Protect Mode depends on a learned baseline that needs a learning period and tuning before it blocks reliably
- ✗No documented coverage of coding agents on laptops or in CI pipelines
- ✗Pricing, supported languages and policy workflows are not published, and there are no G2, Capterra or Hacker News reviews yet
Pricing
Enterprise
Contact for pricing
- ✓Runtime sidecar deployment
- ✓Agent visibility and Protect Mode
- ✓SCA, SAST and API security
- ✓Book-a-demo onboarding
Rein publishes no list prices; the site offers only a demo booking. Arcjet's September 25, 2026 comparison also notes pricing is undisclosed, and a Preqin profile describes the revenue model as mainly annual or monthly subscriptions.
Security & Compliance
Connect
Sources
This page was written from 7 sources, 6 on domains other than reinsec.io.
- 1.reinsec.io — reinsec.iovendor
- 2.calcalistech.com — rkf52cnimg
- 3.cryptobriefing.com — rein security raises 25m ai agent security
- 4.helpnetsecurity.com — helpnetsecurity.com
- 5.arcjet.com — rein vs arcjet
- 6.tfir.io — ai agent production security rein
- 7.enterprisedna.co — rein enterprise ai agent security platform june 2026
Stay Ahead of the Curve
Weekly enterprise AI insights for technology leaders. No spam, no vendor pitches—unsubscribe anytime.
SubscribeRelated Products
SAS AI Navigator
SaaS AI governance inventory for models, agents and use cases, sold through Microsoft Marketplace
Armadin
Autonomous offensive security: AI agent swarms that chain your weak spots into proven attack paths before an attacker does
DeepKeep
AI security platform covering red teaming, an AI firewall, agent attack-surface scanning and runtime controls for coding agents
WitnessAI
Network-layer AI security and governance for employee AI use, models, apps and agents