Riptides gives workloads, services and AI agents their own runtime identities and controls what they can connect to. It is built for platform and security teams that want to remove static secrets from their infrastructure and trace every machine and agent action back to its source.
The difference is where the work happens. Riptides runs at the Linux kernel level, so applications and agent frameworks need no SDKs, sidecars, proxies or code changes to get an identity.
At A Glance
| Vendor | Riptides |
|---|---|
| Category | Workload and AI agent identity, authorisation |
| Identities covered | Workloads, services, AI agents, secrets and API keys |
| Deployment | SaaS control plane, or self-hosted |
| Agentic AI | Agent identity, attribution, secretless AI, tool call security |
| Identity standard | SPIFFE |
| Market classification | Authorisation |
What Riptides Does
Machine identities are usually handled inside the application, with keys in configuration files and libraries added to each service. Every team does it differently, and secrets leak. Riptides moves that work below the application: a kernel module on each host attests which process is running and handles its authentication on the way out, so nothing has to be imported, annotated or rewritten. Identities follow the SPIFFE standard, which makes them usable by any system that speaks the specification. New deployments start in permissive mode, so teams can watch traffic before they enforce policy.
Products
Agentic AI
Each agent request carries two facts: which agent process is running and whose authority it is acting on.
| Agent Identity | Every agent process is identified automatically, even two agents in the same pod, with a SPIFFE identity that records the delegating user. |
|---|---|
| Attribution | Every connection is logged with the agent identity, destination, policy decision and delegating user, so AI actions trace back to a person. |
| Secretless AI | API keys for AI providers are injected by the kernel module before the request leaves the host and never enter the agent’s process memory, so a compromised agent has no key to leak. |
| Tool Call Security | Inspects and authorises each tool call in real time, with egress policies per agent. |
Workload IAM
Workload IAM is for teams that want services to authenticate to each other and to cloud providers without long-lived secrets. Mutual TLS between services is handled by the kernel module, workloads use federated identity with the major cloud providers, and short-lived, rotated credentials are delivered directly to the kernel rather than stored in configuration.
Kubernetes
For Kubernetes clusters, Riptides offers mTLS without a service mesh, so service traffic is encrypted and authenticated without sidecars. It issues standard SPIFFE SVIDs without running SPIRE and builds machine identity from Kubernetes metadata.
How It Works
Enforcement always runs on the customer’s own nodes. The control plane, where users manage workload identities, services and credential sources through the CLI and Kubernetes CRDs, is either hosted by Riptides or self-hosted to keep identities, policies and event data inside the customer’s boundary.
A daemon on each node watches the control plane for policy changes, loads them into the kernel module and handles certificate signing and credential delivery. It runs as a Kubernetes DaemonSet or as a systemd service on VMs and bare metal. The kernel module intercepts TCP connections at the socket level to perform the mTLS handshake and inject credentials. It attests identity from process and environment metadata such as command name, Kubernetes labels, namespace, cgroup and cloud instance metadata, so a compromised process cannot spoof its identity.
Identities are issued locally with no control plane on the critical path. SVIDs are short-lived and rotate automatically, so revoking an agent means stopping new issuance, and when credentials rotate in the source store the kernel module picks up the new values without a restart.
Identities And Systems Covered
| Workloads and services | SPIFFE identities for processes on Kubernetes, VMs and bare metal. |
|---|---|
| AI agents | Per-process agent identity linked to the delegating human user. |
| Secrets and API keys | Short-lived credentials injected at runtime for AI providers and cloud services. |
| Cloud identity | AWS IAM Roles Anywhere, AWS SigV4 signing, Google Cloud Workload Identity Federation and federated identity for Azure. |
Integrations
| Area | Integrations |
|---|---|
| AI providers | OpenAI, Anthropic, xAI Grok, AWS Bedrock, Google Vertex AI |
| Agent frameworks | LangChain, CrewAI, AutoGen, OpenAI Agents SDK, MCP-based agents |
| Secret stores | HashiCorp Vault, AWS Secrets Manager, Kubernetes Secrets, GCP Secret Manager, Azure Key Vault |
Company Background
| Founded | 2025. |
|---|---|
| Headquarters | Hungary. |
| Founders | Zsolt Varga, Nándor István Krácser, Márton Sereg and János Mátyás. |
| Funding | A $3.3 million pre-seed round led by PortfoLion Capital Partners, with KAYA VC, the founders’ fund Cloudbreak and angel investors. |
Note: This overview is compiled from Riptides’s publicly available information as of 4 November 2026. It is provided for information only. NHI Mgmt Group does not endorse any vendor, and nothing on this page represents the opinions of NHI Mgmt Group.