AI RegTech · EU AI ActYour agents.
Your agents.
Your rules.
Your infrastructure.
Your control.
swarmd secures your entire agentic estate — agents, MCPs and LLM gateways — with risk, control and compliance in one place. Deploy on your own infrastructure, built on open-source technology. You own it. No lock-in.
- On-prem
- Built on OSS
- EU AI Act
- No lock-in
One platform
Everything you need to govern your agentic estate.
Risk, control and compliance for every agent, tool call and model request — managed end to end from a single place.
Trace every actionObserve activity in real time without interfering.
Log every eventCapture every interaction in a centralised record.
Audit every changeTrack and verify every system change and action.
Create policyDefine the rules that govern AI.
Settle payments with proofProcess agent transactions with a full audit trail.
Create legal-grade recordsKeep tamper-evident, verifiable records.
Secure every requestVerify every request at every layer.
Keep humans in controlKeep humans in control by enabling review at key decision points.
The gatekeeper
Control the exchange.
Every message between your agents, tools and models passes through swarmd's middleware. We inspect it in real time and take action — tag it, mask it, block it, or hold it for a human — so sensitive data and unsafe calls never slip through.
Middleware we run
- PII detection — Presidio23+ entity types: names, emails, phone numbers, cards, SSNs, medical IDs and more, with confidence scoring.
- Entity detection — AWS Comprehend21 entity types including bank accounts, API keys, URLs and addresses.
- Regex patternsDetect any custom data pattern you define — internal IDs, secrets, keywords.
- Rate limitingThrottle requests per window to keep runaway agents in bounds.
- Skill restrictionsAllow- or block-list exactly which tools and skills an agent may call.
Actions we take
- LogLet it through, but keep an auditable record.
- MaskRedact the sensitive data in place.
- BlockReject the exchange entirely.
- WarnAllow, but raise a warning.
- Human reviewPause and require sign-off.
EU AI Act · reviewed 18 October 2026
The EU AI Act, built into the runtime.
Most AI Act tooling holds documents about your systems. swarmd sits in the path of the agent exchanges routed through it and carries the governance record with them, so it enforces controls on that traffic and evidences the rest. Here is how the provisions that matter for AI agents are covered, and where they are not.
Enforced by swarmd10of 37 provisionsswarmd does it: in the relay, on the traffic the control covers, or by refusing the action that would break it. The limits say where it stops.Tracked and evidenced22of 37 provisionsswarmd runs the workflow, checks it is in place and keeps the evidence. The substance — a document, an assessment, a filing — is yours.Your responsibility5of 37 provisionsOutside what a runtime platform can do. We name it so nothing falls between the gaps.
Built article by articleOne matrix of forty obligations, each cited to its article of the EU AI Act as amended by the Digital Omnibus, applied to each system by its tier, role and use case.
Enforced on the wireLogging runs in the relay on every exchange routed through it. The kill switch stops a frozen agent’s conversations, tool calls and model calls, human review holds the replies you bind it to, and the Article 50 notice is added where an agent replies to a person.
Evidence you can verifyExchanges through the relay and governance decisions are hashed and sealed into signed checkpoints. Each system’s governance record exports as an evidence pack.
Honest about the edgesEach provision says whether swarmd enforces it, tracks it or leaves it to you — and where the control stops.
Chapter by chapter: the obligation, and what covers it.
- Chapters I–IIScope, AI literacy and prohibited practicesArt. 2, Art. 3(1)Art. 4Art. 51 enforced · 2 tracked
- Chapter III · Section 1Classifying high-risk systemsArt. 6(1)–(2), Annex I, Annex IIIArt. 6(3)–(4)1 enforced · 1 tracked
- Chapter III · Section 2Requirements for high-risk systemsArt. 9Art. 10Art. 11, Annex IVArt. 12Art. 13Art. 14Art. 152 enforced · 5 tracked
- Chapter III · Sections 3–5Obligations of providers and deployersArt. 16(b), (h), (l)Art. 17Art. 19, Art. 26(6)Art. 18Art. 20Art. 21, Art. 22Art. 25Art. 26(1)–(2)Art. 26(4)Art. 26(5)Art. 26(7), (9), (11)Art. 27Art. 43, Art. 47Art. 49, Annex VIII3 enforced · 9 tracked · 2 yours
- Chapter IVTransparency for every AI systemArt. 50(1), (5)Art. 50(2)Art. 50(3), (4)2 enforced · 1 tracked
- Chapter VGeneral-purpose AI models in your stackArt. 53(1)(b), Annex XIIArt. 53–551 tracked · 1 yours
- Chapters IX and XIIIMonitoring, incidents, explanations and withdrawalArt. 72Art. 73Art. 3(23), Art. 79Art. 18, 20, 26(6)Art. 86Art. 1111 enforced · 3 tracked · 2 yours
- What the law requires, how swarmd covers it, and where it stops — provision by provision.Explore the analysis →
Ownership & continuity
You own it. And you keep it current.
No lock-in isn't a slogan. It's backed by source-code escrow, on-prem licensing, ongoing updates and full jurisdictional control — so your agentic estate stays yours, and stays current, whatever happens.
No platform lock-in
Leave whenever you want. You own your deployment end to end — there is no proprietary cloud to be trapped in.
Source code in escrow
Your deployment's full source is held with an independent escrow agent and released to you if we ever can't support it.
Licensed on your prem
Run swarmd under licence entirely on your own infrastructure. Your servers, your keys, your data — we never touch it.
Updates on your terms
You keep receiving updates and improvements — and adopt them if and when you choose. Never forced, never frozen on licence day.
Within your jurisdiction
Agents, logs, decisions and data stay inside the legal jurisdiction you choose. Nothing crosses a border you didn't approve.
Built on open-source tech
swarmd runs on a foundation of proven, widely-used open-source technologies — not an obscure proprietary stack you can’t reason about.
Log every action. Enforce every policy. Audit every exchange.
Trace every action
Observe activity in real time without interfering.
Track who changed what and when. Hold every configuration and operation accountable. Capture every interaction—inputs, outputs, tool calls, data accessed, and decisions made—ready for compliance audits or incident investigations
Observe activity in real time without interfering.
Track who changed what and when. Hold every configuration and operation accountable. Capture every interaction—inputs, outputs, tool calls, data accessed, and decisions made—ready for compliance audits or incident investigations
Log every event
Capture every interaction in a centralised record.
Logging creates a permanent and immutable records and alerts of AI interactions, decisions, and data exchanges. These records can be reviewed later to understand what happened, when, and why.
Capture every interaction in a centralised record.
Logging creates a permanent and immutable records and alerts of AI interactions, decisions, and data exchanges. These records can be reviewed later to understand what happened, when, and why.
Enforce compliance in every exchange
Audit every change
Track and verify every system change and action.
Track who changed what and when. Hold every configuration and operation accountable. Capture every interaction—inputs, outputs, tool calls, data accessed, and decisions made—ready for compliance audits or incident investigations
Track and verify every system change and action.
Track who changed what and when. Hold every configuration and operation accountable. Capture every interaction—inputs, outputs, tool calls, data accessed, and decisions made—ready for compliance audits or incident investigations
Create policy
Define the rules that govern AI.
Policies specify what information can flow, what actions are allowed, and under which conditions. They form the regulatory framework the system enforces consistently and automatically.
Define the rules that govern AI.
Policies specify what information can flow, what actions are allowed, and under which conditions. They form the regulatory framework the system enforces consistently and automatically.
Build trust into every layer
Settle payments with proof
Process agent transactions with a full audit trail.
The Payments feature enables billing, automation credits, and monetized AI-agent actions while enforcing regulatory standards. Every transaction is validated, logged, and auditable, ensuring financial transparency and rule-compliant exchanges between systems and organizations.
Process agent transactions with a full audit trail.
The Payments feature enables billing, automation credits, and monetized AI-agent actions while enforcing regulatory standards. Every transaction is validated, logged, and auditable, ensuring financial transparency and rule-compliant exchanges between systems and organizations.
Create legal-grade records
Keep tamper-evident, verifiable records.
swarmd writes the agent actions it relays to an append-only audit log.
Each event is hashed with SHA-256, and every five minutes the hashes are sealed into a chained,
Ed25519-signed checkpoint, which you can also have timestamped by a third-party RFC 3161 authority.
Keep tamper-evident, verifiable records.
swarmd writes the agent actions it relays to an append-only audit log.
Each event is hashed with SHA-256, and every five minutes the hashes are sealed into a chained,
Ed25519-signed checkpoint, which you can also have timestamped by a third-party RFC 3161 authority.
Secure every request
Verify every request at every layer.
Zero-trust architecture verifies every request—internal or external—and grants nothing by default. The platform authenticates and authorises every agent, service, and API call before allowing any data to move.
Verify every request at every layer.
Zero-trust architecture verifies every request—internal or external—and grants nothing by default. The platform authenticates and authorises every agent, service, and API call before allowing any data to move.
Keep humans in control
Keep humans in control by enabling review at key decision points.
AI agents can act independently, but critical decisions still require human authority. swarmd.ai enables structured intervention points so teams can review, approve, or override actions when needed—keeping automation efficient while responsibility remains clear.
Keep humans in control by enabling review at key decision points.
AI agents can act independently, but critical decisions still require human authority. swarmd.ai enables structured intervention points so teams can review, approve, or override actions when needed—keeping automation efficient while responsibility remains clear.
Obtain total oversight of your AI systems
Configure policies your wayIntegrate tools like MS Presidio, regex rules, or AWS Converse — and more to come! — to detect, redact, or validate information in real time. Fully configurable to your compliance needs.
Centralise your MCP serversThe central MCP server brings together auditing, logging, and policy enforcement, giving teams a single source of truth for regulated AI operations.
Discover approved toolsDiscover and deploy approved tools, policies, and middleware to expand functionality while staying within your regulatory framework.
Operate agents at scaleManage, monitor, and regulate agent behavior across systems, ensuring every action is visible, traceable, and compliant by design.
Deploy anywhereDesigned to operate seamlessly in cloud or on-premises setups, without sacrificing control, performance, or security.
Report on what mattersGenerate structured reports that surface trends, violations, and system behavior—ready for internal review or external audits.
Integrate via an API-first platformA fully API-driven, microservice-based platform designed for scale, resilience, and easy integration.
Comply with the EU AI ActDesigned to support compliance with the EU AI Act by enforcing policies, tracking agent activity, and maintaining auditable records across AI exchanges.
Deploy agents on your prem — free for 3 months
Book a demo, or start a free on-prem trial: unlimited, on your own infrastructure. We’ll send the setup steps and a licence key.
Control the exchange
Integrate once. Govern every agent you route through swarmd. Keep the evidence from day one.
Scale with your team
Pick a plan built for growing teams and regulated environments