AI Agent Index
ByHeather MacAvelia·Independently reviewed·Published Oct 12, 2026·Updated Oct 12, 2026
Independently verified against live vendor data on Oct 12, 2026.

Open-source platform for building and running teams of AI agents in regulated EU enterprises. Self-host free under AGPLv3, or buy Exelab's custom-priced managed plans.

How we scored it

Autonomy

4/5

Integrations

5/5

Pricing clarity

4/5

Evidence

Be an early reviewer →

Setup

5/5

The facts

Pricing

Free

freemium ↗

GitHub

—

Stars

G2

—

Rating

MCP

Client

Connects out

Reviews

Be the first →

Polyant is a platform for building and operating teams of specialized AI agents inside an enterprise, built by Exelab, a systems integrator headquartered in Rome. Each agent has a defined role, tools, memory, rules and channels, configured in a web admin panel or in TypeScript. Agents can delegate work to one another, run on schedules, or wake on events through a proactive Room mode. Exelab announced Polyant on May 26, 2026, at its Agent ORG event in Milan, and the self-hosted runtime, admin panel, long-term memory, channels and audit logging are published on GitHub under the AGPLv3 license. On the open-source build, agents talk to people over Slack, Telegram, WhatsApp through Twilio and an OpenAI-compatible HTTP API, and they can be published to outside agent-to-agent (A2A) clients. Managed services add email, HubSpot Conversations, synchronous REST and inbound phone calls through Twilio. Agents can call tools from external MCP servers, a HubSpot plugin covers contacts, deals, tickets and other CRM records, and runs can be traced in LangSmith. The model provider is set per agent: Anthropic, OpenAI, or open-weight models through Nebius and AWS Bedrock. Polyant runs as a Docker container on the customer's own systems, in the customer's AWS, Azure or GCP account, or on Exelab infrastructure in an EU region. Every runtime decision is written to PostgreSQL tables in the customer's database, where compliance teams can query it with a standard SQL client. Polyant can be adopted three ways, and they combine. The self-hosted software is free under AGPLv3, with no license fee and no time limit, and the customer pays only for the infrastructure it runs on. Exelab's managed services come in three plans, each priced per case. Core covers 1 to 3 agents on one business function, with a 99.9% uptime SLA, 90-day audit retention and Exelab infrastructure in an EU region. Advanced covers 1 to 10 agents on two or three functions in a regulated context, with a four-hour response time in business hours and 12-month audit retention. It adds a registry for DORA, IVASS and AGENAS audits and the ten runtime governance controls. Ultra covers 1 to 20 or more agents across divisions, with deployment in the customer's cloud account, on-premises or multi-region, a 24/7 watch and enterprise SSO over SAML or OIDC. Exelab professional services build the agents, priced at the end of a discovery phase. The self-hosted build includes the audit log, while the governance controls, compliance reporting and a contractual SLA come with the managed plans. Self-hosted support runs through the community unless Exelab support is added. Model costs are not included on any plan: the customer pays Anthropic, OpenAI, AWS Bedrock or Nebius directly. With Exelab services, the vendor quotes 4 to 8 weeks to a first agent in production and 8 to 12 weeks for a coordinated squad of agents. Polyant suits enterprises in regulated European industries that want teams of agents on infrastructure they choose, with the code and the audit trail open to inspection. Organizations already standardized on Microsoft 365 that want agents built inside that stack should consider Microsoft Copilot Studio. Teams whose customer data lives in Salesforce should look at Salesforce Agentforce, and technical teams that want self-hosted workflow automation across a wide library of app integrations should look at n8n. Current state Q4 2026: Polyant 1.2.0 was released on October 12, 2026. Exelab runs an ISO 27001-certified information security management system and states that its scope covers the products it builds, including Polyant. Exelab is a HubSpot Elite Solutions Partner and a Twilio Gold Partner. Polyant does not yet have a review profile on G2, Capterra or Product Hunt.

Pricing

freemium · Free

View pricing ↗

Segment

Enterprise

Setup

moderate

Verified

Oct 12, 2026

Transparency

Mostly Public

Contract

—

Data training

Not Disclosed

Human in loop

Optional

Capabilities

autonomousmulti-agent-orchestrationopen-sourcebyokschedulingknowledge-basecontext-memoryobservability

Pros & Limitations

Editorial assessment

Pros

  • ✓The runtime, admin panel, long-term memory, channels and audit logging are open source under AGPLv3 on GitHub, with no license fee for self-hosting.
  • ✓Every runtime decision is written to PostgreSQL tables in the customer's own database, where compliance teams can query it with a standard SQL client.
  • ✓Agents run on-premises, in the customer's AWS, Azure or GCP account, or on Exelab infrastructure in an EU region, with the model provider chosen per agent.

Limitations

  • –The ten runtime governance controls and the regulatory audit registry come with the Advanced and Ultra managed plans, not with the self-hosted build.
  • –Managed plans and Exelab build services are priced per case, and model usage is billed separately by the chosen provider.
  • –Email, HubSpot Conversations and phone-call channels need managed services, and enterprise SSO over SAML or OIDC is on the Ultra plan.

Technical Details

Deployment
self-hostedcloudapi
Model architectureModel-agnostic: the provider is set per agent, with Anthropic and OpenAI models plus open-weight models through Nebius and AWS Bedrock. The customer pays the model provider directly.
Avg setup timeUnder 15 minutes to install the self-hosted build and about 10 minutes to a first agent (vendor docs). With Exelab services, 4 to 8 weeks to a first agent in production (vendor-stated).
Autonomous rateNo autonomous-resolution rate is published. Agents run on schedules, on events through the proactive Room mode and through agent-to-agent delegation. On the Advanced and Ultra plans, runtime governance controls can block, warn on or log each decision.
Integrations
SlackTelegramWhatsAppTwilioHubSpotLangSmith
Security
ISO 27001GDPR

Similar agents

Rating

On Our Radar

Independently reviewed. A full rating unlocks with community reviews or public signal.

Rate this agent
How we score this →

Recognition

Listed 2026
Embed these on your site →

Industries

FinanceInsuranceEnergyManufacturingRetailTelecomLogistics

Built for

EnterpriseOpen SourceB2B

Leave a review

Never displayed publicly.

Agent Stacks

See workflow stacks that feature Polyant.

Compare

Related Content

Manage this listing

Badges, review collection, and listing updates for verified vendors.

Vendor dashboard →

Free · Every Two Weeks

AI Agent Price & Rating Tracker

Price changes, new agent launches, acquisitions, and rating updates across the AI agent market. Verified against live vendor data, not vendor marketing.

No spam. Unsubscribe anytime. We never share your email.