One price,one contract,one interlocutor.
Three components, one interlocutor. Exelab builds the agents, governs production with a contractual SLA, or supports the customer's internal self-host team. Pricing is defined on a case-by-case basis.
Talk to the team →One product, three ways to adopt it.
The entire product core is open source under the AGPLv3 licence. The code is downloaded from GitHub, installed on the customer's cluster, and only the infrastructure it runs on is paid for. No licence, no subscription, no expiring free trial. Suited to customers with an internal technical team and enterprise-grade operational experience who want to keep everything within their own perimeter and take responsibility for managing infrastructure, patching, on-call, and audit prep.
Tailored build
Exelab builds the agents starting from the company's needs, following a path of discovery, design, development, and integration with the customer's systems. Output: an agent squad ready in production, with a documented runbook and hand-off to the operations team. Typically combined with managed services, which govern continuity in production. Project pricing is defined at the end of discovery; first agent in production in 4–8 weeks.
Open professional services →Governed production
Exelab managed services keep the Polyant software in production with a contractual SLA. End-to-end infrastructure management, product maintenance and updates, operational support with response time per profile. Three managed profiles (Core, Advanced, Ultra) tuned to the customer's operational scale, criticality, and compliance.
See the three managed profiles →Self-host software
Polyant is available on GitHub and installs on the customer's infrastructure under the AGPLv3 open source licence. The code stays inside the customer's perimeter; the company's IT team takes on operations. The customer covers the infrastructure costs. Suited to customers with an internal technical team who want to keep everything within their own perimeter and take patching, on-call, and audit prep in-house.
Open quickstart →Core, Advanced, Ultra. Three profiles that differ by operational scale, criticality, and compliance.
A managed profile is a configuration aligned with the customer's specific needs. Operational scale, criticality, and compliance move together. The three profiles cover the operational continuity of the software in governed production, with contractual SLAs. The build of the agents is a separate component: it is typically combined with managed services, but managed can also run stand-alone on top of agents built by the customer's team.
Core
1-3 agents in production on 1 business function
- 99.9% uptime SLA
- Audit retention 90 days (configurable)
- Escalation during European business hours
- Exelab infrastructure in EU region
- Software maintenance and patching
Advanced
1-10 agents on 2-3 business functions in a regulated context
- 99.9% uptime SLA
- 4-hour response time during business hours (9-18 CET)
- Audit retention 12 months (configurable)
- Registry for European regulatory audits (DORA, IVASS, AGENAS)
- Ten built-in runtime controls
- Versioned policy management per instance
- Out-of-hours on-call
Ultra
1-20+ agents multi-division in a critical regulated context
- Service levels defined to the customer's needs
- Audit retention configured to the case
- On-demand reporting
- Customer's cloud account (AWS, Azure, GCP), on-premise, multi-region
- Continuous 24/7 watch
- Enterprise SSO (SAML/OIDC)
- Additional product capabilities defined to the customer's needs
The code stays inside the customer's perimeter.
The autonomous self-host option can be handled by the customer in two different ways. The first is the OSS community path: the customer takes all operations on. As an alternative, the customer can rely on Exelab's professional support: the software stays inside the customer's perimeter, but Exelab can handle installation and configuration, agent creation, or even just support, working alongside on architectural review, maintenance, and updates.
The public software under AGPLv3, downloadable from GitHub. Documentation and reproducible quickstart. Support via the community and public channels. The customer covers infrastructure, patching, on-call, audit prep, and product evolution. Suited to customers with an enterprise-grade internal technical team and a culture of fully internal operations.
Open quickstart →The open-source AGPLv3 software is installed by the customer, with an Exelab professional support channel tailored to the internal team's needs: architectural review before go-live, technical escalation on product defects, and guidance through major releases. Pricing defined on the agreed support perimeter. This does not replace Exelab managed services: it does not include infrastructure monitoring or contractual SLAs on production.
Talk to the team →From qualification to continuous run.
The Polyant adoption path describes how a customer moves from first contact to go-live, and how the agent squad remains in production in the years that follow. The build step applies when the customer chooses Exelab professional services.
Initial qualification
A 30–45 minute conversation with the Exelab team to understand the customer's situation: company profile, starting business area, regulatory constraints, and management systems involved. Qualification closes with the preliminary recommendation of components and a decision on the next step.
Discovery
When the customer chooses Exelab support, we work with the customer's team to define the first agent or the first agent squad, the specific processes it governs, the external systems to integrate, and the operational and regulatory rules to follow. Output: delivery document with scope, integrations, timing, project pricing, and suggested managed tier where applicable.
8-12 weeks for a coordinated squad of multiple agents
Build of the first agent or squad
The Exelab team builds the agents for the customer's case: configuration, integration with the customer's systems, operational validation testing, and hand-off to the operations team. Step pricing is defined in Step 2 based on the actual case. When the customer builds independently or with others, this step does not apply.
Continuous run in managed (or operations in self-host)
When the customer chooses Exelab managed services, the agent squad enters the profile selected in Step 2 (Core, Advanced, or Ultra). Exelab keeps the software infrastructure running, applies patching, responds to operational escalations, and prepares audit documentation. Operational management of the agents stays with the customer's team.
Five starting questions to locate the case.
The decision tree does not replace the conversation with the team — it prepares you for it. The five questions produce a preliminary profile recommendation.
What kind of processes do you want to automate?
How many agents are planned, and across how many business functions?
What is the operational criticality level?
Which regulators apply to the customer's context?
Who builds the agents and who runs operations?
—
—
Ten questions to settle before the conversation.
Yes. The typical combination for fast and effective adoption is professional services + managed services, where Exelab builds the agents and governs production. It is also possible to purchase managed services alone for agents the customer's team builds independently, or to work with Exelab on building the agents only, with self-host operations on the customer's side.
An agent is a coherent configuration of role, tools, memory, rules, and channels, dedicated to a specific operational case. Multiple agents working together on the same function form a squad or team. The operational scale of the managed profile is measured by the number of agents in production, the volume of activity each agent runs, the complexity of the tasks it carries out, the platforms it integrates with, and the number of end users.
The customer moves to the next profile via a contractual migration. The software stays the same; what changes is the SLA, watch window, and compliance depth. The migration is planned with Exelab based on the scale reached and the customer's new operational constraints.
Yes. Polyant is available under the AGPLv3 open source licence in autonomous self-host: download from GitHub and install wherever the customer prefers. It is not a time-limited trial; it is open-source software under a copyleft licence. Exelab professional services and managed services are separate paid services, tuned to the customer's case.
The AGPLv3 licence is explicit on copyleft: anyone who modifies the software and distributes it or offers it as a network service is required to make the source code of the modifications available. For a customer that uses Polyant internally as a work tool (the most common use case), the network clause (AGPLv3 section 13) does not apply. For scenarios involving redistribution or offering the software to third parties, the customer's legal team should assess compatibility with the commercial arrangement. Exelab provides supporting legal documentation during qualification.
In the region chosen by the customer. Deployment options vary by managed profile: Core runs on Exelab infrastructure in an EU region; Advanced runs on Exelab infrastructure in a customer-chosen region; Ultra adds deployment in the customer's cloud account (AWS, Azure, GCP), on-premise, and multi-region data residency. LLM models accessed via AWS Bedrock stay in the customer's cloud account. Contractual data residency is declared in the Exelab contract.
Yes. Two options. The reproducible quickstart at docs.polyant.ai/getting-started/quickstart lets the customer's technical team install Polyant in self-host and build a first test agent. Alternatively, when interest leans toward managed solutions and professional services, a guided demo can be requested from the Exelab team.
Yes. Moving between managed profiles (Core → Advanced → Ultra or vice versa) happens with a contractual migration: the software stays the same, the service perimeter changes. Moving from autonomous self-host to managed is common when the customer concludes that the internal management effort is not justified by the cost of outsourcing the service. The reverse move, from managed to autonomous self-host, is also possible: the customer retains the configuration, the infrastructure, and the agents, and takes operations in-house.
No. LLM model costs remain with the customer, who pays the chosen provider directly: Anthropic, OpenAI, Mistral, AWS Bedrock with access to Amazon Nova, Llama, Qwen, DeepSeek, Mistral, and so on, as well as open-source models hosted internally. Exelab's offering covers the software and, where applicable, managed hosting, support, and maintenance. Per-agent consumption can be monitored through Polyant's interfaces.
For legal: the AGPLv3 licence of the software and the Exelab contracts for professional services and managed services, tuned to data residency and the regulators covered. For procurement: Exelab's legal seat (Rome, Italy), the company's ISO 27001 certification, the active vendor qualification in financial and insurance procurement, the institutional partnerships (HubSpot Elite Solutions Partner, Twilio Gold Partner, AWS Partner).
Talk to a Polyant architect.
A conversation with the Exelab team to understand where the customer's case sits and which combination of components (professional services, managed services, autonomous self-host) fits best. The first agent in production, for typical cases with professional services, takes 4-8 weeks of build after discovery.