Getting started

Glossary of Unless terms

Unless uses a specific vocabulary across its platform, documentation, and dashboard. Understanding these terms helps you navigate the system, configure your AI agents, and communicate effectively with your team. This glossary covers the core concepts you will encounter as an AI Manager, from the two agents and the four lifecycle moments to the engine’s three senses and the billing model.

The agent side

The Customer Agent is the customer-facing AI agent that runs on your marketing site, product, portal, help center, and helpdesk channels. One agent serves all four moments of the customer lifecycle. The Team Assistant is the internal AI assistant that lives inside the helpdesk your team already uses. It drafts replies, surfaces context, and runs Procedures on behalf of your agents.

The four moments are the customer-lifecycle stages where the agent works: Acquisition, Retention, Sales expansion, and Support. Each moment has its own Signals, Personality, Procedures, and Notifications. On unless.com, the Sales expansion moment is sometimes labeled simply as “Expansion” - they mean the same thing.

The engine side

The Engine is the back-of-house platform behind both agents. Two structural ideas live here: the three senses and the loop.

The three senses

Living Knowledge is what the agent knows. It is built from your Sources, structured into the Content library, and published as the Help Center.
Living Memory is who the agent is talking to. It holds per-customer preferences, history, sentiment, and goals, scoped per agent and kept private.
Living Context is where the agent is acting. It connects to your CRM, billing, ERP, helpdesk, and custom APIs so the agent acts inside your systems of record.

The loop

The Engine cycles through four phases: Train, Test, Deploy, and Analyze. Each pass sharpens the agent. Train builds the knowledge and procedures, Test validates before customers see anything, Deploy puts the agent to work, and Analyze measures what happened and feeds the next cycle.

Trust and governance

Trust is the governance layer that sits alongside the loop, built into the platform at the architecture level. It has four pillars: Transparency, Privacy, Security, and Accountability.

The agent boundary is a key safety concept. The agent does real work across the lifecycle, but never takes an irreversible action without a human approving the boundary first. This gives customers help while giving your auditor the receipts.

The model constellation is the set of EU-hosted foundation models the Engine runs on, from providers like Anthropic, Mistral, OpenAI, Google, Cohere, and Amazon Titan. Unless picks the right model per job, balances load, and switches providers in real time if regulation or vendor terms change. Customer data is never used to train public models. You can also bring your own LLM if you prefer to run the Customer Agent on a model you already license.

Billing and outcomes

Unless uses outcome-based pricing, where you pay for results rather than seats or every interaction. An outcome is the single billing unit across both plans. It is recorded when Unless successfully does the job it was set up to do, either a resolution (it successfully answers at least one real question, validated after the interaction ends) or a procedure handoff (a configured Procedure passes the customer to a human or a workflow). You never pay for misfires.

The two subscription plans are Flex and Fixed. Flex costs €0.99 per outcome with a 100-outcome monthly minimum, 10 team seats, 5 audiences, and email support. Fixed costs €1,999 per month with a 12-month minimum, 5,000 outcomes included, unlimited seats, audiences, and standard connectors, plus an SLA, dedicated support, and a named account manager.

Dashboard and data management

The Dashboard is the interface in Unless.com where AI Managers manage the AI and its behavior. It visualizes live customer engagement metrics, AI performance data, and personalization outcomes. Key sections include Data management, Insights, Quality assurance, and User experience.

Sources are external systems and documents that feed into Living Knowledge, such as help centers, wikis, tickets, Slack channels, and recordings. They show connection status as Connected, Syncing, Error, or Disconnected. A Source connector is the integration module that retrieves data from a specified external system.

PII handling is a core compliance feature. PII filtering detects and removes identifiable personal data, PII obfuscation masks it during processing, and a PII filtering whitelist exempts approved data points for specific operational needs.

Security and compliance

Unless follows a compliance-by-design approach with GDPR, DORA, the EU AI Act, ISO 42001, and OWASP built into the architecture. The platform holds the ISO 27001 certificate. Security measures include encryption via Let’s Encrypt certificates, two-factor authentication, VPN access to production infrastructure, and peer-reviewed code changes.

The platform also defends against specific AI threats. Prompt injection is prevented through privilege control and input validation. Model denial of service is limited through context window limits and API rate limits. Training data poisoning is detected through anomaly detection. Secure output handling, secure plugin design, and sensitive information protection are all enforced through validation, sanitization, and data access limits.

Conclusion

This glossary covers the essential terms you will encounter when working with Unless. As you move through the Train, Test, Deploy, and Analyze phases, you will see these concepts recur across the dashboard and help center. When in doubt, the full glossary in the help center provides definitions for every term, from action buttons and audiences to workspaces and variables.