Airia

An Integration Framework provides out-of-the-box connectors linking agents to software-as-a-service platforms, databases and customer relationship management systems, and the vendor states agents connect to thousands of data sources and enterprise systems the customer already uses. Model Context Protocol support is documented as connecting the customer's systems to Airia, making the tool surface open-ended. Agents are described as running across any model, framework and data source. No enumerated connector catalog is published.

Sourceairia.com ai-platform overview and home pages, agent-builder pageread 2026-08-31

A Routing Engine distributes tasks to the most appropriate model, agent or interface and is listed among six integrated platform capabilities as an operational reliability component.

The agent builder page documents designing collaborative agent systems that execute complex workflows while remaining visible, manageable and resilient as they grow in scope, supporting no-code creation and pro-code extension in one controlled environment, with reusable logic patterns reducing policy drift.

Agents are orchestrated across models, frameworks and data sources end to end. No branching, looping or conditional control-flow primitives are named.

Sourceairia.com ai-platform agent-builder, ai-platform overview and frequently-asked-questions pagesread 2026-08-31

The Knowledge section documents an ingestion-to-retrieval pipeline over the customer's data sources: scheduled syncs, ingestion settings, enrichment, knowledge graph extraction, custom knowledge graphs, vocabulary grounding for internal terms, hybrid semantic and keyword search with fusion and reranking, graph-enhanced retrieval and multi-hop or Text-to-SQL retrieval, with permissions enforced on retrieved content.

Sourceairia.ai/docs build/knowledge pagesread 2026-09-29

Agent workflows include Human in the Loop steps that pause execution for a person: an approval step, a Webhook Approval step that waits for an external approval callback, a Form Review step, and a Pause Workflow Execution step; Agent Constraints and runtime guardrails bound what agents may access and output.

Sourceairia.ai/docs build/components human approval, webhook approval and form reviewread 2026-09-29

The vendor's FAQ states Airia supports built-in alignment with GDPR, HIPAA, PCI DSS and SOC 2 (AICPA), FERPA and COPPA, ISO standards and the EU AI Act through automated policy enforcement and data protection controls, which is alignment language rather than an attestation. The deployment page states the fully managed SaaS is hosted on AWS, Azure and Google Cloud with SOC 2 Type II certified operations, attaching the certification to the hosting arrangement ambiguously.

No named audit firm, report request path or published assessment appears on these pages, and the trust center at trust.airia.com renders no certification content. Documented controls are extensive: Security Posture Management, Agent Constraints governing data and tool access, a Routing Engine, Agent Red Teaming, AI Discovery for agent inventory, Responsible AI Guardrails, granular and role-based access controls, and centrally inherited agent policies governing model access, tool usage and deployment.

Sourceairia.com frequently-asked-questions, ai-platform security and deployment pages, agent-builder pageread 2026-08-31

AI Discovery provides a full inventory of agents across the organization, explicitly including agents that were purpose-built, commercially adopted inside licensed software, or connected by employees without approval, with the vendor stating the same controls apply the moment Airia is deployed.

Unified analytics cover usage, cost and impact, with continuous real-time monitoring and audit-ready reporting. The vendor states it automatically generates continuous governance documentation mapped to the EU AI Act, NIST AI RMF and SR 11-7, HIPAA, ISO 42001 and SOC 2, and describes complete visibility into each decision.

Sourceairia.com ai-platform security page, home page and ai-platform overviewread 2026-08-31

Memory Objects persist data across agent executions (user preferences, summaries, workflow state), created per project and scoped as global, user-specific or conversation-scoped, and written and read by Memory Store and Memory Load steps or memory tools inside the agent. No lifetime, expiry or purge path is documented.

Sourceairia.ai/docs/build/components/memoryread 2026-09-29

A dedicated Deployment Options page documents that Airia can be deployed within the customer's own dedicated cloud environments including Amazon Web Services with AWS GovCloud for United States government customers, Microsoft Azure including Azure Government Cloud, Google Cloud Platform and Oracle Cloud Infrastructure.

A fully managed SaaS deployment is offered as the alternative, hosted on AWS, Azure and Google Cloud with a 99.9 percent uptime service level and automated backup and disaster recovery. The vendor separately describes data sovereignty controls keeping sensitive data within required geographic boundaries, and flexible deployment models are named on the agent builder page.

Sourceairia.com ai-platform deployment page, agent-builder pageread 2026-08-31

The vendor's agent builder page documents leveraging trusted agent templates and reusable logic patterns to reduce policy drift and ensure consistent execution across the enterprise, alongside no-code creation and pro-code extension in one governed environment. Policy templates are referenced as a distinct category of supplied asset. No template count or browsable template catalog is published.

Sourceairia.com ai-platform agent-builder and ai-platform overview pagesread 2026-08-31

Users schedule deployed agents from the Catalog to run automatically at specified times, one-time or recurring; Security Event Triggers start an agent the moment a security event fires in Airia; an Airia Agent Trigger starts agents from data source activity; and agents are also reached through the Airia Catalog interface, an API interface, chat, a browser extension and voice.

Sourceairia.ai/docs airia-chat/capabilities/scheduling, build/triggers/security-event-triggers and integrations agent-triggerread 2026-09-29

The vendor states agents are built and run across any model, framework and data source, and describes multi-model orchestration with bring-your-own-model flexibility. A Routing Engine automatically distributes tasks to the most appropriate model, agent or interface and is listed among the platform's six core capabilities as an operational reliability component.

A named customer quote on the platform page describes the ability to use any model within a guided and controlled environment. Every agent inherits centralized policies governing model access, so administrators constrain which models are permitted while builders choose within that boundary.

Sourceairia.com home page, ai-platform overview, agent-builder and frequently-asked-questions pagesread 2026-08-31

An API interface gives each agent programmatic access (a PipelineExecution endpoint taking user input, user and conversation identifiers), API keys act as the user or as a service account with its own roles, a published API reference and a mobile SDK are documented, and Airia exposes its knowledge retrieval and connected tools to external AI clients such as Claude Desktop and Cursor over MCP.

Sourceairia.ai/docs build/interface-options/api-deployment, settings/developer/api-keys, developers-hub SDK pages and build/knowledge/guides/mcp-retrievalread 2026-09-29

Airia Evaluations measure agent and LLM application performance: create an evaluation, select up to three agents and their versions, import test cases from CSV, and compare versions, prompt changes or model updates across hundreds of test cases on metrics such as context accuracy, answer relevance, latency and cost, to catch regressions before they reach users; Agent Red Teaming stress-tests agents separately.

Sourceairia.ai/docs/build/components/evaluate-agent-performanceread 2026-09-29

Agents act exclusively through programmatic routes: an Integration Framework of out-of-the-box connectors to software-as-a-service platforms, databases and customer relationship management systems, Model Context Protocol support, and connection to thousands of enterprise data sources.

Agent Constraints govern which data and tools an agent may access, a permission model over programmatic interfaces. No browser control, navigation, form filling, screen interaction, session automation or desktop capability appears on the home page, platform overview, agent builder, security, deployment or FAQ pages.

Sourceairia.com ai-platform overview, security, agent-builder and frequently-asked-questions pagesread 2026-08-31