FortiAgent is a governed AI agent for enterprise customer support. It answers from your configured knowledge base and retrieves live data from your APIs — order status, subscription details, account history — and responds only within what you have explicitly configured.
It connects to Shopify, Stripe, Zendesk, Salesforce, and any REST API through a secure connector registry. Unlike generic AI chatbots that can hallucinate policies or answer outside their intended scope, FortiAgent is bounded by configuration — and every decision is gated by FortiVault's AI Trust Score.
Governed by FortiVault. Built for enterprises that need AI support they can audit and safely scale.
Capabilities
Generic AI chatbots answer from their training data — which means they can hallucinate policies, invent product details, or drift outside their intended scope. FortiAgent cannot. Every response comes from explicitly configured knowledge sources and live connector data — bounded by what you define, auditable at every step.
Answers only from explicitly configured knowledge sources — FAQs, policies, product guides. Responses cite what was used. Behaviour is bounded by what you provide. FortiAgent cannot answer outside your configuration — no model drift, no training data leaking through.
Calls live APIs when the answer requires real-time data — order status, account details, subscription state, tracking information. Only connectors registered in the connector registry can be called. No unrestricted tool use.
Guidance rules define tone, compliance constraints, and escalation conditions. All behaviour is explicit and auditable — not emergent. Admins can inspect every response: what knowledge was retrieved, what connector was called, what rule was applied.
Multi-step procedures handle operations that modify records — cancel orders, process refunds, update accounts. Each step is validated, audited, and requires explicit configuration. No free-form write-back.
How FortiAgent Works
Every step is explicit, logged, and inspectable. No black-box model behaviour.
Via the embeddable widget or a connected support channel. Streamed in real time.
FortiAgent searches configured knowledge sources for a matching answer — FAQs, policies, product guides.
If the answer requires live data, FortiAgent calls the relevant connector API — order status, account details, billing state.
Tone, compliance constraints, and escalation conditions shape the response before it is sent.
Grounded, accurate, and based only on what you have configured — not on general model knowledge.
Every connector call, knowledge chunk used, and guidance rule applied is logged. Admins can inspect every decision.
Connectors
FortiAgent connects to your existing APIs via a secure connector registry. When a customer asks about their order, subscription, or account — FortiAgent calls the relevant API and returns the live answer, not a templated guess.
Supported integrations include Shopify order and fulfilment data, Stripe and Chargebee subscription details, Zendesk and Freshdesk ticket history, Salesforce and HubSpot account records, DHL and FedEx shipment tracking, and any custom REST API.
Auth, field filtering, timeout protection, and circuit breaking are built in to every connector.
Commerce
Shopify, WooCommerce, Magento, BigCommerce
CRM
Salesforce, HubSpot, Zoho, Pipedrive
Billing
Stripe, Chargebee, GoCardless, Xero, QuickBooks
Helpdesk
Zendesk, Freshdesk, Intercom, HelpScout
Logistics
DHL, FedEx, UPS, Royal Mail, Evri, DPD
Identity
Auth0, Okta, internal SSO
Internal APIs
Any REST API — custom connectors supported
Any REST API can be registered as a connector. Connectors define the tools FortiAgent can call — no others are accessible.
Governance
FortiAgent handles AI execution. FortiVault governs every decision — measuring accuracy, enforcing automation policy, and requiring human review when AI trust has not been earned.
Without a governance layer, automation expands into categories where AI accuracy has not been validated. Billing errors, incorrect policy statements, and wrong account data reach customers before anyone on the team sees it.
AI Execution + AI Governance
FortiAgent resolves support queries. FortiVault determines when it is safe for FortiAgent to do so automatically.
Autonomy + Auditability
FortiAgent acts. FortiVault records every decision — knowledge source, connector call, rule applied — and flags for human review when thresholds are not met.
Execution + Governance
FortiVault decides when FortiAgent is allowed to automate — category by category, based on measured trust.
FortiVault governance rule
Automation Gate
Category: Billing & Refunds
Trust Score threshold: 90
Category accuracy threshold: 92%
Sensitive workflows — billing, refunds, account changes — require proven accuracy before FortiAgent can act without review. FortiVault enforces this automatically.
Enterprise Value
Embeddable Widget
One script tag. FortiAgent appears on any web surface. SSE streaming delivers responses word by word — no page reloads, no polling.
Bounded Autonomy
FortiAgent can only call connectors registered in its registry. No tool calls outside what you have explicitly defined. No hidden behaviour.
Escalation Handoff
When a guidance rule triggers escalation, FortiAgent hands off to a human agent with full conversation transcript — via Zendesk, email, or Slack.
FortiAgent handles execution. FortiVault handles governance. Together they form an enterprise AI support stack you can actually stand behind.
Request a demo to see FortiAgent and FortiVault working together.