Relevance AI review: an AI workforce with approvals on every connection

Contents

1. How approvals work
2. How it is billed
3. Published evidence
4. Strengths and limits
5. Pricing
6. Questions to ask Relevance AI
7. Sources

Type	AI workforce platform
Pricing	Published
Website	relevanceai.com

Relevance AI calls itself the home of the AI workforce: teams of specialised agents, such as a researcher, a meeting scheduler, an outbound prospector and a deal reviewer, connected to more than 1,000 apps including HubSpot, Salesforce, Slack and Gmail⁠1The home of the AI WorkforceRelevance AI.

Best for: Sales and go-to-market teams building a set of agents (research, prospecting, scheduling, deal review) with approval rules on every connection between agents and tools.

Not the right fit for: Small teams wanting a self-serve free plan, or buyers who need RBAC and audit logs without an Enterprise contract.

How approvals work

Relevance AI sets approval on each edge of a workforce, meaning each connection between an agent and a tool or between two agents. Each edge can run automatically, require approval, or let the agent decide. When approval is required, the agent drafts the action it intends to take and waits; the reviewer can approve, reject or add guidance from the task view. Escalations can route to people or to other agents, for example by Slack or email, and a cap limits how many times an automatic edge runs on its own⁠2Approvals and escalationsRelevance AI Docs.

Roles exist at three levels (organisation, project and asset), and an organisation viewer can read the audit logs⁠3Role-based access controls (RBAC)Relevance AI Docs. Role-based access, SSO, audit logs and multi-organisation management are Enterprise features, and RBAC is still being rolled out to Enterprise customers⁠43PricingRelevance AI DocsRole-based access controls (RBAC)Relevance AI Docs.

How it is billed

The public pricing page shows only an Enterprise plan⁠5PricingRelevance AI. Relevance AI's documentation lists self-serve plans: Pro at $19 a month billed annually for 2,500 Actions, and Team at $234 a month billed annually for 7,000 Actions; the free plan has been retired. An Action is one run of a tool, and a failed run still counts. Model costs are passed through as vendor credits without markup⁠64Plans and creditsRelevance AI DocsPricingRelevance AI Docs.

Published evidence

• Qualified. Relevance AI reports that Qualified ran more than 35 agents and generated $7 million in pipeline and $500,000 in closed revenue over six months. Vendor-reported.⁠7How Qualified scaled with 35+ AI agentsRelevance AI · 1 May 2025
• Zembl. Relevance AI reports that Zembl increased customer conversion by 30% without adding headcount and cut average call time by 60%. Vendor-reported.⁠8Zembl built a 24/7 AI sales workforceRelevance AI · 1 Mar 2026

Strengths and limits

Strengths

• The most granular approval model reviewed: set per edge, with the agent drafting its intended action and waiting for a person.
• Reviewers can approve, reject or give further guidance, and escalations can go to people or other agents by Slack or email.
• Publishes named customer results with figures (vendor-reported), from Qualified and Zembl.

Limits

• SSO, role-based access, audit logs and multi-organisation management are Enterprise-only, and RBAC is still being rolled out.
• The public pricing page shows only Enterprise; self-serve prices appear in the documentation, and the free plan has been retired.
• A failed tool run still counts as a billable Action.
• Data residency is claimed on the homepage but not documented in the pages we read.

Choose another option if: You need audit logs and roles on a mid-market budget, or agents that work on your own records rather than through a CRM. OpenLazarus or n8n fit those needs better.

Pricing

Documented: Pro $19 a month billed annually ($29 monthly) for 2,500 Actions; Team $234 a month billed annually ($349 monthly) for 7,000 Actions. Model costs are passed through as vendor credits.

Questions to ask Relevance AI

• Which edges in our first workforce should require approval, and who approves them?
• Do we need Enterprise for audit logs, and is RBAC available to us yet?
• How many Actions would one agent use in a month, including failed runs?
• Where is our data processed and stored?

See how Relevance AI compares with the other agent builders in the main ranking.

Sources

1. The home of the AI Workforce. Relevance AI. Accessed 8 Nov 2026.
2. Approvals and escalations. Relevance AI Docs. Accessed 8 Nov 2026.
3. Role-based access controls (RBAC). Relevance AI Docs. Accessed 8 Nov 2026.
4. Pricing. Relevance AI Docs. Accessed 8 Nov 2026.
5. Pricing. Relevance AI. Accessed 8 Nov 2026.
6. Plans and credits. Relevance AI Docs. Accessed 8 Nov 2026.
7. How Qualified scaled with 35+ AI agents. Relevance AI, 1 May 2025. Accessed 8 Nov 2026.
8. Zembl built a 24/7 AI sales workforce. Relevance AI, 1 Mar 2026. Accessed 8 Nov 2026.

Further reading

• Buyer's guide · Updated quarterlyThe best no-code AI agent builders for company work in 2026An analysis of seven platforms for building AI agents without code, judged on the controls a company needs before an agent changes a real order, invoice or customer record.
• Guide · ControlAI agent permissions and approvals: how to let agents act safelyAn agent's permissions are a job description written in advance. The platforms differ less in whether they offer approvals than in where those approvals apply and what record they leave.
• Guide · CapabilitiesWhat AI agents can do for a business today, and what they cannotAgents now answer mail and calls, key orders, chase invoices and prepare reports. They still fail about one task in three on hard benchmarks, which decides where they should start.