Relevance AI

What is Relevance AI?

Relevance AI is a no-code platform for building AI agents and multi-agent teams that carry out business tasks across connected apps. You describe a job, attach tools and knowledge, and the agent runs it on a trigger or schedule.

The platform groups its building blocks into agents, tools, workforces and knowledge. An agent is an AI worker with instructions and a chosen model. Tools are no-code workflows that let it act, such as sending email or updating a CRM record. Workforces connect several agents so that work passes between them, and knowledge gives them access to your own documents and websites.

Relevance AI targets sales, support and operations teams in mid-sized and large companies, and its homepage shows logos such as Canva, Databricks and Autodesk. The product is sold on paid self-serve plans and custom enterprise contracts, and the older Free plan has now been retired according to the official documentation.

Key Features

Tool-using agents sit at the centre of Relevance AI, and workforces, triggers and phone agents extend them into complete business processes.

• Agent builder: You create an agent by writing its instructions, choosing a model and attaching tools and knowledge. The Invent option drafts a working agent from a plain description, and the Marketplace lets you clone tested templates instead of starting blank.
• No-code tools: Tools are step-by-step workflows that let an agent send email, update a CRM, search the web or call an API. Each time an agent runs a tool, the platform counts one Action against your monthly allowance.
• Workforces: A workforce links several specialist agents on a visual canvas so they can hand work to one another. This suits processes such as research followed by drafting and then CRM updates, where one agent alone would struggle.
• Knowledge sources: Agents can draw on uploaded files, Google Drive, SharePoint, Notion and websites through retrieval. Answers are then grounded in your own material rather than only the model’s general training data.
• Triggers and approvals: Agents can start from Gmail, Slack, Microsoft Teams, WhatsApp, webhooks, API calls or a schedule. The documentation also describes approval steps and escalation rules, so people can check sensitive actions before they run.
• Phone agents: Phone agents place outbound calls, speak with ElevenLabs voices and run tools while the call is live. Numbers can come from Relevance AI or your own Twilio account, and calling is limited to Team and Enterprise plans.

Pros

Model flexibility is the strongest argument for Relevance AI, since it supports five provider routes and passes model costs through at wholesale rates.

• Choice of models: The documentation lists OpenAI, Anthropic, Google Gemini, Azure and OpenRouter as model providers. You can match a cheaper model to simple steps and keep stronger models for reasoning-heavy tasks.
• Wholesale model costs: Relevance AI states that Vendor Credits cover model and tool costs at wholesale rates with no markup. Bring-your-own API keys are also listed, which helps teams that already hold provider contracts.
• Wide set of triggers: Agents can react to email, chat messages, webhooks, API calls and schedules without extra middleware. Sales and support teams can therefore fit agents into channels their staff already use daily.
• Enterprise security basics: The security page lists SOC 2 Type II, GDPR compliance, AES-256 encryption at rest and a choice of US, EU or Australian data storage. It also states customer data is not used for model training.
• Recognisable customers: The homepage shows logos including Canva, Databricks, Confluent, Autodesk and Freshworks. That gives prospective buyers some reassurance that the platform runs in larger, security-conscious organisations.

Cons

Cost predictability is the weak spot, because usage is split across Actions and Vendor Credits and there is no longer any free plan.

• No free plan: The pricing documentation says the Free plan has been retired for new and existing accounts. You must pay at least the Pro rate before building anything, which makes casual testing harder.
• Two meters to track: Actions and Vendor Credits are billed separately, so a busy agent can run out of one while the other remains. Budgeting needs regular checks of both balances in the usage dashboard.
• Costly top-ups: Extra Actions cost $80 per 1,000 under the official documentation. Agents that call many tools per task can burn through the 2,500 Pro allowance quickly and raise monthly costs.
• Steep jump to Team: Calling agents, A/B testing and end users only arrive on Team, which costs $234 a month billed annually. Small teams needing just one of those features face a large price step.
• Learning curve: Agents, tools, workforces, knowledge and triggers are separate concepts that interact. New users usually need the Academy courses or templates before production agents behave predictably.

Who is Using Relevance AI?

Go-to-market and operations teams form the core audience, alongside developers who manage agents through the API and SDK.

• Sales teams: Revenue teams build research and outreach agents for lead qualification. They use them to enrich CRM records and draft first-touch emails.
• Support teams: Support leads deploy agents that triage tickets and draft replies. Approval steps let staff check answers before customers see them.
• Operations staff: Operations managers automate repetitive back-office steps across connected apps. Scheduled triggers run reports and updates without manual effort.
• Developers and agencies: Technical builders use the API, JavaScript SDK and MCP server to manage agents in code. Agencies then hand finished agents to clients.

Pricing

Relevance AI starts at $19 a month on annual Pro billing, with Team at $234 and Enterprise on quote (official documentation, October 2026).

Directories still disagree: Futurepedia lists a free tier and a $199 Standard plan, one pricing blog quotes overage at $40 per 1,000 Actions, and another says annual billing gives no discount, none of which matches the official documentation.

Disclaimer: Relevance AI has changed its plan structure several times, so treat these figures as a snapshot and check the live pricing page for your region and currency.

What Makes Relevance AI Unique?

Pass-through model billing is the clearest differentiator. Relevance AI separates the platform fee (Actions) from model spend (Vendor Credits), states that model and tool costs are charged at wholesale, and lets you swap in your own API keys. Many agent builders bundle model costs into opaque credits instead.

Breadth is the second difference. One account covers single agents, multi-agent workforces, outbound phone agents, retrieval over company documents and developer access through an SDK and MCP server. Teams can therefore prototype in the no-code builder and later manage the same agents from code.

Compatibilities and Integrations

Connections to business apps, model providers and developer tooling are where Relevance AI is strongest, with over 1,000 app integrations cited on its homepage.

• Platform: Relevance AI is a web application, with agents shared through chat links, embeds or connected messaging apps. Data is stored in the US, EU (London) or Australia (Sydney), chosen at signup.
• Apps: The homepage cites over 1,000 app integrations, including HubSpot, Salesforce, Slack, Gmail, Notion, Zendesk and Snowflake. The Enterprise plan description refers to more than 2,000.
• Models: Native connections cover OpenAI, Anthropic, Google Gemini, Azure and OpenRouter. Bring-your-own keys let you bill model usage to your own provider account.
• Developer access: API triggers, a JavaScript SDK and an MCP server allow agents to be called or managed from code. A Claude Code plugin is also documented.
• Telephony: Phone agents use ElevenLabs voices and either Relevance AI numbers or a connected Twilio account. Calling is available on Team and Enterprise plans.

Relevance AI Tutorials

A first working agent takes six steps, and the Marketplace templates shorten the process if your use case is common.

1. Choose a paid plan and select your data region at signup.
2. Use Invent to describe the job, or clone a Marketplace agent.
3. Attach tools, connect apps and add knowledge such as files or a website.
4. Test the agent in the Run tab and watch each tool call step by step.
5. Add approval rules for sensitive actions, then set a trigger or schedule.
6. Track Actions and Vendor Credits in usage reports and adjust models to control cost.

How We Rated It

Breadth and model choice lift Relevance AI to 4.1 out of 5, while the retired free plan and two-meter billing pull its value score down.

Summary

Relevance AI suits sales, support and operations teams that want to build tool-using AI agents without writing code, while keeping control over which models they pay for. Its feature set is broad, covering workforces, phone agents, retrieval and developer access, and its security documentation covers the basics larger buyers expect. The main limitation is cost predictability, because there is no free plan and usage is metered through both Actions and Vendor Credits. Phone agents also need clear consent from the people being called and a check of local calling rules. Start on monthly Pro billing, build one narrow agent and measure Actions per task before committing to an annual plan.