Relevance AI Buyer's Checklist: Plans, Controls, Data and Exit

Relevance AI is a sound buy for a team that wants hosted AI agents and Workforces, can budget two usage meters, and can live without SSO, RBAC or audit logs until it signs an Enterprise contract. The cheapest way in is Pro at $19 a month billed annually or $29 a month billed monthly, and that price covers the whole organization rather than each person who logs in ^[1]. Every organization now needs a paid plan, and Relevance AI's pricing pages describe no free trial. Self-hosting is a hard stop. Everything else below is a question of plan level, usage arithmetic and how cleanly you can leave.

Where the price sits and what one subscription covers

Relevance AI's public pricing page shows only a custom Enterprise plan with a Talk to sales button; the self-serve Pro and Team prices are published in its pricing documentation, and you upgrade to them inside the app. Procurement teams expecting one price sheet often stall here, yet the documented prices are current. The subscription attaches to the organization, so everyone in it shares the plan's Actions, Vendor Credits and features. Build Users and End Users are included counts, not priced seats: Pro carries 2 Build Users and no End Users, while Team carries 5 Build Users and 45 End Users.

Limits are enforced softly. People already over the count stay without extra charge, but nobody new can join until you are back under it, and on Pro only two users can be added at all. Payment is by credit card only, and debit cards are not supported ^[2]. Subscriptions cannot move to another organization, so pay for the one you intend to keep.

• Count who builds agents and who only chats with them. Anyone in the second group pushes you toward Team.
• Confirm that a company credit card, not a debit card, sits on the account.
• Create the organization deliberately: its region and its subscription stay with it.

Two meters, and runs that bill even when they fail

An Action is one run of a Tool. One email or a many-step workflow inside a single tool counts as one Action, and failed runs and retries still count. Failures still bill. Each Eval check also costs one Action. Pro includes 2,500 Actions a month (30,000 a year) and Team 7,000 a month (84,000 a year), and plan Actions reset at each renewal. Vendor Credits are the second meter: model and tool costs passed through at wholesale with no markup, where 10,000 credits equal $20. Pro's allowance is worth $20 a month and Team's $70 a month, and both roll over indefinitely while you stay subscribed.

Top-ups are packs: $80 per 1,000 Actions and $20 per 10,000 Vendor Credits, on paid plans only, bought by an organization admin. Action top-ups roll over to the next billing cycle; Vendor Credit top-ups last while you stay subscribed. By our arithmetic, Pro billed monthly plus four Action packs reaches $349 a month for 6,500 Actions, which equals Team billed monthly with 7,000. Above 6,500 Actions a month on monthly billing, Team wins, if your model spend fits Pro's 10,000 Vendor Credits a month. On annual terms the crossover sits at 62,000 Actions a year (roughly 5,200 a month), where Pro plus 32 packs comes to $2,788 a year against Team's $2,808 a year for 84,000. Heavier model use lowers both crossovers.

• Estimate Actions from Tool runs, and fold steps into one tool where the logic allows it.
• Set usage alerts and hard usage limits; both exist on every plan. Automatic top-ups (Spend Controls) exist on Pro and Team only.
• Treat your own model keys as a Vendor Credit saving only. Bring-your-own-key is optional on every plan, but the subscription and Actions stay unchanged.

Which controls wait behind Team or Enterprise

Pro is a complete builder, but several features a buyer may assume are standard sit one or two levels up. Check them first. The grid below comes from the pricing documentation and Relevance AI's docs pages, and leaves prices out on purpose. Gates decide the plan here. Read the grid against your requirements list, row by row, before the sales call or the upgrade button.

The trigger row catches support and sales teams most often. Salesforce and Zendesk tool steps work on every plan, so an agent can read and update those systems, but an event there can only start the agent on Enterprise; Salesforce trigger connections made before September 2025 keep working. A help desk built around Zendesk tickets therefore changes the plan decision; our customer support guide covers which helpdesk triggers work on each plan. Premium triggers for WhatsApp, LinkedIn and Telegram, by contrast, run on every paid plan. Price the gate, not the plan name. Team costs $234 a month billed annually or $349 a month billed monthly, while Enterprise has no published number.

• Map each requirement on your list to a row above before you talk numbers.
• If an IT policy demands SSO or audit logs, plan a sales conversation from day one.

Data, security and the hosted-only answer

SOC 2 Type II and GDPR compliance appear on every plan in the pricing table. Data travels over TLS 1.2 or later and rests under AES-256 encryption, and Relevance AI lists its subprocessors on its trust site . Relevance AI says it does not use customer data to train its models unless a specific partnership agreement says otherwise. Storage sits in one of three regions: US (N. Virginia), EU (London) or AU (Sydney). The region is chosen when the organization is created and cannot be changed; moving means a new organization arranged through support. Choose once. Data can be exported at any time as CSV, Excel or JSON.

Relevance AI runs as a hosted service, and the docs describe no self-hosted edition. Enterprise buyers can request third-party reports and security questionnaires under NDA. One dated event also belongs in a buyer's file. In May 2026 Relevance AI ran a one-time deletion of Knowledge tables on non-Enterprise accounts left inactive for 182 days, permanently, after a warning email four days earlier. Nothing in the docs says it will recur. It is still a reason to keep your own copies.

• Pick the region your customers and contracts require before signup.
• Keep your own copy of every source file you load into Knowledge.

Volume, public agents and the API

Concurrency covers tasks running at once across the whole organization. Relevance AI's docs give concurrency per plan only as Standard, More or Custom, and tasks above the limit queue until capacity frees up rather than failing. A concurrency chart under Plan and Billing shows real-time use. If a campaign fires hundreds of tasks at once, measure the queue in a pilot before promising turnaround times to anyone.

Sharing carries a billing risk. Turning on any embed option or a cloneable template makes an agent public at an indexable URL, and anyone with it can run the agent on your credits. Since May 2026, admins can switch off public cloning and embedding per project. The API is officially supported only for triggering Agents and Tools, a trigger returns no direct response, and rate limits depend on the plan with no numbers published. The documented SDK is JavaScript and TypeScript. Our API and MCP guide covers keys, regions and endpoints.

• Disable public cloning and embedding on every project that does not need it.
• Confirm your developers can work with results sent to a destination instead of an API response.

Support, reviews and the way out

First-response targets are three business days on Pro, two on Team and one on Enterprise. Pro and Team support runs on weekdays from 9am to 5pm Sydney time with no weekend cover, though engineers are on call around the clock for outages. Enterprise adds a dedicated Slack channel and an account manager, and implementation help is Enterprise-only. A US team on Pro or Team should expect replies on Sydney's working day, not its own. Plan around that.

Relevance AI has only a small number of public user reviews so far, so we score user sentiment as neutral rather than reading a trend into them. Leaving is simple to execute. An organization admin cancels in the app, and the subscription runs to the end of the current paid cycle. Fees are non-refundable for change of mind, and unused Actions and Vendor Credits expire at cancellation without a refund . Downgrades take effect at the next billing cycle and are generally not pro-rated. The full plan math lives in our pricing breakdown.

• Name an organization admin early: only admins buy top-ups, see invoices and cancel.
• Export your data and spend down credits before the cycle you cancel in ends.

Sign Pro, budget Team, or call sales first?

Sign Pro when a couple of builders run agents behind the scenes, Action use stays under about 6,500 a month on monthly billing (about 62,000 a year, roughly 5,200 a month, on annual billing, by our arithmetic), model spend fits Pro's 10,000 Vendor Credits a month, and no one outside the build team needs a seat. Heavier model use moves the crossover lower. Budget Team the moment you need End Users, phone or meeting agents, bulk scheduling or the analytics dashboard. Call sales first if your security review asks for SSO, RBAC, audit logs or evaluations, or if a Zendesk or Salesforce event must start the agent. Walk away only if self-hosting is a requirement. Choose the data region before you create the organization.