Relevance AI for Sales: Research, Outbound and CRM Agents

Relevance AI gives a sales team a place to build its own agents for prospect research, outbound drafts, enrichment, lead scoring and meeting briefs, starting from Marketplace templates you clone and connect to your CRM. The self-serve entry is Pro at $19 a month billed annually or $29 a month billed monthly, for the whole organization, with 2,500 Actions a month. That covers research and outbound. It does not cover everything. Calling agents, meeting agents and bulk scheduling start on Team, the Salesforce trigger needs Enterprise, and the ready-made BDR agent is sold through the sales team. Pick it if RevOps will build and tune agents; skip it if you want a finished SDR in a box.

Morning prep: research briefs before the first call

The sales guide in Relevance AI's docs lists six use cases: prospect research, personalized outbound, lead scoring, CRM data enrichment, meeting briefs and competitive intelligence. Most reps feel the first one daily. The Sales Researcher template builds call-prep briefs covering the company, its contacts, recent news and likely pain points. The docs' prospect research workflow takes a company domain, a contact email or a CRM record and returns an overview, key contacts, pain points and conversation hooks. Sales triggers can be a schedule, a CRM record update or a webhook, so briefs can be ready before the day starts.

Booked meetings get their own template. The Pre-Meeting Prepper pulls CRM activity, past notes, email threads, support tickets and news into one brief, delivered by email, as a calendar event or in Slack. Competitive questions are covered too. The Sales Battlecard Creator monitors competitor sites, news, job boards and social media, then produces battle cards and weekly digests, and the docs suggest the Firecrawl tool step for reading competitor sites. That step needs your own Firecrawl API key, and Firecrawl bills those credits.

After the meeting, the plan starts to matter. Meeting agents send a Notetaker bot into Google Meet, Zoom or Microsoft Teams that records audio and video and writes speaker-labeled transcripts, and they are not on Pro: they start on Team. The bot has to be admitted as a guest, and automatic mode runs on a Calendar trigger plus a Meeting Bot trigger. Phone agents sit behind the same Team gate. They make outbound qualification calls in real time, can run tools during the call, detect voicemail and use ElevenLabs for the voice and Deepgram for transcription.

A Pro team can still work from recordings it already has. The docs' discovery-summary guide takes a recording or transcript, for example from Fireflies, and turns it into a structured summary with action items using the Nate Meeting Notetaker template. The Fireflies integration offers tool steps to find a meeting by ID, find the most recent meeting or upload audio for transcription. That route needs no meeting bot.

Outbound: drafts in a queue, not sends on autopilot

The outbound guide queues drafts for rep approval and suggests two templates. LinkedIn Outreach and Follow up writes personalized connection requests and follow-ups from prospect research, and Outbound Composer writes multi-variant email. ^[1] The same guide advises sending caps and a warm-up schedule on new domains, with high-value accounts routed through human review using Alerts. The docs recommend the built-in Gmail warm-up integration on a dedicated inbox, starting at 4 to 8 warm-up emails a day and capped at 50 a day. New domains need patience.

LinkedIn is where Pro earns its place for SDRs. The LinkedIn trigger is a premium trigger on Pro and above, and it can fire on every message received or only on replies to outreach, for agents and Workforces alike. Connection-accept notifications are tracked only when the request went out through the Relevance AI LinkedIn integration, so mixing manual and agent invitations leaves gaps. On email, sending tools can run in approval mode, where the agent asks before each send, and switch to autopilot later.

Each touch has a price in Actions. Every Tool run counts as one Action, whether it sends one email or runs a many-step workflow, and failed runs and retries still count. A lead that passes through research, a draft and a CRM update built as three separate tools uses three Actions; the same steps combined into one tool use one, and the vendor itself suggests combining steps to save Actions. Model cost runs on a second meter, Vendor Credits, passed through at wholesale with no markup.

CRM hygiene: enrichment, scoring and a third meter

Two Marketplace agents from the RevOps guide carry most of this work. Elli the Enrichment Agent fills firmographics, contact details and job-function classification, and Lia, the LinkedIn Lead Researcher and Qualifier, is the template the lead-scoring guide suggests. A scoring agent returns a score with a reasoning summary and a recommended action, then writes it to the CRM, posts hot leads to Slack or routes them. The RevOps guide chains enrich, score, route and notify, triggered on record creation, a stage change or a freshness schedule, which is a natural job for the Workforce canvas.

The data behind enrichment is billed by someone else. Apollo enrichment uses credits from your Apollo plan, and the docs say the API needs at least an Apollo Basic plan. Lusha, labeled beta, bills its own credits, typically 1 to 2 per contact lookup. Firmable, also labeled beta, bills Firmable credits and does not hold every ANZ contact, so agents must handle empty results. The enrichment guide warns that enriching every record on every event burns credits at scale, and advises gating on lifecycle stage and a freshness window. ^[2] Budget three meters, not two.

HubSpot and Salesforce connect differently. A HubSpot-triggered agent runs through a HubSpot workflow whose action sends a message to the agent, so it can start on a form fill, an email link click, a lead score being reached or a property update. Salesforce is stricter. Its trigger, an SOQL query that sends new matching records such as Leads to the agent, is an Enterprise trigger, even though Salesforce tool steps work on every plan. The integrations guide covers both setups, along with Gmail, Outlook and Slack.

For pipeline review, the Salesforce Assistant template asks for explicit confirmation before creating, updating or deleting any record, and the HubSpot Assistant manages contacts, deals and companies through plain-English conversation. The Deal Health Analyst flags stalled deals, missing critical fields and slippage risk, delivered to Slack, a Notion ops doc or deal-level alerts to account executives. That gives a sales manager a weekly hygiene pass without chasing reps for fields.

Which plan a sales team actually needs

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. ^[3] Every organization now needs a paid plan, and Relevance AI's pricing pages describe no free trial. Payment is by credit card only, and an organization admin buys any top-ups.

Users decide this as often as features do. On Pro only two users can be added to the organization, and Pro includes no End Users, so a group of reps who want to use agents in Chat outgrows it fast. Team includes 5 Build Users and 45 End Users and costs $234 a month billed annually or $349 a month billed monthly for the whole organization, with 7,000 Actions and $70 of Vendor Credits a month. Bulk scheduling on Team turns each spreadsheet row into one task, paced by time zone, work days and tasks per interval.

Volume is the other trigger. Pro includes 2,500 Actions a month. Extra Actions are a top-up pack at $80 per 1,000, and an Action top-up rolls over only to the next billing cycle. By our arithmetic, Pro billed monthly plus 1,000 extra Actions comes to $109 a month for 3,500 Actions, and above 6,500 Actions a month on monthly billing Team is cheaper than Pro with top-ups. On annual billing the crossover sits at about 62,000 Actions a year (roughly 5,200 a month), by our arithmetic. Counting Actions alone, below those lines Pro costs less, if your model spend fits Pro's 10,000 Vendor Credits a month. If you also buy Vendor Credit packs on Pro, Team pays off sooner, since it includes 35,000 Vendor Credits a month.

The BDR agent and the named sales agents are sold, not cloned

Relevance AI's flagship sales product is the BDR agent, earlier marketed as Bosh, a curated multi-agent system in which subagents each own one step of the BDR process. It is reached by booking a demo and discussed with the sales team alongside Enterprise plans. The same holds for the named agents on the use-case pages, such as the Outbound Prospector and the Research and Enricher, which Relevance AI's experts tune with the buyer. You cannot clone them.

The self-serve team can still copy their habits. Relevance AI says the Outbound Prospector holds drafts for review until the buyer trusts the quality bar, and that the Research and Enricher leaves a field blank when no source confirms it. Both habits transfer. Relevance AI says Qualified built 35+ AI agents with it and generated $7 million in pipeline in six months. ^[4] That is a vendor-published story about many specialist agents, not a forecast for a first pilot.

Where a sales team should start on Relevance AI

Start with one workflow that has a clear owner and an output you can check, such as meeting briefs or enrichment for new inbound leads, on Pro with every sending tool in approval mode. Watch the Actions and the data-vendor credits for two weeks before adding outbound volume. Move to Team once you need meeting or calling agents, more than two people in the organization, or volume past the crossover above. Hand lifecycle and MQL work to marketing's agents, and call sales only for the BDR agent, the named use-case agents or Salesforce triggers.