Gong Agents
Gong · Gong Agents: Pre-built and Custom AI Agents for Revenue Teams, Grounded in Sales Calls and CRM Data
Gong Agents are the AI agents inside Gong's revenue platform: pre-built agents that fill CRM fields from calls and emails, review calls and deals, draft follow-up emails, flag deal risk and predict pipeline, plus custom agents you build yourself in plain language. They are for sales, RevOps and customer success teams that already record their customer conversations in Gong, or are willing to adopt it.
Best for
Sales, RevOps, enablement and customer success teams at mid-size and large B2B companies that already record calls in Gong and want CRM updates, call and deal reviews, follow-ups and forecasting signals produced from those conversations
Not ideal for
Teams that do not use Gong and want a standalone agent, small teams that want self-serve pricing, outbound prospecting at volume (an AI SDR is a better fit), and organizations that need self-hosted AI or agents acting across many non-sales systems
Who it's for
Sales reps and managers, RevOps, enablement and customer success teams at B2B companies running Gong
Capabilities
- AI Data Extractor: answers questions you define (for example which competitors came up, who the decision makers are, how many seats were requested) from customer conversations, keeps the answers current as new calls come in, and saves them in Gong or writes them to existing CRM deal or account fields; up to 20 published extractors per workspace
- AI Call Reviewer suggests or automatically fills in answers on call review cards, so managers can give reps feedback more often
- AI Trainer builds role-play scenarios from your own past conversations, with feedback for each rep
- Pipeline agents: AI Deal Reviewer checks deals against your sales methodology, AI Deal Monitor flags risks such as budget or priority changes and lets you drill down to the conversation a signal came from, AI Deal Predictor scores each deal's likelihood of closing, and AI Revenue Predictor forecasts revenue from pipeline signals
- AI Composer drafts follow-up emails from call transcripts and rewrites messages in a chosen tone; AI Tasker suggests the next to-dos for each rep
- AI Briefer produces structured briefs on contacts, calls, deals and accounts from pre-built or custom templates, and can pull data from external systems through Gong's MCP client
- AI Theme Spotter finds recurring pain points, objections and competitive threats across customers; AI Tracker detects defined concepts in calls and emails
- Agent Studio, where admins configure the pre-built agents, preview some before rollout and set where they appear and how much autonomy they have
- On-demand agents: any user can save a personal reusable prompt, optionally with attached files, and run it from Gong Assistant on a call, account or deal
- Automated agents in Agent Builder: one trigger (CRM record change, date condition or smart tracker), optional filters, and actions run in sequence (generate with AI, internal email, Slack message, send a brief, add to a Gong Engage flow, start a play), with a run history for each agent; creating them requires the Manage automations permission, and they are shared with the whole company
- Deep Mode in Gong Assistant: asks clarifying questions, proposes a research plan, then analyzes many calls, deals or accounts and emails you a report with inline citations that you can download as Markdown
- Gong Enrich fills missing contact and account data through a configurable waterfall of data providers, on demand or triggered by an agent
- Read-only Gong MCP server with ask_account, ask_deal and generate_brief tools for ChatGPT, Microsoft Copilot and Copilot Studio, Salesforce Agentforce and private MCP integrations; with personal access, each user sees only data they can already view in Gong
- On-demand and automated agents use the Gong permissions of the person running them, so they only see data that person can view
- Gong's trust page lists SOC 2 Type 2, ISO/IEC 27001 and ISO/IEC 42001 among its certifications and says customer data is not used to train generative models
Limitations
- Requires a Gong subscription: the agents run inside Gong on data Gong has captured, and several pre-built agents belong to add-on applications (for example AI Composer and AI Tasker in Gong Engage, the deal and revenue predictors in Gong Forecast)
- No public prices: Gong quotes per-user licenses plus a platform fee based on user count, and the price of Gong Credits is not published
- Heavier AI work consumes Gong Credits from a shared company pool: automated agents, Deep Mode (Gong estimates 1,000 to 3,000 credits per report), question-based AI Trackers, the MCP server and some APIs
- Gong's help center documents automated agents, but its September 30, 2026 Mission Callisto announcements say automatically triggered custom agents are rolling out through 2026, so check whether they are enabled on your account; manually run custom agents are included with a Gong Foundation seat
- Gong Activate was announced in the future tense at Celebrate '26 without a published release date
- Each automated agent has a single trigger and not every action works with every trigger; on-demand agents are personal to their creator and cannot use triggers, actions or MCP integrations
- Gong markets the Revenue Harness as giving agents a full audit trail, but its help center says agent activity is not available in a centralized audit log: users see their own agent sessions in Gong Assistant history, and automated agents keep a run history
- AI Data Extractor does not follow the configuring admin's permissions: it can analyze any non-private call or email in Gong, and admins testing an extractor see its answers even for calls they cannot open
- Gong connects to one CRM at a time; Salesforce, HubSpot and Microsoft Dynamics 365 are native, while other CRMs upload data through Gong's CRM API, which imports data only and does not support Gong Engage
- Automated agents choose from a fixed list of actions (AI analysis, internal email, Slack message, brief, Engage flow, play); calling arbitrary external tools or APIs from a custom agent is not documented, so this is not a general-purpose agent platform
- The underlying AI models are not disclosed, and no self-hosted deployment is documented
Use cases
- Filling CRM opportunity fields such as competitors, decision makers and next steps from what customers said on calls, instead of asking reps to log them
- Pre-filling call review cards so a sales manager can coach more reps each week
- Flagging at-risk deals and checking methodology gaps before a pipeline or forecast review
- Drafting a follow-up email from a call transcript right after the meeting
- Running an automated agent when an opportunity changes stage or a renewal date approaches, to send a brief to the account owner or post a Slack alert
- Researching why deals are being lost to a competitor across two quarters of calls with Deep Mode
Our take
The case for Gong Agents is the data underneath them: most sales AI works from what reps log in the CRM, while Gong's agents start from the calls and emails themselves, so tasks like filling deal fields, reviewing calls and spotting risk have more to go on. That advantage only exists if Gong already captures your conversations, which makes this a decision about the Gong platform rather than about agents on their own. For existing customers, the pre-built agents are the low-risk starting point; custom automated agents are useful but new, limited to Gong-native actions, and billed in credits, so pilot them on a few workflows and watch credit usage before rolling them out widely.
Who should use it
Existing Gong customers, and teams evaluating Gong, whose reps spend time on CRM updates, follow-ups and deal reviews, whose managers want call coaching at scale, and whose RevOps team wants forecasting and alerts grounded in customer conversations rather than rep-entered fields.
Who should skip it
Teams that want a standalone sales agent without a conversation-intelligence platform, small teams that need transparent self-serve pricing, companies whose main need is high-volume outbound prospecting, and organizations that need self-hosted AI or agents that act across systems Gong does not connect to.
Strengths
- Works from recorded calls and emails, not only what reps type into the CRM
- Pre-built agents cover CRM data entry, coaching, deal review, forecasting and follow-ups
- Custom agents are built in plain language, without engineering help
- Custom agents run with each user's existing Gong permissions
- Read-only MCP server brings Gong insights into ChatGPT, Copilot and Agentforce
Weaknesses
- Only available as part of a Gong subscription
- No public pricing, and heavier AI work draws on Gong Credits
- Automatically triggered custom agents are still rolling out
- Several agents depend on add-on applications such as Engage and Forecast
- No centralized audit log of agent activity, per Gong's help center
Gong Agents pricing
Custom
Custom
- Per-user licenses plus a platform fee based on the number of users
- Manually run custom agents included with a Gong Foundation seat
- Shared pool of Gong Credits based on seat type and count (2,000 credits a year per paid core seat, per Gong's docs), used by automated agents, Deep Mode, question-based AI Trackers, the MCP server and some APIs; more credits can be purchased
- Some pre-built agents come with add-on applications such as Gong Engage and Gong Forecast
Note: Gong does not publish list prices; its pricing page asks for team size and prepares a customized proposal, and says you can integrate your existing tech stack at no extra cost. Credit pricing is not published. Gong's credit docs say each paid core seat includes 2,000 credits a year, shared across the company and reset each contract year; purchased credits last until the end of the contract term, and admins can set monthly credit limits per user or team. Processing rates include 1 credit per call longer than 10 minutes, 0.3 credits per shorter call and 0.1 credits per email, and Deep Mode shows an estimated 1,000 to 3,000 credits before each report. Gong is also sold through Microsoft Marketplace.
Technical specs
Where Gong Agents excels
Keeping opportunity fields current without rep data entry
AI Data Extractor answers questions such as competitors mentioned or decision makers involved from calls and emails, and writes them to the CRM fields you choose, which helps reporting that depends on reps logging those details.
Weekly pipeline review
AI Deal Reviewer, AI Deal Monitor and AI Deal Predictor flag methodology gaps, hidden risks and close likelihood ahead of the meeting, and AI Deal Monitor lets you drill down to the conversation a risk signal came from.
Renewal preparation on a schedule
An automated agent with a date-condition trigger can send a brief to the account owner or post to Slack a set number of days before a renewal date, without anyone remembering to start it.
Win-loss research for planning
Deep Mode analyzes many calls and deals behind a question such as why deals are lost to a competitor, and returns a cited report that can feed a QBR or annual plan.
How Gong Agents fits alongside other tools
Gong Agents and Agentforce
Agentforce builds agents on Salesforce CRM and Data 360 for service, sales and commerce. Gong Agents work from recorded calls and emails captured in Gong. Gong documents how to connect Agentforce to its MCP server, so an Agentforce agent can ask Gong about an account or deal while Salesforce stays the system of record.
Gong Agents and Apollo AI
Apollo combines a B2B contact database, sequencing and outbound agents for prospecting. Gong Agents work on conversations and deals already in progress. Apollo is also one of the data partners Gong Enrich can draw on for missing contact details.
Gong Agents and Clay
Clay is a spreadsheet-style workspace for building custom enrichment and prospecting workflows across many data providers. Gong Agents analyze customer conversations to update the CRM, coach reps and flag deal risk. Gong Enrich now adds a provider waterfall for filling missing contact and account data inside Gong, but Clay remains the more flexible tool for building and researching target lists before customer conversations start.
Frequently asked questions
What are Gong Agents?
Gong Agents are the AI agents built into Gong's revenue platform. Pre-built agents such as AI Data Extractor, AI Call Reviewer, AI Deal Reviewer, AI Deal Monitor, AI Composer and AI Briefer work from the calls, emails and CRM data Gong captures, and you can also build your own on-demand or automated agents.
Can I use Gong Agents without a Gong subscription?
No. The agents run inside Gong on data it has captured. Some come with the core Gong Foundation license, and others belong to add-on applications such as Gong Engage and Gong Forecast. You can bring Gong insights into other AI tools through the Gong MCP server, but that also requires a Gong account.
How much do Gong Agents cost?
Gong does not publish prices. Its pricing page says licenses are priced per user, with a platform fee based on the number of users, and you request a custom proposal. Some AI work, including automated agents, Deep Mode, question-based AI Trackers and the MCP server, consumes Gong Credits from a shared company pool; Gong's docs say each paid core seat includes 2,000 credits a year, and more can be purchased. Manually run custom agents are included with a Gong Foundation seat.
What is Gong Agent Builder?
Agent Builder is where you create automated agents in Gong. Each agent has one trigger (a CRM record change, a date condition, or a smart tracker detecting something on a call), optional filters, and one or more actions such as generating an AI analysis, sending an internal email or Slack message, sending a brief, adding a contact to a flow, or starting a play. Creating automated agents requires the Manage automations permission. Gong's help center documents the feature, but Gong's September 2026 announcements say automatically triggered custom agents are rolling out during 2026, so confirm availability on your account.
What is Deep Mode in Gong Assistant?
Deep Mode is a research agent in Gong Assistant for broad questions, such as why deals are being lost to a competitor. It asks clarifying questions, proposes a research plan, analyzes many calls and deals, and emails you a cited report. Gong made it generally available on September 30, 2026. It is available on Gong Foundation for users with the Use Deep mode permission, and Gong shows an estimate of 1,000 to 3,000 credits before each report.
Does Gong work with Salesforce, HubSpot and Microsoft Dynamics?
Yes. Gong has native integrations with Salesforce, HubSpot and Microsoft Dynamics 365, but connects to one CRM at a time. Other CRMs can send data to Gong through its CRM API, which imports data only. AI Data Extractor can write extracted answers to the CRM fields you choose.
Does Gong support MCP?
Yes. Gong runs a read-only MCP server with tools that answer questions about an account or deal and generate briefs, for ChatGPT, Microsoft Copilot and Copilot Studio, Salesforce Agentforce and private integrations. It returns AI-generated insights rather than raw transcripts, excludes private calls, and consumes Gong Credits. Gong also has an MCP client that lets agents such as AI Briefer pull data from external MCP servers.
Do Gong Agents respect user permissions?
Mostly. Gong's help center says on-demand and automated agents use the Gong permissions of the person running them, and the MCP server's personal access mode limits each user to data they can already view. AI Data Extractor is an exception: it can analyze any non-private call or email in Gong, including calls the admin who sets it up cannot view. Gong also says customer data is not used to train generative models.
Is Gong Activate available?
Gong announced Gong Activate at Celebrate '26 on September 30, 2026 as always-on agents that evaluate every account and opportunity and guide reps and agents through winning plays. The announcement describes it in the future tense and gives no release date, so ask Gong about availability.
Integrations & fit
About Gong Agents
Gong started as conversation intelligence: it records and transcribes sales calls, captures emails, and links that activity to CRM accounts and deals. Gong now calls the whole product a Revenue AI OS, and its agents are the layer that acts on that data. The pre-built agents each do one job. AI Data Extractor answers questions you define from calls and emails (competitors mentioned, decision makers, seat counts) and can write the answers to CRM fields. AI Call Reviewer fills in call review cards for managers, and AI Trainer turns past calls into role-play practice. AI Deal Reviewer checks deals against your sales methodology, AI Deal Monitor flags risks, and AI Deal Predictor and AI Revenue Predictor score deals and the forecast. AI Composer drafts follow-up emails from call transcripts, AI Tasker suggests next actions, AI Ask Anything answers questions about a deal or account, AI Builder turns insights into assets such as scorecards, and AI Briefer, AI Theme Spotter, AI Tracker, AI Activity Mapper, AI Transcriber and AI Translator cover briefs, themes, signals, activity mapping, transcription and translation. Admins tune these in Agent Studio. On top of them, Gong added custom agents in June 2026, under a governance layer it calls the Revenue Harness. Any user can save an on-demand agent, a personal reusable prompt that runs in Gong Assistant on a call, account or deal. Agent Builder adds automated agents, which users with automation permissions create for the whole company: one trigger (a CRM record change, a date condition, or a smart tracker firing on a call), optional filters, then a sequence of actions such as generating an AI analysis, sending an internal email or Slack message, sending a brief, adding a contact to a Gong Engage flow, or starting a play. At Celebrate '26 on September 30, 2026, Gong released Deep Mode, a Gong Assistant agent that researches a broad business question across many calls and deals and returns a cited report, and Gong Enrich, which fills missing account and contact data from providers such as ZoomInfo, Apollo and Lusha. Gong also announced Gong Activate, always-on agents meant to evaluate every account and opportunity against winning plays, described in the future tense with no release date published. A read-only MCP server lets ChatGPT, Microsoft Copilot, Salesforce Agentforce and other MCP clients ask Gong about accounts and deals. The tradeoffs: the agents only work inside a Gong subscription, pricing is quote-based with per-user licenses, a platform fee and Gong Credits for heavier AI work, and although Gong's help center already documents automated agents, Gong's September 2026 announcements say automatically triggered custom agents are still rolling out during 2026.
Updates from Gong Agents
Gong released Deep Mode, an investigative agent in Gong Assistant that consumes Gong Credits, and Gong Enrich for third-party contact and account data, both generally available. Agent Builder gains automatically triggered custom agents (events, conditions, steps and actions, or described in natural language, with reusable skills), which Gong says are rolling out during 2026. Gong also announced Gong Activate, always-on agents that evaluate every account and opportunity against winning plays, without a release date.
Gong introduced the Revenue Harness, a governance and orchestration layer for its agents, and Custom Agents that RevOps leaders and managers can describe in natural language without engineering help. Gong said Custom Agents were generally available that month, with more capabilities to follow.
Are you the founder? Claim this listing →