# TL;DR If you are a B2B SaaS/AI agent developer, a deep Gong integration is now an afternoon's build with Ampersand. [Read the docs](https://docs.withampersand.com/provider-guides/gong#gong) → # Gong in the AI agent era The AI-agent era isn’t “coming” for go-to-market teams; it’s already here. Gong just shipped a fleet of purpose-built revenue agents that live inside its platform and guide reps in the moment of truth ([source](https://www.gong.io/press/gong-redefines-ai-agents-to-deliver-real-revenue-impact/?utm_source=chatgpt.com)). Today, we’re adding to that momentum and expanding AI agent use cases for Gong data across the enterprise stack. **Ampersand is now an official Gong technology partner**, complete with a deep connector (and a roadmap to bring our near-real-time Subscribe engine to Gong [events](https://help.gong.io/v1/docs/search-for-calls#common-filters))**.** What does this mean? Every customer-facing AI agent can treat Gong like a first-class, low-latency data source without having to build customer-specific auth, field mapping, rate limiting, pagination, retry logic, error handling, etc. Your AI agent shouldn’t have to jump through hoops to understand each of your customers’ worlds. Let it instantly blend Gong’s call intelligence with the rest of their stack. From CRM data to support tickets, marketing-automation signals, ad-performance insights, subscription-billing events, and beyond. All inside rock-solid, tenant-by-tenant walls. But first, let’s talk about the macro-trend. The matrimony of structure and unstructured customer data. # WHY meets WHAT: blending customer conversations with CRM precision for smarter agents When your AI agent sees only the **“what”** (closed-lost, churn risk 0.73, ARR $62k) it is basically a spreadsheet with a pulse. Add the **“why”** (the frustrated sigh in a Gong call, the Zendesk/Pylon rant about 3 am broken syncs, the three-exclamation-point email) and suddenly the Agent becomes 10x smarter. The narrative layers supply intent, tone, and subtext: why the customer is unhappy, why the deal is wobbling, why the upgrade feels ripe. The CRM rows pin those hunches to canonical IDs, timestamps, and currencies so the agent can update the proper record, trigger the correct playbook, and stay inside compliance guardrails. Practically, this means your AI CSM can ping your human CSM before the customer rage-tweets, bump license quotas when usage spikes, or nudge engineering with verbatim bug-reproduction steps. The ubiquity of AI notetakers facilitates this new paradigm — voice is a first-class data stream. # Gong conversations, where enterprise nuance lives A single Gong transcript is a gold mine of nuance: the unguarded sigh, the surprise, the pleasant rambling, the hesitation, the pricing exasperation, the half-second pause, and so much more. No rigid pick-list or checkbox combo will ever surface that rich blend of sentiment, intent, and context. Gong’s customer roster is extensive (both mid-market and enterprise), and Gong is steadily diversifying beyond its original tech-SaaS beachhead. If you are building an enterprise AI agent, tapping into the Gong ecosystem is a must! # What type of customer-facing AI agents can be built with Gong and Ampersand Here are three interesting ideas: **Reality-check agent** A live fact-checker that cross-wires Ampersand Salesforce Subscribe with Gong Read to keep sales forecasts honest. When an AE marks a deal as “Commit,” the agent sniffs the latest Gong transcript for buyer jitters, sarcasm, or dreaded “budget freeze” phrases, then scores how well the conversation reality matches the CRM hype. **Bug-triage agent** Whenever multiple transcripts repeat phrases like “keeps timing out,” the agent clusters the calls, grabs matching Zendesk tickets, and either links to or files a Jira/Linear issue that already contains repro steps clipped from the transcripts. **Renewal-risk agent** An agent listens (our Subscribe for Gong is coming soon) for negative sentiment in Gong calls, cross-checks renewal dates in Salesforce, and invoices in Stripe (or Chargebee). When risk crosses a threshold, it writes a `renewalRisk` score to the opportunity and creates an immediate follow-up task for the CSM. Across every example, the flow is the same: Gong supplies high-fidelity voice signals, Ampersand delivers them alongside structured data from the rest of the stack, and the agent performs targeted actions, no extra polling, no cross-tenant leaks, just great acceleration across the Agent's entire customer base. # Using Cursor and the Ampersand MCP to build the “Reality check agent” 🚀 Using Cursor (or any other AI IDE) and Ampersand's MCP doc-server, you fire this prompt: > A live fact-checker that cross-wires Ampersand Salesforce Subscribe with Gong Read to keep sales forecasts honest. The moment an AE marks a deal as “Commit,” the agent sniffs the latest Gong transcript for buyer jitters, sarcasm, or dreaded “budget freeze” phrases, then scores how well the conversation reality matches the CRM hype. I want an Ampersand amp.yaml for this! Your AI IDE will generate an `amp.yaml` that: - Subscribes to Salesforce Opportunity changes, specifically when StageName is set to "Commit". - Reads the latest Gong call transcript using Ampersand's link to that Opportunity (using opportunityId). - Sends the Salesforce event and the Gong call data to your backend (the “fact-checker”) for analysis. Key points for this manifest: - The [Salesforce subscribe action](https://docs.withampersand.com/subscribe-actions) should watch StageName (and optionally OwnerId). - The Gong read action should include the transcript and opportunity linkage. - Both should use the same destination so your backend can correlate the data. - When you receive a Salesforce webhook (e.g., Opportunity StageName changes to "Commit"), your backend can make an on-demand read API call to Ampersand to fetch the latest Gong call(s) associated with that Opportunity ID. - This avoids the need to store or sync all Gong calls in advance—you fetch only what you need, when you need it. ## Set up your amp.yaml ```yaml specVersion: 1.0.0 integrations: - name: salesforceOpportunities provider: salesforce subscribe: objects: - objectName: opportunity destination: factCheckerWebhook inheritFieldsAndMapping: true updateEvent: enabled: always requiredWatchFields: - StageName read: objects: - objectName: opportunity destination: factCheckerWebhook requiredFields: - fieldName: id - fieldName: StageName - fieldName: Name - fieldName: Amount - fieldName: CloseDate - fieldName: OwnerId - name: gongCalls provider: gong read: objects: - objectName: call destination: factCheckerWebhook requiredFields: - fieldName: id - fieldName: startTime - fieldName: endTime - fieldName: participants - fieldName: topics - fieldName: transcript - fieldName: opportunityId ``` ## Make an On-Demand Read API Call Use the [Ampersand API](https://docs.withampersand.com/read-actions#trigger-a-read) to trigger a read for Gong calls, filtering by opportunityId. Example API Call: ```yaml POST https://api.withampersand.com/v1/read Authorization: Bearer <YOUR_API_KEY> Content-Type: application/json { "integration": "gongCalls", "object": "call", "filters": { "opportunityId": "<THE_OPPORTUNITY_ID>" }, "sort": [ {"field": "startTime", "direction": "desc"} ], "limit": 1 } ``` This fetches the most recent Gong call for the given Opportunity. ## Process the Response Analyze the transcript as before. # Ampersand + Gong unlocks your team's most meaningful work, fast "AI isn't about replacing roles, it's about freeing people up to do their most impactful work" - Amit Bendov, CEO and Co-Founder of Gong. With the power of deep integrations across the GTM, CX, Billing stack, from Salesforce to Zendesk to Stripe, the Gong + Ampersand technology partnership can enable AI use-cases where people do only impactful and meaningful work. We cant wait to see the agents you build with us! Got a question? Jump into our [Discord](https://discord.gg/BWP4BpKHvf) or [snag time](https://calendly.com/d/cmdf-fnw-bwv/speak-with-an-engineer-at-ampersand) with one of our engineers — we'll co-build with you.
Back to articles