Turn Meeting Transcripts into Shared Spaces
After a call, sales reps spend 15 to 30 minutes manually reviewing notes, finding relevant content, and building a follow-up Shared Space to send the buyer.
If your reps capture meeting notes in Showpad, Field Seller Agent already automates that step: AI summaries, email drafts, and CRM updates, without any custom build.
But what if your team records full calls through an external platform like Gong, Chorus, or Fathom? The transcript lives outside Showpad, so how do you get the same post-call automation? You can build a pipeline to connect them! When a call ends, your custom pipeline pulls the transcript, extracts the buyer's key themes, finds matching Showpad assets, and creates a ready-to-send Shared Space.
Classic Situation
A sales team uses an external call recording platform like Gong and wants to use those transcripts to create follow-up Shared Spaces in Showpad automatically. Without a custom integration, the team runs into:
- Inconsistent content selection across reps
- Delays between the call and the follow-up
- Missed opportunities to address specific pain points the buyer raised
- Reps spending time on curation instead of selling
Gong Example
Subscribe to call completion events from your conversation intelligence platform, fetch the transcript, send it to your designated LLM to extract buyer themes, query the Showpad API for matching assets, and create a populated Shared Space automatically. The example below uses Gong.
- Plan: Ultimate | Advanced or Expert
- Permissions: Administrator access to Showpad's Admin App
- Config:
- An LLM API key (examples below use the Anthropic API)
- An
ANTHROPIC_MODELenvironment variable set to the Claude model you want to use (see Claude models) - Access to your conversation intelligence platform's transcript API (this example uses Gong's access key and secret)
- Familiarity with Showpad Webhooks and the Showpad API
While this example uses Gong, you can adapt the webhook payload and transcript fetch to match your platform's API.
Steps
1. Receive the Call Transcript from Gong
Gong fires a webhook when a call finishes processing. Set up an endpoint to receive the event, then call the Gong transcript API to fetch the full transcript.
// Receive Gong call-completed webhook
app.post('/webhooks/gong', async (req, res) => {
res.status(200).send('OK');
const callId = req.body.payload?.callId;
if (callId) {
await processCallTranscript(callId);
}
});
// Fetch the transcript from the Gong API
async function fetchGongTranscript(callId) {
const credentials = Buffer.from(`${process.env.GONG_ACCESS_KEY}:${process.env.GONG_ACCESS_KEY_SECRET}`).toString(
'base64'
);
const response = await fetch('https://api.gong.io/v2/calls/transcript', {
method: 'POST',
headers: {
Authorization: `Basic ${credentials}`,
'Content-Type': 'application/json',
},
body: JSON.stringify({ filter: { callIds: [callId] } }),
});
const data = await response.json();
const monologues = data.callTranscripts?.[0]?.transcript ?? [];
// Flatten speaker monologues into a single text string
return monologues.flatMap((m) => m.sentences.map((s) => s.text)).join(' ');
}
2. Extract Buyer Themes with AI
Send the transcript to your designated LLM to identify the buyer's pain points, questions, and interests. Ask for structured output so the results are easy to work with.
The transcript content comes from an external platform and is interpolated directly into the prompt. A crafted transcript could attempt to override your instructions. In production, consider validating or sanitizing transcript text before interpolation, or use a system prompt to clearly delimit untrusted input from your instructions.
import Anthropic from '@anthropic-ai/sdk';
const anthropic = new Anthropic();
async function extractBuyerThemes(transcript) {
const response = await anthropic.messages.create({
model: process.env.ANTHROPIC_MODEL, // Any current Claude model ID
max_tokens: 1024,
messages: [
{
role: 'user',
content: `Analyze this sales call transcript and extract the buyer's key themes.
Return a JSON object with:
- "pain_points": array of specific problems or challenges the buyer mentioned
- "interests": array of product capabilities or features they asked about
- "search_queries": array of 3-5 short search terms to find relevant content (suitable for a content library search)
Transcript:
${transcript}
Return only the JSON object, no explanation.`,
},
],
});
const text = response.content.find((block) => block.type === 'text').text;
return JSON.parse(text);
}
3. Find Matching Showpad Content
Use the extracted search terms to query the Showpad API and collect relevant assets.
async function findMatchingAssets(searchQueries, showpadToken) {
const assetIds = new Set();
for (const query of searchQueries) {
const response = await fetch(
`https://{subdomain}.showpad.biz/api/v4/assets?name=${encodeURIComponent(query)}&limit=5`,
{
headers: { Authorization: `Bearer ${showpadToken}` },
}
);
const data = await response.json();
for (const asset of data.items) {
assetIds.add(asset.id);
}
}
return [...assetIds];
}
4. Create the Shared Space
Use the Showpad API to create a new space and populate it with the matched assets.
async function createSharedSpace(title, assetIds, showpadToken) {
// Create the Shared Space
const spaceResponse = await fetch('https://{subdomain}.showpad.biz/api/v4/shared-spaces', {
method: 'POST',
headers: {
Authorization: `Bearer ${showpadToken}`,
'Content-Type': 'application/json',
},
body: JSON.stringify({ title }),
});
const space = await spaceResponse.json();
const spaceId = space.id;
// Add the matched assets
await fetch(`https://{subdomain}.showpad.biz/api/v4/shared-spaces/${spaceId}/items`, {
method: 'POST',
headers: {
Authorization: `Bearer ${showpadToken}`,
'Content-Type': 'application/json',
},
body: JSON.stringify({
items: assetIds.map((id) => ({ type: 'asset', assetId: id })),
}),
});
return space;
}
5. Tie It Together
Combine the steps into a single pipeline function.
async function processCallTranscript(callId) {
const showpadToken = process.env.SHOWPAD_API_TOKEN;
// Step 1: Fetch and flatten the Gong transcript
const transcript = await fetchGongTranscript(callId);
// Step 2: Extract themes
const themes = await extractBuyerThemes(transcript);
console.log('Extracted themes:', themes);
// Step 3: Find matching content
const assetIds = await findMatchingAssets(themes.search_queries, showpadToken);
console.log(`Found ${assetIds.length} matching assets`);
// Step 4: Create the Shared Space
const space = await createSharedSpace('Follow-up resources', assetIds, showpadToken);
console.log('Shared Space created:', space.publicUrl);
// Notify the rep via your preferred channel (Slack, CRM, email)
}
Ask the LLM to limit search queries to 3 to 5 terms. More queries increase API calls without meaningfully improving content coverage, and may pull in less relevant assets.
To improve relevance over time, track which Shared Space assets the buyer actually views and feed that signal back into
your content selection logic. The Content Reporting API records these views as
sharedspace-viewed-asset events.
Was this page helpful?