Skribby

Practical examples for common meeting intelligence workflows using the Skribby TypeScript SDK.

AI Meeting Summary

The most common use case: waiting for a meeting to end, then sending the transcript to an LLM like GPT-4.

import { createClient } from '@skribby/sdk';
import OpenAI from 'openai';

const skribby = createClient({ api_key: process.env.SKRIBBY_KEY });
const openai = new OpenAI();

// This would typically be inside your Webhook handler
async function handleWebhook(payload: any) {
    if (
        payload.type === 'status_update' &&
        payload.data.new_status === 'finished'
    ) {
        // 1. Fetch the full bot data
        const bot = await skribby.getBotById(payload.bot_id);

        // 2. Extract the transcript text
        const fullTranscript = bot.data.transcript
            .map((s) => `${s.speaker_name}: ${s.transcript}`)
            .join('\n');

        // 3. Send to AI
        const response = await openai.chat.completions.create({
            model: 'gpt-5.2',
            messages: [
                {
                    role: 'user',
                    content: `Summarize this meeting into 5 bullet points:\n\n${fullTranscript}`,
                },
            ],
        });

        console.log('Summary:', response.choices[0].message.content);
    }
}

Real-time "Who is Speaking" UI

Build a live dashboard that highlights the current speaker using WebSockets.

const bot = await skribby.getBotById('bot_uuid');
const realtime = bot.getRealtimeClient();

realtime.on('started-speaking', (data) => {
    console.log(`${data.participantName} is talking...`);
    // Update your UI state: setHighlightedSpeaker(data.participantId)
});

realtime.on('stopped-speaking', (data) => {
    console.log(`${data.participantName} stopped.`);
    // Update your UI state: clearHighlightedSpeaker()
});

await realtime.connect();

Export Recording to S3

Move your meeting assets to your own cloud storage as soon as they are ready.

import { S3Client, PutObjectCommand } from '@aws-sdk/client-s3';
import axios from 'axios';

const s3 = new S3Client({});

async function exportToS3(botId: string) {
    const bot = await skribby.getBotById(botId);

    if (bot.data.recording_url) {
        // Download the WebM file from Skribby
        const response = await axios.get(bot.data.recording_url, {
            responseType: 'arraybuffer',
        });

        // Upload to your S3 bucket
        await s3.send(
            new PutObjectCommand({
                Bucket: 'my-meeting-recordings',
                Key: `${botId}.webm`,
                Body: response.data,
                ContentType: 'video/webm',
            }),
        );

        console.log('Export complete!');

        // Optional: Delete from Skribby to save on storage/privacy
        await bot.delete();
    }
}

Live Sentiment Alerts

Analyze the live transcript stream for specific keywords or sentiment.

realtime.on('ts', (segment) => {
    const urgentKeywords = ['cancel', 'unhappy', 'refund', 'manager'];

    const isUrgent = urgentKeywords.some((word) =>
        segment.transcript.toLowerCase().includes(word),
    );

    if (isUrgent) {
        sendSlackAlert(
            `Urgent keyword detected in live meeting! "${segment.transcript}"`,
        );
    }
});