Built My Own Chat Summarizer with Gemini Flash — Because Who Has Time for 150+ Messages? 😅📲
🚀 𝗧𝗵𝗶𝘀 𝗪𝗲𝗲𝗸 𝗜 𝗕𝘂𝗶𝗹𝘁 𝗦𝗼𝗺𝗲𝘁𝗵𝗶𝗻𝗴 𝗖𝗼𝗼𝗹 𝘄𝗶𝘁𝗵 𝗚𝗲𝗺𝗶𝗻𝗶 + 𝗟𝗟𝗠𝘀!
You know that feeling when you return to your phone after a busy day and find 150+ messages on a WhatsApp group planning a trip? You don’t have the time (or patience 😅) to scroll through every message, but you really need to know and enthusiastic what was finalized?
✨ 𝘛𝘩𝘢𝘵’𝘴 𝘸𝘩𝘦𝘳𝘦 𝘐 𝘥𝘦𝘤𝘪𝘥𝘦𝘥 𝘵𝘰 𝘢𝘱𝘱𝘭𝘺 𝘈𝘐.
This week, I tried building a chat summarization tool using Google’s 𝗚𝗲𝗺𝗶𝗻𝗶 𝗙𝗹𝗮𝘀𝗵 (𝗴𝗲𝗺𝗶𝗻𝗶-𝟮.𝟬-𝗳𝗹𝗮𝘀𝗵-𝗹𝗶𝘁𝗲-𝟬𝟬𝟭) model — right from my local laptop.
🔍 𝗪𝗵𝗮𝘁 𝗱𝗼𝗲𝘀 𝗶𝘁 𝗱𝗼?
Whenever there’s a long chat (e.g., WhatsApp, Teams, customer support logs), just hit “Summarize Chat” — and boom, it pulls out the key points, actions taken, and resolutions. It’s like having your own Copilot for conversations. 🧠
🛠️ 𝗧𝗲𝗰𝗵 𝘀𝘁𝗮𝗰𝗸:
- Python-based FastAPI to host the summarization endpoint
- Gemini Flash for fast, efficient summarization — Model I have chosen but can use any other LLM in market at free of cost
- Deployed locally for quick prototyping
💻 𝗖𝗼𝗱𝗲 & 𝗦𝗲𝘁𝘂𝗽 𝗚𝘂𝗶𝗱𝗲:
I’ve made everything public — including step-by-step documentation on how to:
- Create a Google Cloud project
- Enable Vertex AI API
- Use Gemini for summarization
💡 𝗠𝗼𝗿𝗲 𝗶𝗱𝗲𝗮𝘀 / 𝗨𝘀𝗲 𝗖𝗮𝘀𝗲𝘀 𝘆𝗼𝘂 𝗰𝗮𝗻 𝘂𝘀𝗲 𝘁𝗵𝗶𝘀 𝘁𝗼𝗼𝗹 𝗳𝗼𝗿:
🧳 Trip planning groups (yes, I actually used it for that!)- Screenshot provided
💬 Customer service chat summaries (past 24 hrs or across agents) — Improves customer satisfaction and turn frustrated 😡 to friendly 🥰customers
🏢 Internal team discussions or meeting logs — Helpful if you are doing multi tasking and missed some conversations🤹♀️
🧾 Long email threads — Quickly respond or take actions of what is going on with the mails ⚡
📚 Student group chats summarizing assignments — Gets you some hints on prioritized tasks.💯
𝘼𝙄 𝙙𝙤𝙚𝙨𝙣’𝙩 𝙝𝙖𝙫𝙚 𝙩𝙤 𝙗𝙚 𝙘𝙤𝙢𝙥𝙡𝙚𝙭. 𝙎𝙤𝙢𝙚𝙩𝙞𝙢𝙚𝙨, 𝙞𝙩 𝙟𝙪𝙨𝙩 𝙣𝙚𝙚𝙙𝙨 𝙩𝙤 𝙨𝙖𝙫𝙚 𝙮𝙤𝙪 𝙩𝙞𝙢𝙚 🕒.
