Telegram + Google Gemini: review insights you can trust
You look up reviews, open ten tabs, skim for patterns, then realize you still can’t answer the only question that matters: “So… what are customers actually saying?” By the time you’ve pulled quotes and formed an opinion, you’ve burned an hour and you still don’t fully trust it. That’s why Telegram Gemini insights are so useful.
Brand managers feel this when they’re prepping a quarterly update. Growth marketers hit it when a campaign tanks and they need fast feedback. And founders doing their own research? Honestly, you’re the one paying the opportunity cost.
This workflow turns a single Telegram message into an executive review summary powered by Google Gemini, based on real scraped review pages. You’ll see how it works, what you need, and what results to expect in the real world.
How This Automation Works
The full n8n workflow, from trigger to final output:
n8n Workflow Template: Telegram + Google Gemini: review insights you can trust
flowchart LR
subgraph sg0["Get Message Flow"]
direction LR
n0@{ icon: "mdi:brain", form: "rounded", label: "Google Gemini Chat Model", pos: "b", h: 48 }
n1@{ icon: "mdi:swap-vertical", form: "rounded", label: "Loop Over Items", pos: "b", h: 48 }
n2@{ icon: "mdi:brain", form: "rounded", label: "Google Gemini Chat Model1", pos: "b", h: 48 }
n3["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/telegram.svg' width='40' height='40' /></div><br/>Get Message"]
n4@{ icon: "mdi:swap-vertical", form: "rounded", label: "Set Search Query", pos: "b", h: 48 }
n5["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/code.svg' width='40' height='40' /></div><br/>Extract URLs"]
n6["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/code.svg' width='40' height='40' /></div><br/>Clean HTML"]
n7@{ icon: "mdi:robot", form: "rounded", label: "Analyze Sentiment", pos: "b", h: 48 }
n8["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/code.svg' width='40' height='40' /></div><br/>Merge Results"]
n9@{ icon: "mdi:robot", form: "rounded", label: "Draft Final Report", pos: "b", h: 48 }
n10["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/telegram.svg' width='40' height='40' /></div><br/>Reply to User"]
n11@{ icon: "mdi:cog", form: "rounded", label: "Google Search1", pos: "b", h: 48 }
n12@{ icon: "mdi:cog", form: "rounded", label: "Scrape Page Content1", pos: "b", h: 48 }
n6 --> n1
n3 --> n4
n5 --> n12
n8 --> n9
n11 --> n5
n1 --> n8
n1 --> n7
n4 --> n11
n7 --> n1
n9 --> n10
n12 --> n6
n0 -.-> n7
n2 -.-> n9
end
%% Styling
classDef trigger fill:#e8f5e9,stroke:#388e3c,stroke-width:2px
classDef ai fill:#e3f2fd,stroke:#1976d2,stroke-width:2px
classDef aiModel fill:#e8eaf6,stroke:#3f51b5,stroke-width:2px
classDef decision fill:#fff8e1,stroke:#f9a825,stroke-width:2px
classDef database fill:#fce4ec,stroke:#c2185b,stroke-width:2px
classDef api fill:#fff3e0,stroke:#e65100,stroke-width:2px
classDef code fill:#f3e5f5,stroke:#7b1fa2,stroke-width:2px
classDef disabled stroke-dasharray: 5 5,opacity: 0.5
class n3 trigger
class n7,n9 ai
class n0,n2 aiModel
class n5,n6,n8 code
classDef customIcon fill:none,stroke:none
class n3,n5,n6,n8,n10 customIcon
The Problem: Review research is slow, messy, and easy to misread
Manual review research looks simple until you do it under time pressure. You Google “Brand + reviews,” bounce between forums and marketplaces, and try to keep a running tally in your head. The problem is that your brain is not a database. You’ll over-weight the loudest complaint, miss quieter patterns, and accidentally cherry-pick quotes that fit what you already suspect. Then comes the write-up, which is usually the same thing every time: sentiment guess, a handful of screenshots, and “we should look into this.”
The friction compounds. And it shows up in a few predictable places.
- You spend about 2 hours reading, yet you still can’t summarize the themes with confidence.
- Key quotes get lost in tabs, so your “evidence” ends up thin when you present it to someone else.
- When you repeat this weekly, you end up repeating the same search and skim cycle instead of building a trackable view of sentiment over time.
- Small errors creep in, like mixing competitor reviews into your notes or missing context because you only read the first paragraph.
The Solution: Telegram in, executive review summary out
This n8n workflow starts with a simple trigger: you send a brand or product name to your Telegram bot. n8n takes that text, builds a search phrase like “[Name] reviews,” and runs a web search through Decodo (a scraping API that can both find pages and extract content). It collects top review URLs, fetches the page HTML, and strips out the markup so you’re left with clean text that AI can actually read. Then it loops through each page, asking a Gemini-powered AI agent to assess tone (positive, neutral, or negative) and pull out the key topics that show up in the review content. Finally, a second Gemini agent combines all those mini-analyses into a single executive summary with a sentiment score, direct quotes (“customer voice”), and an actionable verdict you can share.
The workflow begins in Telegram, moves through search and scraping, then runs a batch-style analysis pass across every review page found. The end result lands right back in Telegram as a clean report you can forward to your team in seconds.
What You Get: Automation vs. Results
| What This Workflow Automates | Results You’ll Get |
|---|---|
|
|
Example: What This Looks Like
Say you’re reviewing a product before a landing page refresh. You usually check about 12 review pages, spend maybe 8 minutes skimming each, then another 30 minutes pulling quotes and writing a summary. That’s roughly 2 hours, and it’s easy to miss patterns. With this workflow, you send “XQ Pharma” in Telegram (10 seconds), Decodo gathers and scrapes the pages, and Gemini produces a consolidated executive summary after the batch analysis finishes (often about 15 minutes). You get your morning back.
What You’ll Need
- n8n instance (try n8n Cloud free)
- Self-hosting option if you prefer (Hostinger works well)
- Telegram for sending the brand name trigger
- Decodo to search and scrape review pages
- Google Gemini API key (get it from Google AI Studio / Gemini API console)
Skill level: Intermediate. You’ll connect accounts, paste API keys, and tweak prompts or queries without touching heavy code.
Don’t want to set this up yourself? Talk to an automation expert (free 15-minute consultation).
How It Works
Telegram message trigger. You send a product or company name to your bot, and n8n captures that text instantly using the Telegram trigger.
Search phrase and web search. n8n builds a consistent query (like “Brand reviews”) and runs it through Decodo to pull back a list of top URLs worth checking.
Scrape, clean, then batch analysis. Each URL is fetched, the HTML is stripped down to readable text, and the workflow loops through sources in batches so the AI agent can score sentiment and extract recurring topics without choking on huge payloads.
Executive summary delivery. A second Gemini agent consolidates everything into a sentiment score, key themes, quotes, and an actionable verdict, then sends the finished report back to Telegram.
You can easily modify the search query to target specific sites (like Reddit) or change the output destination to email or a spreadsheet based on your needs. See the full implementation guide below for customization options.
Step-by-Step Implementation Guide
Step 1: Configure the Telegram Trigger
Set up the workflow to start whenever a Telegram message is received.
- Add and open Incoming Telegram Update.
- Set Updates to
message. - Credential Required: Connect your Telegram credentials (add a Telegram bot token) because Incoming Telegram Update needs access to your bot.
Step 2: Build the Search Query and Run the Web Search
Create a Google search phrase from the incoming Telegram message and execute the search.
- Open Build Search Phrase and set the query value to
={{ $json.message.text }} reviews. - Open Run Web Search and set Operation to
google_search. - Set Query to
={{ $json.query }}. - Credential Required: Connect your decodoApi credentials in Run Web Search.
Step 3: Extract Links and Fetch Page Content
Collect target URLs, fetch HTML content, and clean the text for analysis.
- Open Collect Link Targets and keep the provided JavaScript for recursive URL extraction.
- Open Fetch Page HTML and set URL to
={{ $json.url }}. - Credential Required: Connect your decodoApi credentials in Fetch Page HTML.
- Open Strip Markup Text and keep Mode set to
runOnceForEachItemto clean each page’s HTML.
Step 4: Analyze Sentiment and Aggregate Results
Analyze each cleaned review for sentiment, then combine all AI outputs into a single report.
- Open Assess Review Tone and confirm the Text prompt includes
{{ $json.text_clean }}. - Ensure Gemini Chat Engine A is connected as the language model for Assess Review Tone; add credentials to Gemini Chat Engine A (not the agent node).
- Open Batch Through Links to keep default batching behavior for iterating over results.
- Open Combine Findings and keep the JavaScript that concatenates AI outputs into
aggregated_report.
Step 5: Summarize and Send the Telegram Report
Generate a concise Telegram-ready summary and send it back to the original chat.
- Open Compose Summary Brief and confirm the prompt references
{{ $json.aggregated_report }}. - Ensure Gemini Chat Engine B is connected as the language model for Compose Summary Brief; add credentials to Gemini Chat Engine B (not the agent node).
- Open Send Telegram Report and set Text to
={{ $json.output }}. - Set Chat ID to
={{ $('Incoming Telegram Update').first().json.message.chat.id }}to reply in the same chat. - Credential Required: Connect your Telegram credentials in Send Telegram Report.
Step 6: Test & Activate Your Workflow
Verify the workflow end-to-end and then turn it on for production use.
- Click Execute Workflow and send a test message to your Telegram bot.
- Confirm Run Web Search returns results, Strip Markup Text outputs
text_clean, and Assess Review Tone produces a JSON sentiment response. - Verify Compose Summary Brief returns a concise summary and Send Telegram Report posts it back to the same chat.
- Switch the workflow to Active to enable live processing.
Common Gotchas
- Telegram bot credentials can expire or the bot can be blocked from reading messages in certain chat setups. If things break, check your bot token and chat permissions in the Telegram Bot settings first.
- If you’re using Wait nodes or external scraping, processing times vary. Bump up the wait duration if downstream nodes fail on empty responses.
- Google Gemini prompts that are too generic will give you vague summaries. Add your preferred tone and what you consider “actionable” inside the AI Agent system prompt early, or you will keep editing the same report forever.
Frequently Asked Questions
About 30 minutes if your API keys are ready.
No. You’ll connect Telegram, Decodo, and Gemini, then adjust a few fields and prompts.
Yes. n8n has a free self-hosted option and a free trial on n8n Cloud. Cloud plans start at $20/month for higher volume. You’ll also need to factor in Decodo scraping plan costs and Gemini API usage.
Two options: n8n Cloud (managed, easiest setup) or self-hosting on a VPS. For self-hosting, Hostinger VPS is affordable and handles n8n well. Self-hosting gives you unlimited executions but requires basic server management.
Yes, and it’s basically the same flow. You just send the competitor name to Telegram, then tweak the “Build Search Phrase” step to target what you care about (for example, “pricing complaints” or “alternatives”). Many teams also edit the AI Agent prompt in “Assess Review Tone” to focus on switching triggers, deal-breakers, and feature gaps, then adjust “Compose Summary Brief” to output a comparison table-style verdict.
Usually it’s the bot token or chat permissions. Regenerate your Telegram bot token if needed, confirm the bot is added correctly to the chat, and make sure you’re testing by sending a plain text message (not a forwarded post with weird formatting). If it works in some chats but not others, it’s often a privacy setting or group permission issue.
Practically, it depends on how many URLs you scrape and your plan limits, but most teams comfortably run batches of about 10–30 pages per request cycle.
It can be, mainly because this workflow relies on looping through batches, cleaning scraped HTML, and running multi-step AI analysis. n8n handles branching and batching without getting awkward or expensive as the logic grows. Zapier and Make can still work for simpler review alerts or single-page summaries, but they tend to feel cramped once you need “search → scrape many pages → analyze each → merge results.” If you’re unsure, Talk to an automation expert and explain what “done” looks like for your team.
Once this is running, review research stops being a recurring chore and becomes something you can do on demand. The workflow handles the repetitive reading. You make the call.
Need Help Setting This Up?
Our automation experts can build and customize this workflow for your specific needs. Free 15-minute consultation—no commitment required.
