🤖 Promo seeker: Auto-find promo codes with SerpAPI, Gemini & Telegram
⚡ 359 views · 🤖 AI Chatbots & Agents
💡 Pro Tip — HTTP Request scraping tends to break when sites update their markup. If you’re scraping a major platform, check if ScraperNode covers it — it has maintained scrapers for LinkedIn, Instagram, TikTok, YouTube, and 20+ other platforms that return structured data.
Description
Promo Seeker automatically finds, verifies, and delivers active promo codes to users via Telegram or email using SerpAPI + Gemini (OpenRouter). Saves hours of manual searching and deduplicates results into an n8n Data Table for fast reuse.
Why Use This Workflow?
Time Savings: Reduces manual promo hunting from ~2 hours to ~5 minutes per query.
Cost Reduction: Cuts reliance on paid scraping tools or manual services — potential savings of $50–$200/month.
Error Prevention: Cross-references sources and enforces a 30‑day recency filter to reduce expired-code hits by ~60% vs single-source checks.
Scalability: Handles hundreds of queries per day with Data Table upserts and optional scheduling for continuous discovery.
Ideal For
- Marketing / Growth Managers: Quickly discover competitor or partner discounts to promote in campaigns.
- Customer Support / Operations: Respond to user requests with verified promo codes via Telegram or email.
- Affiliate / Content Teams: Aggregate and maintain a clean promo feed for newsletters or site widgets.
How It Works
- Trigger: Incoming request via Webhook, Telegram message, Google Form submission, or scheduled run.
- Data Collection: The LangChain agent uses SerpAPI search results and Gemini (OpenRouter) to locate recent promo codes.
- Processing: Filter for recency (last 30 days), extract code, value, terms, and expiry.
- Intelligence Layer: Gemini 2.5 Pro (OpenRouter) + LangChain agent structure and verify results, outputting a standardized JSON.
- Output & Delivery: If a code exists in the Data Table, notify the requester via Telegram and Gmail; otherwise, return results and upsert them into the Data Table.
- Storage & Logging: Results stored/upserted in an n8n Data Table to prevent duplicates and enable fast lookups.
Setup Guide
Prerequisites
| Requirement | Type | Purpose |
|---|---|---|
| n8n instance | Essential | Execute the workflow — import JSON to your n8n instance |
| SerpAPI account | Essential | Web search results for promo code discovery |
| OpenRouter (Gemini) account | Essential | Language model (Gemini 2.5 Pro) for extraction and verification |
| Telegram Bot (BotFather) | Essential | Receive queries and send promo notifications |
| Gmail OAuth2 | Essential | Send rich email notifications |
| n8n Data Tables | Essential | Store and deduplicate promo code records |
Get your n8n instance here: n8n instance — import the JSON and begin configuration. (Repeat link for convenience) n8n instance
Installation Steps
- Import the JSON workflow into your n8n instance.
- Configure credentials (use n8n’s Credentials UI — do NOT paste keys into nodes):
- SerpAPI: paste your SerpAPI API Key from your SerpAPI dashboard.
- OpenRouter API: paste your OpenRouter API Key (for Gemini 2.5 Pro).
- Telegram API: create bot via @BotFather, then add Bot Token to Telegram credentials.
- Gmail OAuth2: use n8n’s Gmail OAuth2 credential flow and authorize the account.
- Update environment-specific values:
- Webhook path: /v1/promo-seeker (configured in the Webhook node) or replace with your preferred path.
- Data Table ID: point to your own Data Table
- Form/webhook recipient fields (email/chatId mappings).
- Customize settings:
- Adjust the LangChain agent prompt (Promo Seeker Agent) for different recency windows or regional focus.
- Change max results/limit on the Data Table node (default limit = 3).
- Test execution:
- Trigger via Telegram or a POST to the webhook with sample payload:
{ "platform":"example.com", "email":"you@domain.com" } - Confirm notifications arrive and Data Table rows are upserted.
- Trigger via Telegram or a POST to the webhook with sample payload:
Technical Details
Core Nodes
| Node | Purpose | Key Configuration |
|---|---|---|
| SerpAPI | Fetch web search results | Provide credential; adjust search params in agent prompt |
| Gemini 2.5Pro (OpenRouter) | Extract & verify promo details | Use OpenRouter credential; model google/gemini-2.5-pro |
| Promo Seeker Agent (LangChain) | Orchestrates search + parsing | System prompt enforces 30‑day recency & result format |
| Structured Output Parser | Validates agent output | JSON schema example for platform/code/value/terms/validUntil |
| Data Table (Get row(s)) | Lookup existing promos | Filters by platform; limit = 3 |
| If (Code Exist?) | Branching logic | Checks existence of platform field |
| Data Table (Upsert row(s)) | Insert or update promo | Mapping from agent output to Data Table columns |
| Telegram Trigger / Telegram | Receive queries & notify users | Webhook-based trigger; parse_mode = HTML for messages |
| Gmail | Send rich HTML emails | Uses Gmail OAuth2 credential |
| Webhook / Form Trigger | Alternate inputs | Webhook path /v1/promo-seeker and Form trigger for manual submissions |
Workflow Logic
- On trigger, the Platform Set node normalizes the incoming query and receiver.
- Get row(s) checks Data Table for existing promos.
- If found: notify via Telegram and send Gmail (email template included).
- If not found: the Promo Seeker Agent runs SerpAPI searches, parses structured output, then Upsert row(s) saves results and notifications are sent.
- Structured Output Parser enforces correct JSON to avoid bad upserts.
Customization Options
Basic Adjustments
- Recency window: change the agent prompt to 7/14/30 days.
- Result limit: increase Data Table Get limit or change Upsert batch size.
Advanced Enhancements
- Add Slack or Microsoft Teams notifications (moderate complexity).
- Add caching layer (Redis) to reduce repeated SerpAPI calls (advanced).
- Parallelize searches across multiple search engines (higher API usage & complexity).
Performance & Optimization
| Metric | Expected Performance | Optimization Tips |
|---|---|---|
| Execution time | 8–30s per new search (depends on SerpAPI + LM response) | Reduce SerpAPI page depth; cache recent results |
| API calls | 3–10 SerpAPI calls per complex query | Batch queries; use higher-quality search params |
| Error handling | Agent retries on malformed output | Use retry nodes and set onError strategy for downstream nodes |
Troubleshooting
| Problem | Cause | Solution |
|---|---|---|
| No results returned | Query too vague or rate-limited API | Improve query specificity; check SerpAPI quota |
| Gmail send fails | OAuth scope not granted or token expired | Reconnect Gmail OAuth2 credential in n8n |
| Telegram webhook not firing | Incorrect bot token or webhook setup | Recreate Telegram credential and check bot permissions |
| Duplicate rows | Upsert mapping mismatch | Ensure promoCode mapping in Upsert matches structured output |
| Agent returns malformed JSON | LM prompt too permissive | Tighten the agent system prompt and validate with Structured Output Parser |
Created by: khaisa Studio Category: Marketing Automation
Tags: promo-codes, coupons, serpapi, telegram, gmail, openrouter, data-tables
Need custom workflows or help adapting this template? Contact us
🔗 Nodes Used
Webhook, Telegram, Telegram Trigger, Gmail, Schedule Trigger, AI Agent
📥 Import
Download workflow.json and import into n8n:
Workflow menu → Import from File