Sometimes, software projects don't start with a business plan, a startup pitch or an ambitious product roadmap.
Sometimes, they start with a conversation.
A friend of mine works in the User-Generated Content (UGC) space. Like many creators and creative professionals, he needed a more efficient way to discover brands, identify potential opportunities, find the right contacts and manage outreach.
The process involved a lot of manual research.
Finding brands that might need creators, checking their advertising activity, looking for contact information and keeping track of conversations can quickly become time-consuming.
He approached me with an idea and asked if I could help build a solution.
I took that problem and started designing a system around it.
That became UGC Deal Engine.
The idea behind the project
The objective was to build a platform that could handle much of the repetitive work involved in finding and approaching potential UGC clients.
Instead of manually searching for opportunities across different platforms, the system would discover relevant signals, qualify potential leads and prepare personalized outreach.
The goal wasn't simply to build another email automation tool.
It was to build a workflow specifically designed around how UGC creators and creative agencies find business opportunities.
- Discovering brands with buyer intent
The first major component is the discovery engine.
UGC Deal Engine integrates Google search intelligence through Serper.dev to find opportunities across different sources, including:
- X and LinkedIn posts.
- Creator application pages.
- Google Forms.
- Notion briefs.
- TikTok-related searches.
It also integrates Apify-powered scrapers for TikTok and Meta advertising libraries.
This makes it possible to identify brands actively advertising in selected niches.
The idea is to look for signals that indicate a potential opportunity rather than simply collecting random business names.
- AI-powered opportunity qualification
Finding a potential lead doesn't automatically mean it's worth pursuing.
That's where Google Gemini comes in.
The AI qualification component examines discovered information, filters irrelevant results and assigns an intent score.
It also generates two personalized three-second video hook concepts based on the brand's product or advertising activity.
Additionally, it drafts concise outreach sequences.
Where Gemini isn't configured or its quota is exhausted, the application has a rule-based fallback mechanism.
- Contact enrichment
After qualifying an opportunity, the system attempts to identify relevant contact information.
Hunter.io integration helps discover professional email addresses associated with brand domains.
The enrichment system prioritizes roles that may be relevant to creator partnerships and marketing, including:
- Marketing.
- Growth.
- Founder.
- Social Media.
- Creative.
There are also fallback mechanisms for extracting publicly available contact information.
- Automated outreach
This is one of the most important components.
UGC Deal Engine supports a threaded three-step email outreach sequence.
It uses SMTP to send messages and IMAP to monitor the inbox.
The system checks for:
- Replies.
- Email bounces.
- Unsubscribe requests.
- Negative responses.
- Automated out-of-office messages.
If a recipient replies, the follow-up sequence stops.
If someone opts out, their email is added to a suppression list to prevent future contact.
The system also maintains email thread references to keep follow-ups connected to the original conversation.
- Engineering reliability and safety
I didn't want automation to come at the expense of reliability.
The project includes:
- SQLite with WAL mode.
- Atomic email dispatch claims to prevent duplicate sends.
- Historical contact deduplication.
- Daily sending limits.
- Business-hour restrictions.
- Randomized sending intervals.
- Dry-run mode enabled by default.
- CSRF protection.
- Brute-force login throttling.
- Secure HTTP response headers.
- CSV formula injection protection.
The dashboard also allows users to review discovered leads, edit generated hooks, approve emails and manage their pipeline.
The technology stack
The project was developed using:
- Python 3.10+
- Flask
- SQLite
- Google Gemini
- Serper.dev
- Apify
- Hunter.io
- SMTP and IMAP
- APScheduler
- Pytest
- Ruff
- Docker
The architecture separates the web dashboard from the background worker, allowing discovery and outreach tasks to run independently.
An unexpected outcome: LeadRadar
Something interesting happened while working on this project.
As I explored the problem of discovering brands with potential buying intent, I started thinking about how the same general concept could apply beyond the UGC industry.
What if you could monitor multiple platforms, identify people and businesses expressing a need, qualify those signals and surface opportunities across different industries?
That line of thinking contributed to the ideas behind LeadRadar, another project I'm developing.
It's interesting how solving one person's problem can lead you to think about entirely different opportunities.
Open source
UGC Deal Engine is available under the MIT License.
You can explore the source code, inspect the architecture, contribute or adapt it to your own workflow.
GitHub repository:
https://github.com/frankdotdev/ugc-deal-engine
Final thoughts
This project reinforced something I strongly believe about software development.
Good software doesn't always start with a completely original idea.
Sometimes, it starts with listening to someone explain a problem they're experiencing.
The developer's job is to understand that problem, think through the workflow, make sensible engineering decisions and build something useful.
That's what I set out to do with UGC Deal Engine.
And there's always another problem worth solving.
Frank Oge
Software Developer | Hackverse Technologies
https://frankoge.com
Top comments (0)