AdAlchemist

Inspiration

When you first build a startup, you are in a race against your 'runway' to understand the different personas of users who would find your product valuable and use insights derived from monitoring what they are talking about to find a scalable way to reach out to more of these users with messages that best appeal to them in a cost effective manner. We decided to focus on early stage open source software tools aimed at developers as our initial niche

What it does

We scrape data from a company site, using this to gain an understanding of;

• what problems the company is trying to solve
• who would likely find their product(s) useful
• the broader market they operate within

How we built it

• we use LangChain to interact with an LLM (OpenAI)
• we use custom tools to enable LLM to search for relevant discussions on X and to create Ads using GoogleAds

Challenges we ran into

• we struggled getting Nolita to work, we therefore had to resolve to using the X API to retrieve relevant discussions (we wanted to use X, Reddit, Hacknews and Dev.to)
• getting the information we wanted from different APIs

Accomplishments that we're proud of

• we got a working demo

What we learned

• how to use the GoogleAds API to programmatically create ads

What's next for AdAlchemist

If we find an actual user that would find this valuable (willing to pay) we would have to:

• search for relevant discussions from Reddit, HN, Dev.to, Medium
• Use the Canva API to create visuals for each ad
• Use the keyword planner using the Google Ads API to factor in keywords that have a high number of searches
• Use the GitHub API to monitor competitors (ie. what issues are people discussing about your competitors)
• Monitor analytics for campaigns using the Google Ads API and using these insights to iteratively improve (A/B test) different ad variations using AI

Built With

• google-cloud
• langchain
• x-api