In venture capital, time is an essential resource, particularly for analysts managing a constant flow of inbound deal opportunities. One major bottleneck we face is handling the sheer volume of newsletters that land in our inboxes daily. These often contain valuable information on startups, funding rounds, and market movements—but manually filtering and extracting relevant details can be both time-consuming and error-prone.
To address this, I developed a simple automation solution that speeds up the process of screening newsletters and ensures no important deal slips through the cracks. Here’s a breakdown of how I approached the problem and the tools I used:
Tools Used:
- Gmail: For email filtering and categorization.
- Zapier: Automation tool for processing and transferring data.
- GPT-4 API: For extracting key information from emails.
- Notion: A collaborative workspace to store and organize the extracted data.
The Problem: Newsletter Overload and Unstructured Information
On any given day, I was tasked with reviewing upwards of 15 different newsletters. They varied in format, relevance, and presentation, making it difficult to quickly identify which ones contained valuable startup insights. The lack of a structured approach to managing this inflow meant time was lost, and potential leads could be overlooked.
The goals were clear:
- Aggregate relevant information from multiple sources without drowning in irrelevant data.
- Create a centralized database accessible to the team.
- Keep the process cost-effective and avoid building a solution that would be underused.
The Solution: Automating Newsletter Screening
Here’s the process I implemented to solve these challenges:
- Email Filtering and Labeling (Gmail)
- Content Processing (Zapier)
- Data Extraction (GPT-4 API)
- Data Structuring (Zapier)
- Centralized Database (Notion)
The first step was to set up automated filters in Gmail. This involved creating custom labels to categorize incoming newsletters based on keywords and sender information. This helped sort the newsletters into relevant categories, ensuring that only emails with deal-related information were processed further.
Next, I leveraged Zapier to automate the flow of data from the filtered emails. Zapier acted as the central automation hub, extracting the raw email content while removing unnecessary HTML, JavaScript, and other embedded code for security reasons. This step was critical in preparing the data for further analysis.
Once the email content was cleaned up, I used the GPT-4 API to extract specific details from the text. The model was trained to recognize and pull out key information such as startup names, funding amounts, URLs, and any significant announcements. It also detected and summarized key events, such as new product launches or notable hires, providing a concise overview of each opportunity.
After extracting the relevant data, Zapier formatted the information into structured JSON files. These files were then automatically transferred to our team’s database, ensuring all the extracted details were stored in an organized and searchable format.
Finally, I set up a Notion workspace as the central repository for all the extracted data. Each JSON file was uploaded to this database, where it could be filtered, searched, and categorized based on specific criteria like funding stage, industry, or event type. This allowed the investment team to quickly access the most relevant information without wading through emails.
The cherry on top? It costs about $0.20 per month. That’s right—20 cents for a system that saves hours of manual sorting and ensures no hot lead goes unnoticed.
What’s Next?
This project was just the beginning. With this automated setup, I’m now exploring ways to refine our dealflow dashboards to better track the source of new deals and see how long opportunities are sitting in our pipeline. The goal is to make data-driven decisions about where to focus next, without letting anything slip through the cracks.
In short, this small automation project has turned the dealflow chaos into something that’s (almost) fun to manage. And all it took was a little AI magic and some creative thinking with existing tools. So, next time you’re facing a mind-numbing task that feels repetitive and wasteful, ask yourself: “Could I automate this?”
Chances are, the answer is yes.