MailFlame

Inspiration & The Problem

Our inspiration comes from a conversation between two team members and a marketing agency director during a trip. As they listened to the struggles with creating personalized email content for cold lead outreach - and doing it at scale - a lightbulb went off. We knew we could leverage AI to bridge this gap, and we're ready to build a game-changer.

The Solution

Our current solution is targeting marketing agencies or sales with the need to reach out to cold leads at a large scale. During this hackathon, we built a web platform that allows the user to create a campaign, input import lead information with one click, input sales information, and generate personalized emails draft ready for outreach. On this platform, the user can also track the engagement data and status of emails and send out follow-up emails with the assistance of AI.

How we built it

Frontend Development

In the frontend development of MailFlame, we utilized Next.js and Tailwind CSS, with the entire codebase written in TypeScript. Next.js provided us with a robust framework for building React applications, enabling efficient routing, server-side rendering, and API integration. We leveraged Next.js's component-based architecture to create reusable UI components and ensure a smooth user experience. Tailwind CSS was our chosen CSS framework, offering a utility-first approach that allowed us to quickly style our components. Its extensive set of pre-defined classes enabled us to achieve consistent and responsive designs without writing custom CSS from scratch. We also took advantage of Tailwind CSS's customization capabilities to tailor the styles to match our desired visual aesthetics.

Backend Development

For the backend development of MailFlame, we utilized MongoDB, Next.js, and TypeScript to build a robust and scalable architecture. MongoDB served as our database, providing a flexible and schema-less document structure that seamlessly integrated with our TypeScript codebase. We leveraged the powerful querying capabilities of MongoDB to efficiently retrieve and store lead and campaign data. Next.js acted as our backend framework, allowing us to create server-side logic and API endpoints effortlessly. We defined RESTful API routes in Next.js to handle various functionalities such as lead creation, campaign management, and email sending. TypeScript added an extra layer of type safety and improved developer productivity, ensuring code quality and minimizing potential errors. To enhance the email generation and personalized communication features, we integrated OpenAI's ChatGPT and LangChain. Through prompt engineering techniques, we fine-tuned the interactions with these AI models to generate accurate and context-aware email drafts. PineCone was employed to cache vector embeddings, enabling faster retrieval of lead information and optimizing performance. Additionally, we utilized web scraping techniques using libraries like Cheerio to extract personalized information on leads. This allowed us to gather data from websites and incorporate it into the email templates, ensuring a personalized touch in the outreach process.

Challenges we ran into

Avoid being flagged as spam or promotion: we set an upper limit of the email being sent out today, and the user and also adjust the number through settings.

Accomplishments that we're proud of

We're proud of both the quality and quantity of what we delivered. We managed to implement the core features which are email generation and personalized communication, as well as a comprehensive platform that demonstrate the product vision.

We're also proud to build our product from a real user pain point, which makes its impact more significant. By providing a solution that streamlines processes and accelerates efficiency, we are using the power of AI to transform the way users handle email outreach. This innovative approach to addressing user challenges is what truly sets our product apart.

What's next for MailFlame

Our next step is to implement the discovery research within the lead database, transforming the experience of cold lead discovery with AI as a co-pilot.

Built With

• langchain
• mongodb
• next.js
• tailwind
• typescript