MarketingWala

MarketingWala offers AI-generated social posts, automatic posting, and push notifications, making social media management fast and effortless

Comment

Inspiration

Our inspiration came for this project when we saw a renowned company giving something very similar as a problem statement. we were very intrigued by the idea and were able to brainstorm all the possible applications and sectors where our solution can make a change.

What it does

MarketingWala can create as well as auto execute AI generating marketing solutions. It has 2 features AI Image and caption generation with scheduling and Push notification generation and execution.

MarketingWala can produce Ai generated social media post based on product/companies information. User can give a reference image too. The app produces multiple post from which the user can select and post directly on social networking sites through our site We can also generate push Notification and schedule them.

How we built it

We built MarketingWala using Django and React, with PostgreSQL as the backend database. For AI image generation, we integrated OpenAI’s DALL·E 3. Cloudflare’s API powers both captions and push notification text.

Some of our RESTful API endpoints are hosted on Cloudflare servers for improved performance and scalability. Additionally, we utilized Conductor for orchestration such as manage, renew OAuth 2.0 tokens and interacting with Firebase Cloud Messaging to trigger push notifications

Conductor linkes

Renew Auth token https://play.orkes.io/execution/3d67949b-8e9b-11ef-b6be-323517a1f1a2 Valid Auth token - no need to renew https://play.orkes.io/execution/042992d2-8e9f-11ef-85a8-ba73d777fd9d

We used conductor to send puch notifications to FireBase Cloud messaging.

  1. We fetch all mobile users access tokens from get resquest
  2. check if the Auth2.0 tokens are expired or not
  3. If expired generate new token
  4. Send push notification with content to Firebase viw POST http v1 request

CloudFlare

C3 for restful api hosting , @cf/meta/llama-3-8b-instruct for text generation

Challenges we ran into

  1. Had to figure out how to make custom task in conductors for regenrating access tokens
  2. Since we were interacting with multiple services we had had time maintaining the secrets and access tokens. We accidently pushed some of our access token on github public repo causing our google service account to be suspended.
  3. Had to change models to achieve better image generation

Accomplishments that we're proud of

  1. We were exposed to meta developers account for the first time and had to go through its documentation to enable Automatic posting using meta API's
  2. We created a bunch of complex workflow task in conductor for managing , securing and access tokens from users
  3. We had lot of microservices interacting with each other

What we learned

  1. Creating workflows.
  2. Hosting on Cloudflare
  3. how to create RestFull APIs
  4. Team Collaboration

What's next for MarketingWala

we plan to expand functionalities in marketwala, from just generating to editing images, video-audio-text analysis for post generation,A/B testing and profile monitoring ,Most importantly we want to go out and let other brands use our product to see how much realistic our approach is towards expanding marketwala

Share this project:

Updates