Inspiration
The idea for peek.ai was inspired by the need to stand out in the competitive professional landscape. Traditional LinkedIn profiles can be static and text-heavy, making it challenging to capture the attention of recruiters and potential collaborators. We wanted to create a tool that leverages the power of video to present professional profiles in a more dynamic and engaging way.
What it does
peek.ai is a web platform with a companion Chrome extension that transforms your LinkedIn profile into an engaging video summary. It uses supporting AI features to extract key details from your profile and generates a concise, visually appealing video that highlights your skills, experience, and accomplishments. The generated videos come with shareable links, making it easy to distribute across various platforms.
The feature of highlight apart from the video is that you get a preview of your generated summary as a dynamic banner image when the shareable link is sent over platforms like WhatsApp so the receiver gets to see this digest of information even before clicking the link.
How we built it
We built peek.ai with Google's Gemini AI and Canva Connect APIs to solve the pivotal requirements of the project. The Chrome extension integrates seamlessly in order to analyze LinkedIn profile data. We developed custom algorithms/logic to identify and summarize the most important information, and used Canva's Autofill APIs for video and social preview generation to create visually appealing content. Our team focused on ensuring the extension and web dashboards are user-friendly to generate video stories.
Challenges we ran into
- Explored Canva Connect APIs and Autofill jobs for the first time and built various POCs to test each functionality in isolation.
- One of the biggest challenges was ensuring the AI accurately identifies and summarizes key profile details.
- Chrome extension scripts work differently in various contexts (background_script, content_scripts, popup, etc.), requiring planned and effective message passing among them.
- Tackled the lazy loading constraint of LinkedIn to retrieve all necessary data.
- Connect user profile using LinkedIn OAuth to ensure safety to their profile data.
- We built guardrails and schema checks to ensure that only the most reliable and accurate data generated using AI with LLMs is saved to the database.
- We also had to implement queueing for AI job requests and Autofill jobs to prevent rate limits and prevent from holding the user's server request.
Accomplishments that we're proud of
The positive feedback from our beta testers, who found the tool easy to use and effective in enhancing their professional presence, is a major accomplishment. Most of all, we are happy to convey that we were able to build the entire product at a significantly accelerated pace considering its underlying design and robust nature.
What we learned
The development of peek.ai has been a sprint of various technical and non-technical learnings. On the technical front, we got to experience and plan out how the behind-the-scenes operations must be automated to handle user workloads with Canva Connect APIs for image/video generations and Google Gemini APIs to make AI inferences. On the other hand, we got the experience of interacting with some test users and their LinkedIn profiles to ensure that our systems are robust enough to handle a variety of profiles where we also could spot and handle certain edge cases.
What's next for peek.ai
Looking ahead, we plan to enhance peek.ai by introducing more customization options for users, such as different video styles and templates right from their Canva editors. We are also exploring the integration of additional features like voice-overs and interactive elements in the video summaries. In addition, peek.ai would have the following in the coming times:
- Custom shortened peek URLs
- Rich analytics of user interactions
- Migration to other LinkedIn APIs to reduce synchronisation times.
Our goal is to continuously improve peek.ai to make it an indispensable tool for professionals seeking to make a memorable impact.
Built With
- azure
- canva-connect-api
- chrome
- cloudflare
- docker
- figma
- git
- google-gemini-ai
- next.js
- node.js
- postman
- tailwind
- typescript



Log in or sign up for Devpost to join the conversation.