Overview

We've worked with Rep'd for more 7 years through many pivots and iterations to arrive today at an AI-assisted gov-tech video platform that is scaling successfully in the market.

Rep'd is now both a city government software provider for residents to ask city officials questions directly and get a video answer from them as well as a civic engagement platform that allows voters to ask  candidates, constituents and elected officials questions about their campaigns and why they should vote for them.

For many voters, participation in an election goes beyond voting to include wanting to directly connect with candidates and contribute their time and/or money to a campaign.

Rep’d’s Co-Founders (and brothers), Mike and Dave Baumwoll, identified this trend and had the vision of a platform to empower transparent democracy and create a more direct connection between voters and candidates.

V2 Challenge‍

In V2, the goal was to add AI-driven features and multilingual support to meet the needs of a broader, more diverse user base. The challenge here involved integrating AI components for real-time video transcription and automated response assistance, requiring seamless collaboration between AWS and OpenAI tools. Adding Polimorphic integration for local government queries introduced new data privacy and API rate-limit considerations. Furthermore, dynamic UI translation for Spanish and Creole meant maintaining both accuracy and efficiency without impacting load times, critical for a platform anticipating high user engagement across multiple languages.

‍

V1 Challenge‍

For the first version of Rep’d, Mike and Dave envisioned a platform that would facilitate direct voter engagement by allowing voters to submit and upvote policy questions to candidates, receive email notifications when their questions were answered, and easily contribute to campaigns. The platform also needed to provide candidates with tools for analyzing voter data and trends. This required implementing a white-label architecture with a candidate-side web portal that included secure login, management oversight, and platform analytics, all within the constraints of a tight budget. Balancing performance, scalability, and cost while maintaining rapid deployment for new clients was the primary challenge.

The first step was to create detailed wireframes. Rep’d’s designer initially provided a design but it required changes over the years for which we stepped in to make updates.

Then our development team analyzed each task in Jira, wrote the API document and data schema prior to going into code.

V2 - 2024

In V2, we expanded Rep’d’s functionality with AI-based features and multilingual support, tailored to meet the needs of a more diverse user base.

AI Video Transcription: AWS-based AI transcription generates subtitles, improving accessibility.

Multilingual UI: Dynamic and static translations in Spanish and Creole further expand reach.

Polimorphic Integration: Provides authoritative responses to user questions on local governance.

AI-Assisted Complexity Analysis: OpenAI-powered analysis matches users with video responses based on question complexity.

Teleprompter Note Generator: A custom OpenAI integration creates notes based on user questions, helping city officials respond effectively to resident inquiries.

‍

V1 - 2023

Anticipating a broad audience for Rep’d, we focused on both performance and server cost management. To achieve a lightweight structure, we avoided heavy frameworks like Angular or React, opting instead for preprocessors. This streamlined approach allowed rapid deployments, especially important in a white-label environment where onboarding new clients quickly to their sub-domain is key.

With a limited budget, our team provided project management hours and several additional features at no extra cost. Mike and Dave, Rep’d’s co-founders and long-time collaborators, were patient and supportive throughout. As shareholders in Rep’d, we were committed to adding value, despite tight resources.

‍

Key features for V1 included:

Advanced Analytics in DataStudio: Enhanced analytics for each white-label client, facilitated by a custom stats database that syncs with Google Analytics and enables direct DataStudio queries.‍

Custom Teleprompter Mode: We developed a teleprompter feature within the admin panel to allow live recording. A Safari compatibility issue required us to convert recorded videos from .webm to .mp4 via the API.‍

NYC Candidates Page & Universal Login: Users could find their voter district information through an integrated government API, and Universal Login enabled seamless access across all candidate pages.‍

NGP VAN Integration: Added for advanced voter data analytics, providing access to an extensive database for tracking grassroots voter trends.

‍

Initial Product Version - 2017

Mike and Dave had worked with us on an earlier start-up back in 2017 called Minque (Minute-Questions), a question and answer platform dedicated to connecting curious minds with educated video answers. For Minque, we had written a custom video player and video feed similar to Tiktok but where the creators were professors, for iOS and uploaded and stored the user-generated videos on AWS while keeping the backend API separate on Heroku for easy control over speed of scaling.

At the starting gate Rep’d signed 11 candidates and counting for a variety of political races throughout the country and raised investment to build out the platform. Rep’d is scaling fast with a very flexible product/market fit for candidates and major organizations.
‍

After pivoting to local government in 2023, it got better, with 31 local cities and municipalities pitched and 13 live.
‍

As of October 2024, Rep’d is entering the next phase of its growth.

They have been onboarding a new customer roughly every week, unlocking new market segments through their just-launched AI product, and are seeing 7x ROI from marketing efforts.