AI election chatbot and a fully AI-generated launch film

Retrieval-augmented chatbot for university elections plus a 24-hour AI-only launch commercial.

An education-focused company built and later exited by the SimpleAI founder. Operated at the intersection of higher education and student life, with a focus on connecting students to decisions that shape their university experience.

Problem

University elections have a structural information problem. Every year a new slate of candidates runs, and every year most students have no idea who the candidates are or what they stand for. Students vote based on the candidate they happened to see in a poster, a friend of a friend, or whichever candidate had the best-looking Instagram campaign. Substance rarely makes it into the voting decision. The design question: can a tool let a student get to know an unfamiliar candidate in the same time it takes to scroll through a feed?

How the industry typically approaches this

Conversational AI for public information has had strong recent results. BallotBot, a retrieval-augmented chatbot for California ballot initiatives, was tested in a randomised experiment with 1,500 voters and improved understanding, especially among voters with low initial knowledge. The City of Carson launched 'Ask Carson' on Google Dialogflow CX for 24/7 self-service, with 27% of conversations after hours. Universities have adopted the pattern at scale: Arizona State University and Thompson Rivers University both run admissions assistants handling the majority of student inquiries autonomously.

Our approach

A retrieval-augmented chatbot students could use to learn about candidates in their upcoming university election. Grounded in candidates' own platforms and public material, so answers were based on what each candidate had actually said. Students could ask questions in natural language, scan a candidate's code to pull up that profile and start a conversation, and compare what different candidates said about the same issue. Alongside the chatbot, produced a fully AI-generated launch film. No crew, no actors, no physical shoot. Produced in 24 hours of focused work using the AI image and video generation tools available at the time. It ran at the company's booth at a London-based education fair.

• •~3,000 chatbot conversations during the election cycle
• •Fully AI-generated launch film produced in 24 hours
• •Screened at a London education fair as part of the booth

What the industry has achieved with similar solutions

External benchmarks from comparable deployments. Sourced and labelled as third-party evidence, not our own results.

• BallotBot, California ballot initiatives

  Retrieval-augmented chatbot. 1,500-voter randomised experiment: reduced perceived effort, increased engagement, improved voter understanding. AEA RCT Registry

• Thompson Rivers University 'Penny'

  83% of student inquiries handled autonomously. Comm100

• City of Carson 'Ask Carson'

  Dialogflow CX chatbot. ~400 calls per month deflected. 27% of conversations after hours. Insight