# Built for SF by SF: AI Solutions Helping Our City Thrive Source: https://www.youtube.com/watch?v=EDqQtysycT4 Recap page: https://rapidrecap.app/video/EDqQtysycT4 Generated: 2025-10-08T17:35:05.064+00:00 --- ## Quick Overview The presentation showcases how AI, specifically models like GPT-5 Nano and Sora, is being integrated into San Francisco's municipal operations through the "Solve SF" initiative and the City Science Lab to streamline issue reporting, improve resource allocation for homelessness and public safety, and visualize urban planning scenarios, ultimately aiming to make the city safer, cleaner, and more vibrant. **Key Points:** - Solve SF, an AI solution built by SF residents, allows users to report issues like graffiti or trash via photos, with GPT-5 analyzing the image and automatically pre-filling 311 forms. - The analysis reduces the reporting process from potentially 12 steps down to just two clicks, significantly speeding up response times for issues like homelessness and public safety concerns. - The City Science Lab uses AI models (including GPT-5, Whisper, and text-embedding-3-large) to simulate future urban development scenarios, such as housing projects, to better understand impact. - The City Science Lab data platform centralizes information from nine disparate city departments (like Public Health, Fire, DEM, Homeless Services) into one searchable dashboard. - The City Science Lab demonstrated using AI to generate visualizations of proposed housing developments in different architectural styles (Modern, Victorian, Contemporary) based on user input. - The overall goal is to leverage AI to help San Francisco address complex challenges like homelessness and public safety by providing data-driven insights and efficient response tools. ![Screenshot at 00:24: 17:The City Science Lab dashboard is displayed, showing an analysis of a city block with different parcels color-coded based on development potential \(Eligible for development, Not economically viable, Prohibited\).](https://ss.rapidrecap.app/screens/EDqQtysycT4/00-00-24.png) **Context:** This segment from the OpenAI DevDay [2025] features presentations from key stakeholders involved in leveraging AI for municipal improvements in San Francisco. Cory Decker (OpenAI Global Head of Events) introduces the session, followed by Mayor Daniel Lurie discussing the city's commitment to innovation. The core of the presentation involves Patrick McCabe (Founder of Solve SF) demonstrating an AI-powered issue reporting app, and Kate Connolly and Peter Hirschberg (Co-founders of City Science Lab SF) showcasing how they use AI to model urban development and resource allocation across various city departments. ## Detailed Analysis The presentation transitions from OpenAI's general advancements to specific San Francisco initiatives using AI. Cory Decker introduces Mayor Daniel Lurie, who emphasizes the city's commitment to innovation and leveraging AI to solve local issues, noting that crime is down 30% downtown due to these efforts. Patrick McCabe then introduces the 'Solve SF' app, which uses GPT-5 to streamline reporting of street issues (like graffiti or trash) to the 311 service, reducing a 12-step manual process to just two clicks. He highlights that this AI capability was developed rapidly during a weekend hackathon. Following this, Kate Connolly and Peter Hirschberg from the City Science Lab SF present their work, which unifies data across nine separate city departments (including DEM, Public Health, Fire, and Homeless Services) into a single searchable dashboard. They demonstrate how AI models can simulate the impact of proposed developments, like adding housing to a block, by visualizing changes in community sentiment and commercial activity. They also showed the ability to generate photorealistic renderings of proposed buildings based on user-specified architectural styles (like Modern, Victorian, or Contemporary), aiding in urban planning discussions and decision-making. ### Introductions and Overview - Cory Decker introduces Mayor Lurie, who credits AI-driven efforts for reducing downtown crime by 30% and setting a global standard for urban AI. ### Solve SF App Demonstration - Patrick McCabe showcases the Solve SF app, which uses GPT-5 to analyze photos of street issues (graffiti, trash) and automatically populate 311 reports in two clicks, improving response times for first responders and 311 teams. ### City Science Lab SF Vision - Kate Connolly and Peter Hirschberg introduce their initiative, which integrates data from nine separate city departments into one dashboard to track progress on homelessness and safety issues. ### AI-Powered Urban Planning - The City Science Lab uses AI (GPT-5, Whisper, text-embedding-3-large) to simulate the impact of new development scenarios, such as housing construction, providing data-driven insights for city planners and advocates. ### Visualizing Change - The tool allows users to select a block and render proposed buildings in various architectural styles (Modern, Victorian, Contemporary) to help visualize future urban landscapes and maintain neighborhood character. ![Screenshot at 00:05: 05:The screen displays "OpenAI DevDay \[2025\]" as Cory Decker welcomes the audience.](https://ss.rapidrecap.app/screens/EDqQtysycT4/00-00-05.png) ![Screenshot at 00:24: 05:Mayor Daniel Lurie emphasizes that AI is used to reduce crime and improve civic life in San Francisco.](https://ss.rapidrecap.app/screens/EDqQtysycT4/00-00-24.png) ![Screenshot at 00:44: 19:Patrick McCabe highlights that Solve SF uses AI to automate the complex 12-step process of filing a 311 report down to two clicks.](https://ss.rapidrecap.app/screens/EDqQtysycT4/00-00-44.png) ![Screenshot at 01:06: 04:Mayor Lurie details the focus areas for AI in San Francisco: reducing crime, improving housing, and increasing civic engagement.](https://ss.rapidrecap.app/screens/EDqQtysycT4/00-01-06.png) ![Screenshot at 01:55: 03:Patrick McCabe demonstrates how the Solve SF app uses AI to classify issues like graffiti from a photo, reducing manual input.](https://ss.rapidrecap.app/screens/EDqQtysycT4/00-01-55.png) ![Screenshot at 02:31: 16:A slide lists the three core goals for VoiceReach: using data for resource allocation, broadening use cases for first responders, and expanding open-source tools.](https://ss.rapidrecap.app/screens/EDqQtysycT4/00-02-31.png) ![Screenshot at 04:49: 46:The screen shows an overview of the multi-departmental structure being unified under the Mayor's office via the Neighborhood Streams initiative.](https://ss.rapidrecap.app/screens/EDqQtysycT4/00-04-49.png) ![Screenshot at 07:22: 22:Patrick McCabe demonstrates the Solve SF app, showing how a photo of graffiti is automatically classified and ready for submission.](https://ss.rapidrecap.app/screens/EDqQtysycT4/00-07-22.png) ![Screenshot at 10:05: 05:A graphic displays the various tech stacks \(Swift, Kotlin, AWS, Python, etc.\) integrated by VoiceReach.](https://ss.rapidrecap.app/screens/EDqQtysycT4/00-10-05.png) ![Screenshot at 13:16: 16:A diagram illustrates the unification of nine separate city departments into five neighborhood stream groups, powered by AI.](https://ss.rapidrecap.app/screens/EDqQtysycT4/00-13-16.png)