Devin deploys a City Explorer prototype
Sambhav Gaur reports Devin scaffolding, testing and deploying a City Explorer dashboard with FastAPI, React and placeholder external integrations in about ninety minutes.
medium.com
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Build a City Explorer preview in this empty repository. Mark placeholder APIs clearly, test routes and UI, open a review PR and deploy a prototype with a list of incomplete integrations.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Sambhav Gaur
- Original account
- Read the source ↗
- Published
- 13 Aug 2025
- Source link checked
- 15 Oct 2026
- Reported result
- Sambhav Gaur reports Devin scaffolding, testing and deploying a City Explorer dashboard with FastAPI, React and placeholder external integrations in about ninety minutes.
- Human involvement
- Gaur approves or clarifies prompts and assesses the prototype. Images, news and weather integrations are explicitly placeholders.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Gaur gives Devin an empty GitHub repository and a City Explorer dashboard brief.
- Devin clones the repository and scaffolds FastAPI and React.
- It adds query routes, environment configuration and placeholder external integrations.
Sambhav Gaur tried Devin on a weekend prototype called City Explorer. He supplied an empty GitHub repository and requested a dashboard for a selected city's images, weather and news. He reports a deployed application after roughly ninety minutes, with only one or two approval or clarification interruptions.
The described session begins with Devin cloning the empty repository, noticing that no code exists and asking about other branches. It then scaffolds a FastAPI backend and React frontend, configures environment files and routes city queries through the application.
The account says the integrations for images, weather and news were placeholders. Devin starts the app locally for testing, pushes changes, opens a pull request and deploys a public URL. Those steps support a deployed prototype, but they do not establish that every external data source was live or correct.
Gaur also describes discomfort with limited visibility into model choice and internal decisions. His comments about future uses, legacy-code performance and automatic learning are broader observations, not measurements from this particular build. The useful case is a bounded empty-repository experiment with an inspectable interface and explicit placeholder limits. The reported duration is his observation, not a controlled comparison with a human developer.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: medium.com · Sambhav Gaur
Frequently asked questions
Were the external data integrations complete?
The account explicitly calls the images, weather and news integrations placeholders.
What did Devin deploy?
A City Explorer prototype with a FastAPI backend and React frontend.
Is ninety minutes a benchmark?
No. It is Gaur's timing for this individual experiment.
More details
Remaining steps
- It starts the app for testing, pushes a PR and deploys a public prototype.
About this account
Devin generated the city-explorer backend and frontend, tested the application and prepared a deployed prototype with incomplete external integrations.
Firsthand account opened and read. Outcome is source-reported, not independently reproduced. Suggested prompt written by Agents In Use. Recovered readable Medium original after local 403; prototype with placeholder integrations, not a verified live-data application.
Credit: Sambhav Gaur
Source published 13 Aug 2025
Source checked 15 Oct 2026
Tools and topics
Devin · FastAPI · React · GitHub
Devin · FastAPI · React
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.