What people do with Devin
Cognition's autonomous software engineer. 31 real use cases so far, most often for software. Each links to its source and comes with a prompt to try.
31 use cases · 7 kinds of job · Updated Oct 7, 2026
All 31 Devin use cases
Test features on screen and merge fixed PRs
Ryan Carson uses Devin's Test App feature to record and narrate feature clicks, fix bugs, and merge PRs with a !land command.
How it went Shipping velocity and dev happiness increased drastically with the cloud VMs and gpt-5.5.
2 accounts
Build a name-and-face quiz site from a spreadsheet
Scott Wu had Devin build and deploy a mobile-friendly React quiz from a TSV of speaker names and photos, plus a QR-code site to share it.
How it went The final site worked, with a streak counter that resets on a wrong answer, covering roughly 170 speakers. The first version showed names under the photos, which gave away the answer, because Wu hadn't specified that precisely.
2 accounts
Turn Slack feedback into PRs and debug infra
Sid Nanda's company uses Devin in Slack to triage feedback, turn user issues into PRs, debug async flows and infra read-only, and let non-technical staff coauthor PRs.
How it went Sid ranked top 200 global Devin users. Time-to-value was strong enough to justify premiums despite high cost for Devin Review, which caught good bugs.
Build a WordPress marketing site from Figma overnight
bnistordev connected Devin to a Figma file via MCP and had it build an entire WordPress marketing site overnight.
How it went Output quality matched work he would quote €2.5k for, though the site is still unfinished.
Split a React Native to Swift migration across agents
Cognition's Jared Zoneraich uses Devin fan-out so child agents work in parallel on a migration split into six pieces.
How it went The migration completed via parallel child agents. No specific completion time, code volume, or shortfalls are stated.
Ship more pull requests with Devin across tools
Ryan Carson uses Devin from iPhone, Slack, browser, GitHub and Linear for real work and reports higher PR shipping velocity.
How it went PR shipping velocity rose higher than before.
Test Devin on pulling Notion data into Sheets
Answer.AI testers found Devin completed only 3 of 20 tasks, succeeding at moving Notion data to Google Sheets and building a planet tracker.
How it went Devin finished 3 of 20 tasks: the Notion-to-Sheets move, the planet tracker, and a Discord bot research task. Three were inconclusive and 14 failed. Simple-looking tasks often took days, and it was hard to predict which would work.
Test Devin on integrations, migration and security review
Answer.AI tried Devin on an observability integration, an nbdev migration and a vulnerability review; it found some real issues, hallucinated others, and failed the migration.
How it went The Braintrust integration was abandoned and redone in Cursor. The nbdev migration failed on basic setup. The security review found some real issues but also false positives and hallucinated ones. Overall 14 failures, 3 successes and 3 inconclusive.
Migrate an Electron app from Vue 2 to Vue 3
Kearney gave Devin repo access to migrate a production Electron app to Vue 3 and update npm packages, committing after every five components.
How it went Devin opened a PR after four hours and reported a passing build, but the build failed on Kearney's machine. After about two days of fix cycles, the code still didn't run. He gave up and used roughly half his ACUs.
Build a Pong game and website from scratch
A tester asked Devin to create a basic Pong-style game and a website, and it finished both in under 20 minutes.
How it went Per Vance, Devin finished both the Pong-style game and the website in under 20 minutes. The article gives no detail on the quality of the results. It says more complex tasks take longer.
Build and deploy a Game of Life simulation site
In a vendor launch demo, Devin built an interactive Game of Life website, added requested features step by step, and deployed it to Netlify.
How it went The demo shows the site built and deployed end to end. The post gives no timings, costs or failure details, and it is a vendor showcase, not an independent test.
Hand narrow code changes to Devin for pull requests
A team used Devin for a month on small fixes like chart formatting and API edge cases, with tests and PRs; a complex full-stack task went poorly.
How it went Small single-component tasks took a minute or two to write up and five minutes to test, against roughly an hour by hand. A full-stack task with a new table, API call and button failed, and the team rewrote it themselves.
Add a third-party API integration via Devin PR
Devin opened a PR with the endpoint and backend wiring that was merged, but missed API key logic, code style and abstraction, needing review fixes.
How it went After 2-3 review rounds the code was production-ready, tests passed, and the PR merged. The whole task took about an hour, against roughly 15 minutes if done by hand. The author spent about 5 minutes reviewing.
Build and test a simple endpoint project with Devin
A reviewer had Devin set up a project, write code, install dependencies, test endpoints and add error handling in about 4 minutes on a simple task.
How it went It finished in about 4 minutes, against the reviewer's estimate of 5 minutes by hand. The reviewer says Devin's advantage is small on tasks this simple, and that it was slower than doing the work yourself on other tasks.
Turn Slack and Linear tasks into PRs and fix CI
Cognition staff assign Devin work in Slack or Linear; it posts PRs, fixes bugs flagged by review bots and works CI and lint failures until checks pass.
How it went Cognition merged 659 Devin PRs in one week, up from 154 in its best 2025 week. The source gives no failure rates, and says complex work should be split into small tasks.
Build apps, debug repos and set up CI/CD with Devin
A review lists Devin building a full-stack to-do app, debugging a public repo with a PR, and doing Docker setup, deployment and CI/CD.
How it went The review calls Devin strong on boilerplate and clearly specced features. It is weaker on legacy refactoring, complex logic and vague instructions. One beta tester said Devin did 90% of a Node.js backend setup, leaving cleanup.
Write, review and document code across a product's PRs
Founders report Devin touching all PRs at ChatPRD and being Gumroad's top contributor with over 1,500 merged PRs.
How it went Claire Vo calls Devin the #2 contributor to ChatPRD, behind only her. Sahil Lavingia says it is Gumroad's top code contributor, with over 1,500 merged PRs, about 10 a day. Both are founder claims with no independent measurement.
Generate internal dashboards and data migration scripts
A backend engineer says their team uses Devin for internal dashboards and migration scripts, treating output as a starting point that saves about four hours of boilerplate.
How it went The engineer says it reliably saves about four hours of boilerplate. The article says this is an extrapolation from 25–45 minutes saved per task, and it gives no controlled data.
Set up weekly changelogs and fix docs from GitHub issues
Neon's docs team has Devin create the weekly changelog from a playbook and open or revise doc PRs from GitHub issues, for repetitive work only.
How it went This helped them work through a backlog of older docs issues without auditing each one by hand. Devin struggled with open-ended tasks, and a bulk term rename needed 20+ manual cleanup commits.
Rewrite a README after a rearchitecture
Devin scanned a rearchitected codebase and delivered a new README as a PR with accurate setup and API details, but struggled with design rationale and purpose.
How it went The README had accurate setup steps, clear API endpoint specs, and examples inferred from the code. It struggled to explain design rationale and the project's overall purpose, which the codebase doesn't contain.
Build and deploy an app over a weekend
The author had Devin build and deploy an app, finding it quick on PR feedback but weak on legacy monorepos and costly on broad requests.
How it went A working, deployed app took about 90 minutes from an empty repo. Devin is weaker on older monorepos and bespoke infrastructure, and broad requests burn more compute units.
Click through apps to verify its own work
With computer use, Devin built and played a desktop game, ordered on Amazon, and tested its own work by clicking through the app.
How it went Engineers ran 10 to 20 Devins in parallel, each with its own dev server. Approved test runs per day more than doubled. Timing-sensitive UI like toasts confuses it, and it sometimes cheats by running JavaScript instead of clicking.
Answer data questions and build dashboards from the warehouse
Cognition's team uses a Devin variant called DANA to query their data warehouse, build dashboards and answer data questions, and Devin fills out report forms.
How it went Ad-hoc questions that used to wait days in an engineer's queue can now be answered by anyone in Slack, with the SQL included so the logic can be checked. The post gives no numbers for this.
3 accounts
Surface unanswered emails from the past week
Author had Devin build a Gmail API assistant that lists unanswered emails from the past week, since Attio lacks that tab; he wrote the auth part himself in Cursor.
How it went It produced a working function, a front end, and a public deployment in one late-night session. It worked at human speed: the Supabase setup ran over an hour, which he says he could do in under 10 minutes.
Turn Notion PRDs into scoped Linear tickets
Devin's docs describe reading a Notion PRD to create Linear tickets, producing implementation plans for a backlog, and posting customer health summaries to Slack.
How it went The page states no results, numbers, or limitations. Open the linked prd-to-tickets page for the actual prompts and steps.
Build a chess game against a language model
An early tester asked Devin to build a chess game against a language model; it made good progress but eventually hung.
How it went Devin made remarkable progress, but at some point it hung and the game wasn't finished. Gao's other tests had mixed results: a Chrome extension worked, and an Antarctic temperature visualization was unsatisfactory.
Test an iPhone game and report what passed
Cognition had Devin on GPT-6 Astra test the game Otter Run in a simulator, returning a recording and a report of checks passed and untested areas.
How it went The source gives no pass counts or timings. It presents the example as a step toward engineers reviewing less code by hand, which Cognition expects over time and has not yet shown.
Hunt jobs, tailor resumes and apply automatically
An agent called Resume finds and ranks jobs around schedule and pay wants, drafts tailored resumes and submits applications; the author reports 3 interviews and 1 offer.
How it went Secured 3 interviews and 1 job offer with start date the following week that aligned with existing schedule and finances.
Try Devin on research tasks (mostly failed)
Answer.AI had Devin fetch an author's recent papers and research transcript summarization; it got stuck parsing HTML and returned generic results, rated a failure.
How it went Both failed. On the scraping task Devin got stuck parsing HTML and went to sleep. On the research task it returned related but unhelpful material and code that missed the core problem.
Explore code to write technical documentation
A technical writer uses Devin to explore code before documenting it, plus a Slack writing agent built on Devin, reporting about 20,000 lines of docs per month.
How it went She reports about 20,000 new lines of docs per month across roughly 500 pages. She reviews and edits every page, often heavily. Fern Writer's first drafts sometimes got the structure wrong or skipped a changelog entry.
Hunt business-logic flaws and open fix PRs
Parallel Devin agents each read part of a codebase, find business-logic and auth bypass flaws, confirm them in a sandbox, then open remediation PRs (vendor description).
How it went On Cognition's own 50-vulnerability benchmark, Devin Security scored 72% recall at $90.23 per run, against 68% for Claude Security at $131.87. This is a vendor benchmark, not an independent one.
2 accounts
Five new use cases in your inbox every morning
The best things people got an AI agent to do, each with the prompt to try it.