What can you do with AI?
30 practical jobs, each with the best-scored tools, prompts to copy and a workflow.
- Use cases
- 30
- Tools behind them
- 49
- Audiences
- 5
- Coding agents14 tools, top: gemini-cli 708
- Agent skills & MCP9 tools, top: superpowers 624
- Automation & agent frameworks14 tools, top: n8n 706
- Research & web data6 tools, top: firecrawl 785
- RAG, knowledge & memory3 tools, top: ragflow 693
- Local AI5 tools, top: ollama 733
- Media & creative1 tool, top: remotion 431
Top 10 by public usage data
What people actually use AI for, from published studies. Ranked by cited evidence, not by our scores.
- 1AI pair programming in the terminalAnthropic, McKinsey, Stack Overflow6
- 2AI web research reports with sourcesMcKinsey, OpenAI; indirect: Stack Overflow5
- 3AI customer support botMcKinsey, Microsoft4
- 4AI debugging with an agentAnthropic, Stack Overflow4
- 5Business process automation with AI agentsMcKinsey, Microsoft4
Pick by who you are
- 25DevelopersCoding agents, code review, tests, debugging, MCP servers and agent frameworks.
- 11Small businessAutomation, customer support, private chat and knowledge assistants.
- 6Marketing and contentResearch, competitor monitoring, cited articles, automation and video.
- 7Researchers and analystsSourced reports, scraping, document conversion and document Q&A.
- 3Privacy-focused teamsLocal models, private chat and offline coding.
All 30 use cases
Grouped by job. The number is the highest tool score for that use case.
Coding agents
- AI pair programming in the terminal5 toolsgemini-cli 70
- AI code review4 toolsclaude-code 65
- AI refactoring and code migrations4 toolsOpenHands 68
- AI test writing4 toolsclaude-code 65
- AI debugging with an agent4 toolsOpenHands 68
- AI documentation generation4 toolsgemini-cli 70
- AI for GitHub issues and pull requests4 toolsOpenHands 68
- Up-to-date library docs for AI coding4 toolsclaude-code 65
Agent skills & MCP
- Give AI agents new abilities with MCP and skills4 toolsskills 53
- Browser automation for AI agents3 toolsbrowser-use 62
- Repeatable agent workflows with Agent Skills3 toolssuperpowers 62
- Turn recorded work into reusable agent skills3 toolssuperpowers 62
Automation & agent frameworks
- Workflow automation with AI3 toolsn8n 70
- Multi-agent AI teams4 toolscrewAI 57
- AI customer support bot4 toolsn8n 70
- Business process automation with AI agents4 toolsn8n 70
- Build AI agents in Python or TypeScript6 toolslangchain 54
- Persistent AI agent workspaces3 toolslobehub 69
Research & web data
- AI web research reports with sources3 toolsbrowser-use 62
- Scrape websites into clean data for AI2 toolsfirecrawl 78
- Convert documents to Markdown for LLMs4 toolsfirecrawl 78
- Competitor and web monitoring with AI3 toolsfirecrawl 78
- Write cited, long-form articles with AI2 toolsgpt-researcher 31
RAG, knowledge & memory
- Chat with your documents (RAG)4 toolsragflow 69
- Long-term memory for AI agents2 toolsmem0 55
- A company knowledge assistant3 toolsragflow 69
Local AI
- Run AI models locally3 toolsollama 73
- A private AI chat for your team3 toolsollama 73
- An offline AI coding assistant4 toolsollama 73
Media & creative
- Programmatic video with React1 toolremotion 43
No use case matches that search.
Questions about AI use cases
What are AI use cases?
Concrete jobs you can do with AI tools, such as reviewing code, researching a topic, answering questions from your documents or running a model privately. Each page here explains the job, gives example prompts and a workflow, and lists the tools we track for it.
How are the tools on these pages chosen and ranked?
Only tools in our weekly dataset (52 tracked GitHub projects) appear. Each has a quality score from public signals (maintenance, adoption, docs) and a separate automated safety label. Scores are relative and nobody here tested the tools hands-on. See the methodology page.
Are the example prompts tested?
No. Prompts are labelled "Example prompt" because they are starting points we wrote, not results from a test run. A prompt would only be labelled tested after someone ran it and we recorded when and with what.
Does a green safety label mean a tool is safe?
No. It means our automated scan of dependencies and code patterns found no issues. It is not an audit. Give agents the least access they need and review what they do.
Can I use these tools at work?
Often yes, but check each licence and your organisation's data rules first. Every tool card shows its detected licence and last activity.
How we pick tools
- Quality and safety, separately. Every tool has a relative quality score from public GitHub signals and an independent automated safety label.
- Real data only. Scores, labels, licences and dates on these pages are computed from our weekly dataset, never typed by hand.
- Honest limits. Nobody here tested the tools hands-on and example prompts are untested. Read the methodology.
Get the board every Monday
One email a week: the new top 10, the biggest score changes and every safety label that changed.
Free. Double opt-in: you confirm by email first, unsubscribe in one click.