diff --git a/docs/getting-started.md b/docs/getting-started.md new file mode 100644 index 0000000..b4d971e --- /dev/null +++ b/docs/getting-started.md @@ -0,0 +1,178 @@ +# Getting Started with AI for Curators + +Welcome! This guide will help you begin integrating AI into your curation workflows. Whether you're maintaining ontologies, knowledge bases, or other curated resources, AI can serve as a powerful assistant to enhance your productivity. + +## Prerequisites + +Before diving in, you should have: + +- Basic familiarity with your knowledge base or ontology editing workflow +- Access to GitHub (most of our tools integrate with GitHub-based workflows) +- A willingness to experiment and iterate (AI assistants improve as you learn to work with them) + +## Understanding the AI Mindset + +Before jumping into specific tools, it's helpful to understand different ways to think about AI in your workflow: + +### AI as Junior Curators +Think of AI as a team of junior curators who are eager to help but need clear instructions. They: + +- Benefit from detailed technical documentation and context +- Can handle routine tasks efficiently +- Need guidance on domain-specific conventions and best practices +- Should have their work reviewed before finalizing + +### AI as Agents +Modern AI can go beyond answering questions - they can: + +- Develop comprehensive plans for complex tasks +- Execute multi-step workflows autonomously +- Search, analyze, and synthesize information from multiple sources +- Make changes to files and create pull requests + +### AI as Personal Assistants +AI can serve as your coding and curation scribe: + +- Document your decisions and rationale +- Draft boilerplate text and code +- Help with repetitive formatting tasks +- Assist with documentation writing + +## Learning Path + +We recommend following this progression to build your skills effectively: + +### 1. Start with the Fundamentals + +**Read first:** [Using AI Right Now: A Quick Guide](https://www.oneusefulthing.org/p/using-ai-right-now-a-quick-guide) by Ethan Mollick + +This accessible guide covers: + +- Which AI models to use and when +- Practical prompting strategies +- Privacy and cost considerations +- Real-world examples of effective AI use + +**Time investment:** 15-20 minutes +**Key takeaway:** AI is most useful when you learn to work *with* it iteratively, not just throw requests at it + +### 2. Choose Your Primary Tool + +Based on your workflow and needs, select one of these tools to start: + +#### Claude Code (Recommended for most curators) + +Best for: Integrated coding and curation workflows, GitHub-based projects + +**Resources:** + +- [Claude Code Overview](https://docs.claude.com/en/docs/claude-code/overview) - Official introduction +- [Claude Code Best Practices](https://www.anthropic.com/engineering/claude-code-best-practices) - Essential reading +- [DeepLearning.ai Short Course](https://www.deeplearning.ai/short-courses/claude-code-a-highly-agentic-coding-assistant/) - Comprehensive hands-on course (highly recommended) + +**When to use:** + +- Working with files in a GitHub repository +- Need to make systematic changes across multiple files +- Want an AI that can explore your codebase and understand context +- Prefer a conversational, iterative workflow + +**See also:** Our [Claude Code reference](reference/clients/claude-code.md) + +#### GitHub Copilot + +Best for: In-editor assistance while coding or writing + +**When to use:** + +- Working directly in VS Code, Visual Studio, or other supported editors +- Want inline suggestions as you type +- Prefer suggestions integrated into your existing editor workflow + +#### DragonAI (GitHub Agent) + +Best for: Automating GitHub-based workflows + +**When to use:** + +- Want an AI agent that responds to GitHub issues and comments +- Need automated pull request generation +- Building CI/CD pipelines with AI assistance + +**See also:** Our guide on [instructing the GitHub agent](how-tos/instruct-github-agent.md) + +#### Goose + +Best for: Command-line workflows and automation + +**When to use:** + +- Prefer working in the terminal +- Need scriptable AI interactions +- Want to integrate AI into existing shell scripts + +**See also:** Our [Goose reference](reference/clients/goose.md) + +### 3. Apply to Your Workflow + +Once you've chosen a tool and completed initial training: + +1. **Start small** - Pick a single, well-defined task +2. **Provide context** - Give the AI information about your domain and conventions +3. **Iterate** - Refine your prompts based on results +4. **Review carefully** - Always validate AI-generated content before committing + +**Practical guides:** + +- [Instruct the GitHub agent](how-tos/instruct-github-agent.md) +- [Make identifiers hallucination-resistant](how-tos/make-ids-hallucination-resistant.md) +- [Integrate AI into your KB](how-tos/integrate-ai-into-your-kb.md) +- [Ontology editing with AI](tutorials/ontology-editing-with-ai.md) + +### 4. Expand Your Skills (Advanced) + +Once comfortable with the basics, explore advanced topics: + +- **Multi-agent workflows:** [Multi-agent Research System at Anthropic](https://www.anthropic.com/engineering/multi-agent-research-system) +- **Knowledge graph construction:** [Agentic Knowledge Graph Construction Course](https://www.deeplearning.ai/short-courses/agentic-knowledge-graph-construction/) +- **Claude Skills:** Our guide on [Claude Skills](reference/claude-skills.md) for creating reusable AI capabilities + +## Quick Reference: Tool Selection + +Not sure which tool to use? Here's a quick decision tree: + +``` +Do you need to make changes to files in a GitHub repo? +├─ Yes → Are you working on systematic, multi-file changes? +│ ├─ Yes → Claude Code or DragonAI +│ └─ No → GitHub Copilot (in-editor) or Claude Code +└─ No → Are you primarily asking questions or exploring ideas? + ├─ Yes → Claude Desktop or web interface + └─ No → Command-line workflows? → Goose +``` + +## Common Pitfalls to Avoid + +1. **Expecting perfection on the first try** - AI tools work best with iteration and refinement +2. **Not providing enough context** - Share relevant documentation, examples, and conventions +3. **Skipping validation** - Always review AI-generated content, especially for critical data +4. **Using AI for everything** - Some tasks are still faster done manually +5. **Ignoring privacy** - Don't share sensitive data with AI tools unless you understand their privacy policies + +## Getting Help + +- Check our [FAQ](faq.md) for common questions +- Review our [Glossary](glossary.md) for terminology +- Explore topic-specific guides in our [how-tos](how-tos/instruct-github-agent.md) section +- See the [Reference section](reference/client-apps.md) for detailed tool documentation + +## What's Next? + +Now that you understand the landscape, we recommend: + +1. **Complete the foundational reading** (Ethan Mollick's guide + your chosen tool's documentation) +2. **Pick one tool** and complete its tutorial or short course +3. **Try one real task** from your actual workflow +4. **Share your experience** - contribute back by documenting what worked (or didn't!) + +Remember: the goal is to enhance your curation workflow, not replace your expertise. AI tools are most effective when they amplify your knowledge and judgment. diff --git a/mkdocs.yml b/mkdocs.yml index 159af61..21e82f1 100644 --- a/mkdocs.yml +++ b/mkdocs.yml @@ -36,6 +36,7 @@ plugins: nav: - Home: index.md + - Getting Started: getting-started.md - FAQ: faq.md - Curator how-tos: - Instruct the GitHub agent: how-tos/instruct-github-agent.md