In the future this slide won't be necessary
removed for time - consider if re-running at 4 credits --- ## Exercise: Planning a Task <div class="columns"> <div class="column text-sm"> **Tasks** - Prepare a literature review on library services for non-binary youth - Develop a collection development plan for a graphic novel section in an academic library - Organize a community workshop on digital privacy tools for seniors - Create a research survey and sampling strategy to assess student information literacy skills - Design a metadata schema for a digital collection of historical photographs - Develop a recipe book with family favorites that accommodates dietary restrictions </div> <div class="column text-sm"> 1. Break into small groups (3-4 people) 2. **Spend 5 minutes listing the specific steps you would take to complete this task** 3. For each step, identify: - What information you need - What tools or actions would be involved - What output you expect, and how you'd combine the prior parts 4. Share your approach with another group (on Zoom, ping me when you're at this step) 5. Discuss: How much of this can be approached by an AI Agent? </div> </div> --- ## Agent Components - **Environment**: Defines where the agent operates (web, code editor, game, etc.) - **Tools**: Extend the agent's capabilities - Knowledge augmentation - add access to source documents, rather than relying on the model's learned knowledge (e.g. retrieval augmented execution) - Capability extension - the things that Language Models can't do themselves (e.g. calculator, code execution, image generation) - Write actions (file editing, sending emails) - **Planning**: The agent's ability to reason about how to accomplish tasks <p class="footnote"> <a href="https://huyenchip.com/2025/01/07/agents.html">Huyen, 2025</a> </p> ---
FUTURE - remove terminal examples altogether
removed to get to the lab quicker ---- ## Coding Agents for Information Professionals LIS-relevant use cases where coding agents reduce friction: <div class="text-sm"> - **Data wrangling**: "I have a messy spreadsheet with 10,000 rows — clean the dates, remove duplicates, flag missing values" → Python script written and run - **SQL queries**: "Write a query that finds all patrons who borrowed more than 5 items in the last 3 months" → working SQL, no memorizing syntax - **Format conversion**: batch convert MARC records to Dublin Core; transform XML metadata; parse JSON-LD - **Small tools**: a duplicate ISBN checker; a citation formatter; a simple web scraper for a research project - **Analysis**: "Here's a CSV of circulation data — what patterns do you see? Visualize by month" - **Documentation**: "Here's a Python script I inherited — explain what it does and add comments" </div> ----