By Teresa Torres · producttalk.org · @ttorres on X · LinkedIn
Teresa Torres follows up her home-use AI article with a practical guide to using large language models at work, organized into increasingly complex categories: translation, doing the work, research, writing, and coding. She argues that skills built in low-stakes personal use (prompting, context, verification) transfer to professional settings, while work demands extra care about when AI is appropriate. Her examples emphasize keeping humans in the loop, building reusable CustomGPTs and Projects, and automating repetitive workflows over time. The piece matters to product people because it offers a concrete, incremental path to building personal AI fluency and a toolbox of workflows.
01Key takeaways
- Start with low-stakes AI tasks like translation, then progress to more complex workflows as your confidence grows.
- Keep a human in the loop for high-stakes outputs such as customer communications, so AI speeds up human judgment rather than replacing it.
- Use CustomGPTs or Projects to store recurring context and prompts so you stop re-entering the same instructions.
- Ask an LLM to write the prompts you will reuse; this tends to produce more consistent, higher-quality output.
- Feed AI rich, specific context such as customer profiles, examples, and source documents, since context quality drives recommendation quality.
- Use AI as a thought partner for feedback and brainstorming rather than letting it do all the thinking yourself.
02Key sections
- Translator
- Starts with low-stakes uses like translating emails and parsing international addresses, plus creating CustomGPTs to save repetitive prompts.
- Do the Work
- Shows AI handling multi-step tasks such as template matching for support, LinkedIn analytics analysis, automated article summaries with a human review step, and contract review.
- Researcher
- Describes using AI for vendor evaluation during a platform migration and for quarterly deep-research reports on academic literature, now automated.
- Learning to Use AI as a Thought Partner
- Introduces the idea that LLMs are most valuable as thinking partners that give feedback and help get unstuck, while the author keeps authorship of her ideas.
- Writing Partner
- The author explains her reluctance to outsource writing because drafting is how she figures out what she thinks, and reframes the LLM as a feedback partner.
03From the post
“I recently shared 15 ways I'm using AI at home—from fixing cooking disasters to researching school bonds. Each use case helped me build new AI skills: learning to chat with a large language model (LLM), providing the right context, verifying results, and much more. Now it's time”
04Frameworks mentioned
Summary and takeaways written by PM Atlas; quotes are short excerpts. © the original author.