Your literature review has 40 abstracts. How long did the first pass take?
Three hours — read each one, take notes, decide if it's relevant. And I'm still not sure I caught everything important.
One function call handles that first pass in two minutes. Agent(model).run_sync(prompt).output — that's it. Pass an abstract in, get a relevance verdict back as a string. The sandbox wires the model automatically. This week every lesson is one new thing you can do with that call: count its output, add a summarisation prompt, lock it to a single label, chain two calls together.
But I can't cite an AI summary in my thesis. What's the actual use case here?
Triage, not citation. You use the agent to filter 40 abstracts down to 8 worth reading carefully. You still read and cite those 8 yourself — the agent saved you from reading the 32 that weren't relevant. That's the research assistant pattern. Week 2 makes the output typed and structured so it fits into a methodology table.
Agent(model).run_sync(prompt) and .output.split()system_prompt that turns any agent into a summariserGoal: by Day 7 you can call an AI model, process its output as a string, and wire two agent calls together in one function.
Create a free account to get started. Paid plans unlock all tracks.
Your literature review has 40 abstracts. How long did the first pass take?
Three hours — read each one, take notes, decide if it's relevant. And I'm still not sure I caught everything important.
One function call handles that first pass in two minutes. Agent(model).run_sync(prompt).output — that's it. Pass an abstract in, get a relevance verdict back as a string. The sandbox wires the model automatically. This week every lesson is one new thing you can do with that call: count its output, add a summarisation prompt, lock it to a single label, chain two calls together.
But I can't cite an AI summary in my thesis. What's the actual use case here?
Triage, not citation. You use the agent to filter 40 abstracts down to 8 worth reading carefully. You still read and cite those 8 yourself — the agent saved you from reading the 32 that weren't relevant. That's the research assistant pattern. Week 2 makes the output typed and structured so it fits into a methodology table.
Agent(model).run_sync(prompt) and .output.split()system_prompt that turns any agent into a summariserGoal: by Day 7 you can call an AI model, process its output as a string, and wire two agent calls together in one function.