claim accepted goes to: this-edition A all three
Tools changed everything: the model answers with a request, the program runs it, and the result goes into the next request
The first ChatGPT and Claude had only their training, a compression of the world that stopped at a date. Now a model can return a tool call instead of an answer (search the web, call a service, read a file); the harness decides whether to run it, runs it, adds the result to the conversation and sends it all back. The model never runs a tool, which is where the control lives. From no tools in November 2022 to an open standard for them in November 2024.
Decided by. Dinis, in Lenses, roles and loops; the editor records it in the vault
Note. Dates as commonly recorded; 'training as compression' is a metaphor, the article says so.
Where it comes from
- Lenses, roles and loops Tools changed everything; Precisely: the model never runs a tool (article)
What it answers in the 2018 book
- Machine Learning and AI answered written, Technologies
Where it goes
Chapter Lenses, roles and loops, part The frame (drafting); topic How to work with agents: briefs, voice, one shared state.
Other ideas proposed for the same chapter: A prompt is a lens: translating into another language is, to a model, the same as explaining for a teenager or a doctor, A role is a model with a description and a question; an agent is that role on a loop with a goal, The model remembers nothing between requests: every request carries the whole conversation, which is why there is a context window, Long chats are compressed into a lower-resolution summary, so details drop out: work in many focused threads with named agents.
Node idea:tools-changed-everything in graph.json; see it in the graph. Source: admin/content/graph/ideas.json.