Prompting
In this section
- AI prompting guide 2026The thirteen prompting concepts from The AI Agent Factory, carried in full: how AI knows things, how to brief it with the right context, how to neutralize agreement bias, the brainstorm-iterate loop, images, apps and data, permissions, model choice, and models checking models. The rules any instruction, system prompt or brief written from this record must follow.human:rrizwan98
- How AI knows things: novice vs power user, pretrained knowledge, three retrieval modesConcepts 1 to 3 of the book's prompting guide: what separates a novice prompt from a power-user prompt, where a model's knowledge is strong, sparse or absent, and how to tell whether an answer came from memory, a web search or deep research, with the phrasings that steer each mode and the three lines that make web search reliable.human:rrizwan98
- Context is the whole game: the six layers, your own layer, memory, context rot and projectsConcept 4 of the book's prompting guide in full: what is on the model's desk for every answer, the system prompt you never see, how to write and prune your own standing instructions, the checklist before any non-trivial prompt, the memory layer the tools now write about you, when a conversation has gone stale, and how projects front-load context once.human:rrizwan98
- Reasoning and sycophancy: think hard, frame neutrally, force a numberConcepts 5 and 6 of the book's prompting guide: how to invoke a model's thinking mode and when not to, why models are biased toward agreement and the measured size of that bias, the bait phrasings and their neutral rewrites, the objective-rubric pattern, and the instruction that turns a grade into a next step.human:rrizwan98
- The brainstorm-iterate loop, and one loop for four kinds of askConcept 7 of the book's prompting guide, the single highest-leverage habit: load context, demand options, give explicit feedback, iterate, expand only the winner, then grade and re-iterate until the score plateaus; outline before drafting; research before either; and the dial that loosens the spec for options and tightens it for execution.human:rrizwan98
- Beyond text: images, audio, small apps and data analysisConcepts 8 to 10 of the book's prompting guide: what AI sees well and badly in an image, how to prompt image generation and its failure modes, the diagram recipe, audio in and out, the three-slot recipe for a one-prompt app and where it runs, and how to make the model actually run code on your data instead of guessing.human:rrizwan98
- Safety, cost and checking: permissions, which model when, models checking modelsConcepts 11 to 13 of the book's prompting guide: the plan-then-approve workflow and permission ladder for file-aware AI apps, the cost and speed stack, why no single model is best and how to keep tooling current, and the full recipe for making models from different families grade each other, with the single-model self-critique loop for everyday work.human:rrizwan98
- Twelve prompts to try, in orderThe book's closing exercises for the prompting guide, verbatim: twelve prompts that take about twenty-eight minutes in any chat tool and exercise every concept you can exercise from a chat tab, from a web-search trigger to a cross-model review.human:rrizwan98
From one task to an AI-native company
The ladder from a single automated workflow to a company where work enters by itself and people sit only at the review points.
AI prompting guide 2026
The thirteen prompting concepts from The AI Agent Factory, carried in full: how AI knows things, how to brief it with the right context, how to neutralize agreement bias, the brainstorm-iterate loop, images, apps and data, permissions, model choice, and models checking models. The rules any instruction, system prompt or brief written from this record must follow.