Playbooks are ready-made prompts for the work admins and consultants do most often. Each one is written for a situation rather than a feature, and each returns a finished document you can share, not a chat answer you have to tidy up.
Find them at Playbooks in the left-hand navigation, or go straight to app.swantide.com/playbooks.
The quickest way to understand them
Open any playbook and you will see the full prompt, with a few words highlighted. Those highlighted words are the parts you can change, and they already have sensible values in them.
That means you do not have to fill anything in. A card that says "Fill 2 blanks" is telling you the prompt has two things you can adjust, not that you have work to do first. Open it, read it, and press Send to Assistant. Two clicks.
Change the highlighted words when you want a different object, a different process, or a different output format. Leave them alone when the defaults suit you.
What is in the library
Grouped by what you are trying to achieve.
Group | Use it when | Playbooks |
Assess | You need to understand what is really in the org, and rank what to do about it | Security and access review · Single-object deep dive · Technical debt register · Turn Health Check into a plan · SOW and gap analysis from a discovery call · Executive summary · Health score with a rubric |
Design | You have a requirement and need something you can defend in review | Solution design document · User stories with acceptance criteria · Impact and blast-radius analysis · Migration plan |
Document | Something is only in somebody's head and needs writing down | Business process documentation · Object documentation · Admin handover runbook |
Enable | People need to use or test what you built | End-user quick reference guide · Test plan and QA script · New-admin onboarding guide |
Extract | You want it in a spreadsheet so you can sort, filter and share it | Inventory extract as CSV · Configuration workbook |
What you get back
Every playbook produces a document rather than a chat reply. Depending on the playbook that is an interactive HTML report, a PDF, a Word file or a CSV, and it appears in the panel beside the conversation with a download control.
The content is drawn from your live configuration, so the findings name your actual components. An assessment will tell you which flows duplicate each other, which automation fires on every save regardless of what changed, and which test components are still active in production, by name.
A playbook takes a few minutes to run. Larger orgs and broader questions take longer.
Choosing the right one
The description under each title is the fastest way to pick. They are written as situations, so the one that matches yours is usually the right one.
About to change something and want to know what it will break: Impact and blast-radius analysis.
Inheriting an org or handing one over: Admin handover runbook, or New-admin onboarding guide.
Someone senior is asking how healthy the org is: Health score with a rubric if you want it scored against your own standards, Executive summary if they want one page.
You have had the discovery call and need to scope the work: SOW and gap analysis.
You need the detail in a spreadsheet: Inventory extract, or Configuration workbook for field-by-field detail.
Getting more out of them
The playbooks already follow the habits that make a prompt work well, so you do not need to learn them first. If you go on to write your own prompts, these are the ones worth copying.
Say what you want to end up holding. The playbooks ask for a report, a register, a plan, a workbook. Naming the artifact gets you the artifact.
Keep one object or one domain per conversation. Scope is the biggest single lever on answer quality, especially in a large org.
Ask for components by name. Every playbook does this, which is what makes the output checkable.
Keep going in the same conversation. Follow-up questions are where the depth comes from, because the Assistant keeps what it has already established.
After it runs
The document is yours to edit. Open it, read it, and ask for changes in the same conversation: a section expanded, a finding removed, the tone adjusted for a different audience. The document updates in place rather than being regenerated from scratch.
