Connect Outlign to your AI assistant (Claude, ChatGPT, Cursor or any tool that supports MCP) and the admin that eats your week (turning meeting notes into tasks, setting up new projects, chasing status) becomes a single prompt. These playbooks are agency-tested prompts for the jobs that come up on every project. Copy one, swap the bits in [square brackets] for your own details, paste it into your AI, and let it do the clicking.
Each playbook has the same four parts:
When to use it: the moment in a project where it saves you time
What you need: the connectors and inputs your AI needs before it starts
The prompt: copy and paste this
What the AI does: so you know what to expect and what to check
You don't need to be technical to use these. If you can write a brief, you can write a playbook.
1. Connect Outlign to your AI tool. Outlign has an MCP server, so it works with any AI tool that supports MCP connectors, including Claude, ChatGPT and Cursor. Add Outlign in your tool's connector or integration settings and sign in with your Outlign account. Your AI can only see and do what your Outlign user can, so connect as someone with access to the projects you want it to work on.
2. Connect the tools your inputs live in (optional). The playbooks work fine if you paste or upload things straight into the chat. But if these are connected, your AI can go and get them itself:
Connector | What your AI pulls from it |
|---|---|
Your meeting notetaker (Fathom, Fireflies, Supernormal, Otter, etc.) | Call transcripts and summaries |
Qwilr (or your proposal tool) | Accepted proposals, scope, pricing, timeline |
Google Drive / SharePoint | Briefs, feedback docs, schedules |
Gmail / Outlook | Client feedback that came in by email |
Google Calendar | Meeting dates for milestones |
3. Know the two sides of an Outlign project. Every project has an internal side (your team only) and a client side (what your client sees). The two often have phases with the same name, like "Concept" on both. Good prompts always say which side, and a good assistant will ask if it's unclear. Internal and client tasks can be linked, so the client sees a clean version of the job while your team works the detail on the linked internal task.
4. Tell your AI your house rules once. Put these in your AI tool's custom instructions (a Claude project, a ChatGPT project or custom GPT, and so on) so every playbook picks them up:
When working in Outlign:
- Always search for an existing task before creating a new one. If it exists, add a comment instead of duplicating it.
- Never make anything client-facing (tasks, messages, documents) unless I say so.
- Show me the plan before you create more than 5 things.
- Our team: [Name, role], [Name, role], [Name, role].When to use it: straight after a client call, while the actions are fresh.
What you need: the transcript (pasted, uploaded, or in a connected notetaker) and the Outlign project name.
The prompt:
I've just had a call with [Client] about [Project]. Here's the transcript: [paste it, attach it, or say "grab the latest [Client] call from Fathom"].
In Outlign, on the [Project] project:
1. Pull out every action item. For each one, tell me who owns it (us or the client), what it is, and any date mentioned.
2. Before creating anything, search the project for existing tasks that cover the same thing. If one exists, add the new detail as a comment on it instead.
3. For OUR actions: create an internal task in the right internal phase, with a clear title, the context from the call, and assign it to [Name] (or whoever was named on the call).
4. For actions the CLIENT needs to see or do: create a client-facing task in the matching client phase, written in plain, friendly language for the client. Link it to its internal task so both sides stay paired.
5. Add due dates wherever one was agreed.
Show me the list and where each task will go before you create anything. Then give me a summary of what you created, updated and linked.What the AI does: reads the transcript, sorts actions into internal and client, checks for duplicates, creates the tasks, then uses Outlign's linking so each client task is paired with its internal twin. Where the work starts internally, it creates the internal task first and copies it to the client side already linked.
Check before you approve: the client-facing wording. The AI writes it for the client, but you know the relationship. Also check that nothing internal (pricing chat, team notes) has slipped onto the client side.
When to use it: the client has signed. You want the project live in Outlign the same day, built from your templates, with the team added in.
What you need: the signed proposal (PDF upload, or connected Qwilr so your AI can find it), the client name, and who's on the team.
The prompt:
[Client] has signed. The proposal is [attached / in Qwilr, titled "[Proposal name]"].
Set up the project in Outlign:
1. Read the proposal and summarise the scope, deliverables, timeline and anything out of scope.
2. Check whether [Client] already exists as a client in Outlign. If not, create them.
3. Look through our Outlign templates and recommend the best fit for this scope. Tell me why, and flag any deliverables the template doesn't cover.
4. Once I confirm, create the project from that template, titled "[Client] - [Project name]".
5. Tailor it to the proposal: rename or add phases for deliverables the template is missing, remove tasks for things that are out of scope, and put the proposal's key details in the project info.
6. Set phase due dates from the proposal timeline, and add milestones for [kick-off / launch / other key dates].
7. Add [Name], [Name] and [Name] to the project, and assign the first phase's tasks to [Name].
8. Post a welcome message on the client side introducing the team and what happens first. Show me the draft before posting.What the AI does: reads the proposal, picks and justifies a template, builds the project, trims it to the actual scope, puts dates on the timeline and adds your team.
Good to know: Your AI can add people who already belong to your agency or the client's organisation in Outlign. Brand new people (a new client contact, a freelancer) need to be invited from Outlign first, then your AI can add them to the project.
When to use it: Friday afternoon, or before a WIP meeting. Run it per project, or across all your active projects at once.
What you need: just Outlign.
The prompt:
Give me a status check on [Project / all my active projects] in Outlign.
For each project:
1. What got done this week (look at recent activity and completed tasks).
2. What's overdue or due in the next 7 days, and who owns it.
3. Anything waiting on the client.
4. Any phase that's slipping against its due date.
Then draft a short, friendly update for the client covering progress, what's next, and anything we need from them. Keep internal detail out of it. Show me the draft, and once I approve, post it as a client-facing message on the project.What the AI does: reads recent activity, tasks, phases and milestones, gives you an internal snapshot, then writes a client-ready update and posts it to the client feed when you say go.
Variation: swap the last paragraph for "Post an internal message listing overdue tasks and who owns each one" to use it as a team nudge instead.
When to use it: the client has sent back a round of feedback on a design, website or document, usually as a long email, a marked-up PDF or a spreadsheet of comments.
What you need: the feedback (pasted, attached, or in a connected inbox) and the Outlign project.
The prompt:
[Client] has sent round [1/2/3] feedback on [deliverable]. It's [attached / in my inbox from [Name], subject "[Subject]"].
In Outlign, on [Project]:
1. Break the feedback into individual changes. Group them by page, section or deliverable.
2. Flag anything unclear, contradictory, or that looks out of scope, and list those separately for me to raise with the client.
3. Find the existing task for this deliverable. Add the grouped change list to it as a comment, rather than creating a new task for each change.
4. If a change needs its own task (it's big, or it's a different person's job), create an internal task, assign it to [Name], and link it to the client-facing task for this round.
5. On the client-facing task, add a short comment confirming we've received the feedback and when they can expect round [X+1].What the AI does: turns messy feedback into a clean, grouped change list on the right task, raises scope creep before it becomes free work, and keeps the client in the loop.
When to use it: at kick-off, or when dates move and you need to reshuffle the whole schedule (client delays content, launch date changes).
What you need: the new dates (from a proposal, a schedule doc, an email, or just typed in).
The prompt:
The timeline for [Project] in Outlign needs [setting up / updating]. [The launch date has moved to [date] / Here's the agreed schedule: paste or attach].
1. Show me the current phases, their due dates and any milestones.
2. Propose new phase due dates that work back from [launch date], keeping the same gaps between phases unless I've said otherwise. Allow [X] working days for client review at each stage.
3. Add or move milestones for [kick-off, content deadline, go-live, etc.].
4. List any tasks with due dates that now fall after their phase's due date.
Show me a before-and-after table first. Once I approve, update everything, then draft a client message explaining the new dates.What the AI does: sets phase due dates (what shows on the project timeline), adds milestones (the marker dates on the calendar), and catches tasks left stranded by the change. It knows the difference between task dates, phase dates and milestones, so ask for the one you mean.
When to use it: you want the thinking behind the work (a creative brief, workshop notes, a strategy summary) living in the project next to the tasks, not lost in someone's Drive.
What you need: your source material (transcript, workshop board, notes, proposal) and the project.
The prompt:
Using [the attached workshop notes / the transcript from today's [Client] call], write a [creative brief / meeting summary / strategy doc] for [Project] in Outlign.
Structure it as: [Background, Objectives, Audience, Key messages, Deliverables, Timing / or your own headings].
Save it as an internal document on the project. Then link to it in a comment on the [relevant task] so the team can find it.
If I ask for a client version later, create a separate client-facing document with only what the client should see.What the AI does: writes the document in your structure and saves it to the project as internal. Outlign documents are internal by default, and your AI won't make one client-visible unless you ask.
Tip: pair this with Playbook 1. One transcript can become a set of linked tasks and a meeting summary document in the same run.
The best playbooks come from things you already do every week. Pick a repeat job, write down the steps you'd give a new hire, and that's your prompt. A few habits that make them work:
Name the project and the side. "Client-facing Concept phase on the Acme website project" beats "the Concept phase".
Ask for the plan first. "Show me before you create anything" costs you ten seconds and saves cleaning up 20 wrong tasks.
Search before creating. Tell your AI to check for an existing task and comment on it. Duplicates are the fastest way to make a board messy.
Keep internal stuff internal. Say explicitly when something should be client-facing. Everything else stays on the team side.
Link, don't paste. Use linked tasks to connect internal and client work, rather than pasting task URLs into descriptions.
Save the good ones. Once a prompt works, save it in your AI tool (a Claude project or skill, a custom GPT, a saved prompt) so the whole team can run it with one line.
Got a playbook that's saving your team hours? Send it to us and we'll add it here.