Welcome to Outloop.
In the next few minutes, we'll create your first Workspace, connect a
client's advertising accounts, use your AI agent to connect a new service,
and give Ollie its first real task.
We'll also show how Ollie can pick up new requests automatically, so you
don't have to start every task yourself.
I'll show you where to click, what to choose, and how to know it's working.
More clients.
Same team.
Everything starts with a Workspace.
Think of it as one home for one client or project.
In Outloop, click Create workspace.
Give it a name, choose your AI environment, and click Create and connect.
That's it.
Outloop creates the local folder and prepares the instructions your AI agents need.
Claude Code, Codex and Hermes can work with this environment.
No separate Claude Cowork project or manual folder connection is needed.
Now open the Workspace Context.
This is where you keep the client's goals, key performance indicators, reference
folders and working instructions.
You can edit these yourself or ask an authorized agent to help you update them.
Next, click AI Worker.
Meet Ollie.
Open Settings, choose the engine, select a Work owner and your Reviewers,
and confirm their permissions.
Once Ollie shows Active, you can give it work directly.
Want Ollie to receive tasks by email too?
Select your connected mailbox, choose which clients or team members can
send requests, and confirm activation.
Now Ollie can pick up authorized requests without waiting for
you to open a conversation.
Let's give Ollie access to the services it needs.
Here's where Outloop becomes especially useful for agencies.
Let's start with Google Ads.
Your agency connects its Google Ads manager account once.
Now open another client's Workspace.
Go to API Access and choose Assign shared key.
Select your existing Google Ads connection.
Enter the client's customer account ID, review the account access, and click Assign.
Done.
The new Workspace can now use that approved client account
without reconnecting Google Ads or entering the credential again.
Meta Ads works with the same idea.
Select the existing agency connection, choose the client's advertising
account, and assign it to the Workspace.
You can do the same with eligible Google Analytics properties
and other supported services.
Need a separate credential for one particular client?
Choose the dedicated option instead, and assign that connection
only to the intended Workspace.
Let's check Live Activity.
Here you can see the service requests and the Workspace they belong to.
That's the advantage.
Connect your agency services once.
Assign each client's accounts.
Start working.
What if the service you need isn't already in Outloop?
That's where our Custom API engine comes in.
Open the Outloop website's Skills Catalog.
Find the Custom API Setup Skill, download the package, and return to Outloop.
Open Skills, choose Import, and select the downloaded package.
Set it up as a Global Skill.
Now your configured Global Skill is available across your Workspaces,
while each Workspace keeps its own approved service access.
Let's try it.
Open a conversation with your authorized agent and ask:
"Connect PandaDoc to this Workspace using the Custom API Setup Skill."
The agent researches PandaDoc's official documentation and prepares the connection,
including its destination, authentication method and supported operations.
When a credential is required, enter it directly into Outloop's secure form.
Never paste it into the conversation.
Outloop keeps the credential in your Mac's Keychain, and the agent works
through the approved connection.
Now let the agent validate the setup and perform a real test.
Here is the result.
The PandaDoc request succeeded, and we can inspect the matching record in Outloop.
Your agent can also perform approved write operations when the service
and Workspace permissions allow them.
So you're not limited to a fixed list of integrations.
With Outloop and the right Skills, your agent can help connect almost
any service with a compatible API.
And if you want a Skill customized for one client, open that
Workspace's Skills and create or update its dedicated instructions.
Now let's put everything together.
Return to AI Worker and open Ollie.
Type your request, and give Ollie the information it needs.
Here, we share one week of synthetic campaign figures for a demo skincare
brand, and ask for a short review and an action plan for our account manager.
Click Send.
Ollie opens the conversation and starts working.
It works inside this Workspace, with its Context and Skills.
Let's follow the result.
Ollie calculated the return on ad spend for each channel, compared it with the
target, and prepared an action plan.
It also separates what the figures show from what still needs investigation.
Ollie works within the access you approved.
When an action needs a person's decision, it asks your Reviewer first.
Once the work finishes, open Conversations.
Here you'll find the request and the recorded outcome.
You can return to Ollie, or open a new chat in Claude Code when
you want a more hands-on session.
But here's where Ollie becomes more than a chatbot.
Open Routines.
Routines let you automate work instead of requesting the same task again and again.
You can schedule a weekly performance report, or use a supported trigger
to start work when something happens.
Remember the email requests we enabled earlier?
That's a practical example.
When an authorized client or team member emails a request, Ollie
intake picks it up automatically.
It works in this client's Workspace, uses its Context and approved
tools, and starts the work.
You don't need to open a chat or manually start every request.
And you can follow the work and its results in Conversations.
That's the difference between chatting with an AI and having an AI worker
that keeps your business moving.
And that's your first setup.
One Workspace for each client.
Approved advertising accounts you can assign in just a few clicks.
Global and dedicated Skills that help your AI worker follow your processes.
A Custom API engine that helps your agent connect more services.
And Ollie, ready to handle direct requests, scheduled work and
authorized incoming tasks using your client's Context and approved access.
Your team stays in control of the important decisions, while Ollie
handles more of the execution.
More clients.
Same team.
Below this video, you'll find the step-by-step instructions, Skills
Catalog, full Custom API walkthrough and detailed connector guides.
Welcome to Outloop.