Learn · Multi-client operations

How do you keep an AI task in the correct client account?

Last updated:

In short

Outloop is Workspace infrastructure for deploying and operating AI workers.

Tie the work to its intended client Workspace, approved accounts and specific resources. Outloop’s recorded Google Ads and Meta evidence shows approved requests succeeding and out-of-scope requests being refused. A successful login alone does not establish the right client target.

Open in: ChatGPTClaudePerplexity
An agent in Client A's workspace correctly acting on its own client, while a request crossing into Client B's workspace is blocked by a shield before any backend call
The right key, aimed at the wrong client, is still the wrong outcome — every request proves which client it's for before the backend call runs.

Client AWorkspace

Context
Client A’s brief & rules
Accounts
Google Ads A · Meta A
Resources
CRM A · Drive A
Browser
Approved identity A
Access
Approved for Client A

Work stays with this client.

Client BWorkspace

Context
Client B’s brief & rules
Accounts
Google Ads B · Meta B
Resources
CRM B · Drive B
Browser
Approved identity B
Access
Approved for Client B

Work stays with this client.

Illustrative client setup. Each service and resource needs its own approved access; Local Chrome uses the Workspace’s configured profile; Managed Browser uses a shared profile with approved identity and access. Diagram, not recorded execution.

Where do wrong-client mistakes start?

A copied brief still names the previous customer. An agent opens the wrong ad account under a manager login. A report is saved to the wrong client folder. These mistakes can happen with valid access and a plausible-looking result.

Repeated or scheduled tasks make checking the setup especially important: the next run may repeat the same mistake.

What should you configure before the first run?

  1. Create one canonical Workspace for the client.
  2. Add that client’s approved Context and working methods.
  3. Approve the intended accounts and resources, including targets behind shared agency logins.
  4. Confirm the browser identity and the resource shown in the website.
  5. Verify a safe allowed action and an out-of-scope refusal.
  6. Define human review and check the actual output before treating work as complete.

Real allowed and refused actions

Outloop’s Live Activity from one demo workspace running Google Ads and Meta Ads work with approved client access. Approved requests execute; requests for an unapproved resource are refused.

Outloop Live Activity for demo workspace 018: Google Ads and Meta Ads requests, each marked Allow with result OK, including grouped batches of 10 and 14 requests.

Swipe sideways to read the log · open full size

Allowed: approved API work executes

  • Demo Workspace 018 · Google Ads and Meta Ads requests
  • Every row: Allow · OK — including grouped batches of 10 and 14 requests
  • The approved requests returned results in this client Workspace
Outloop Live Activity for demo workspace 018: allowed google_ads and meta_ads requests interleaved with three refused requests — two marked Deny with RESOURCE_ID_NOT_ALLOWED (one google_ads, one meta_ads) and one meta_ads request marked Deny with META_OBJECT_NOT_APPROVED.

Swipe sideways to read the log · open full size

Refused: these wrong-resource requests did not run

  • Deny · RESOURCE_ID_NOT_ALLOWED (google_ads and meta_ads) — the request named an account outside this Workspace’s approved resources; the requested action was refused
  • Deny · META_OBJECT_NOT_APPROVED (meta_ads) — a campaign, ad set or ad that does not belong to the approved ad account; the requested action is not sent to Meta
  • Approved requests in the same workspace keep executing around the refusals — Allow · OK

Real execution in an Outloop demo workspace (workspace 018), published in the September 26, 2026 proof pack — not a customer case study, and not evidence of ad spend, delivery or performance. Enforcement covers requests routed through Outloop. See the runtime access walkthrough for the recorded flow.

What if the login legitimately reaches several clients?

Then the resource matters as much as the identity. A Google Ads manager account can reach multiple customer accounts; the current task is still approved for particular client resources. Read the shared-login and resource-boundary guide.

What are the limits?

Correct setup matters: approving the wrong resource does not become correct because the software follows the configuration. This proof covers access through Outloop, not unrelated tools. It does not decide whether an allowed budget change, recommendation or creative choice is a good business decision.

Start with the first-Workspace setup guide. For the broader operating model, see what an AI Workspace keeps and Outloop for multi-client agencies.

The intended client, approved action and reviewable result

  1. 01

    Client task

    Start with the intended client and a defined result.

  2. 02

    Approved Workspace

    Use this client’s Context, accounts and resources.

  3. 03

    Supported action

    Work through the approved API, browser or file path.

  4. 04

    Reviewable result

    Inspect the output and involve a person where required.

  5. 05

    Work history

    Keep the work record with the intended Workspace.

Conceptual workflow. Supported actions depend on the configured service, runtime and approved resources.

See the boundary before another loop runs.

Follow a workspace-scoped request and inspect the recorded denials. This shows the access decision behind the wrong-client risk explained above.

This is evidence of the shown requests, not a universal isolation test.

  • Workspace-scoped action requests
  • Labeled dashboard views and recorded denials
  • Redacted API evidence
Start 14-day trialDownload for Mac

Create your trial, then install on your company-controlled Mac. Set up your first workspace.

56 seconds · One request, the right client · No form required

The opening is an architecture illustration. Dashboard views and recorded denials are labeled in the video; API evidence is redacted. Shown campaign execution stays PAUSED.

Open video directlyRead the English transcriptExplore the complete 8:53 workflow →

What Outloop enforces — and where the boundary is

What Outloop enforces

  • Keep supported work tied to the configured client accounts and resources.
  • Retain approved client Context and reviewable work records.

What it does not guarantee

  • Support depends on the service, runtime and configuration.
  • Complete session, provider and machine migration is not promised.

What you still configure

  • Configure approved accounts, resources and Context separately for each Workspace.
  • Validate the client boundary and output before expanding.

Is this for you?

A good fit if…

  • You deploy repeatable AI work into real client systems.
  • You need separate client context, resources and review.

You may not need Outloop if…

  • You need universal model or machine migration today.
  • You need automatic deployment cloning or automatic cost routing.
Frequently Asked Questions

Wrong-client access in agent loops — FAQ