A launch date moves, the offer changes, and the campaign drafts still tell the old story. Someone has to reconcile the landing page, email copy and social posts before the team can ship. OpenAI Dots invites a practical question for business owners and marketing leads: which ongoing responsibilities would be worth handing over?

What are OpenAI Dots?

OpenAI introduced Dots on September 29, 2026, as agents that can continue working between conversations. They have a cloud computer and use the apps you connect, with feedback helping them learn your preferences. The official Dots announcement describes the product and its intended role.

Our operational takeaway is to start with a responsibility you can define and a result you can assess. A useful assignment gives the reviewer a shared standard. Asking an agent to handle all your marketing leaves too many choices unresolved, from the target audience to the authority to change an offer.

Before testing a workflow, write down the context your colleagues usually carry in their heads: the current product story, the source of approved claims, the person who signs off on pricing, and the location of the latest files. This exercise can expose conflicting assumptions before they become instructions.

Three marketing workflows worth testing

These are pilot scenarios proposed by La Fábrica. Their feasibility depends on the tools and permissions available in your account. Treat them as test designs with defined deliverables and reviewers.

• Keep launch materials aligned. Start with an approved offer sheet and request a review of draft emails, landing copy and social posts. Ask for a list of outdated statements, an explanation of each conflict and suggested edits. Your brand lead reviews any change to a customer promise before the revised copy moves forward.
• Build useful content from an interview. Provide an authorized transcript and request an article outline plus channel-specific drafts. Specify which quotations need checking against the recording and which details should be anonymized. Assess whether each asset contributes something useful while keeping the speaker's meaning intact.
• Prepare a client conversation. Use approved account notes and product documents to request a summary of needs, unanswered questions and missing materials. Require a source for each factual statement and a visible label for unknowns. Your sales team decides what commitments it is ready to make.

Choose the scenario that addresses a visible bottleneck. If approvals are already backed up, generating more drafts may add review work. A good starting point is a deliverable you can compare with a recent example from your existing process.

For a regional team, choose a market as well as a channel. An English campaign aimed at an international buyer may need a different opening, proof point and call to action from a Spanish campaign for local customers. Include that editorial decision in the assignment.

Write a brief with a clear finish line

We recommend a short pilot brief covering the goal, permitted inputs, expected output, deadline and reviewer. Add approved examples and explain why they work. A reference file shows a style; a short explanation helps communicate the judgment behind that style.

If the business serves several countries, include the market, language variant, currency and offer restrictions. Keep those details in an authoritative source. Ask for missing commercial information to be flagged so the reviewer can resolve it before anyone relies on the draft.

A first assignment could look like this:

“Review these campaign drafts against the attached approved offer sheet. Return a list of inconsistencies and suggested edits, with a link to each document. Keep the prices and conditions in the offer sheet. Flag unresolved questions for me. Prepare the work for my review and ask for my approval before publishing or sending any materials.”

Define completion in terms the team can check: identified documents, traceable versions, accessible sources and visible open questions. Decide where the reviewer will receive the work and what signals that it has been accepted. This avoids mistaking a polished draft for a finished handoff.

Build permissions and review into the workflow

OpenAI's Dots safety explanation says proactive research uses read-only tools. Those tools can gather permitted information but cannot send messages or change connected app content. Follow-up actions remain subject to action rules and checks, including action review and custom rules.

For a pilot, we recommend documenting what may be read, what may be drafted and what requires a named reviewer. Assign responsibility for publishing and for commercial conditions. Give the team a route for resolving uncertainty, so a decision does not depend on who happened to see a notification first.

Review data handling before connecting client information. The official privacy and safety FAQs say disconnecting an app stops new access without deleting context already obtained. Business, Enterprise and Edu workspace data is not used for training by default; personal plans follow the applicable data controls.

Start with project materials that have a clear owner and are authorized for the intended use. Record when key documents were last reviewed. If permission to use a file is still unresolved, settle that before adding it to the pilot. Good input management also makes the output easier to assess.

Check availability before promising a workflow

When checked on October 23, 2026, OpenAI's getting-started guide describes a gradual rollout to eligible Pro and Business Premium users. Enterprise workspaces can try the beta when an administrator enables it. The first dot is included in Pro or Business Premium, with an allowance for deeper work. Account and workspace eligibility still apply.

Confirm access, required connections and applicable limits before putting the workflow on a client delivery schedule. Test it with representative materials from your business so the team understands the actual handoff and review effort.

Run a small pilot and assess the whole process

Our suggested starting point is one workflow over one week. This is a planning recommendation from La Fábrica. Keep a recent sample of the existing process as your comparison, and count the effort spent preparing inputs and reviewing outputs.

1. Pick one deliverable and name the person who will accept it.
2. Gather current sources, approved examples and action boundaries.
3. Request a first version and log the corrections it needs.
4. Repeat the same type of assignment with that feedback included.
5. Compare total effort, substantial corrections and usefulness to the team.

If fixing the result takes longer than preparing the work, revisit scope and context before expanding the pilot. If the deliverable arrives with clear sources and fewer inconsistencies, test one additional responsibility. The business decision is whether this form of delegation improves execution and gives people more room for brand judgment, customer conversations and decisions that need their experience.