Five Dot Workflows for a Marketing Team of One
If you are the whole marketing team, the work that slips is rarely the writing. It is the checking: whether the calendar still matches the plan, whether last week’s release got an announcement, whether the automation you built is actually ready to turn on.
OpenAI’s dots, launched September 29, 2026, are built for that kind of standing responsibility. This post gives five workflows to hand to one. Each is written as the four parts from what a dot workflow is: goal, tools, trigger and rules. The tools are from Marqeable’s MCP server, and setup is in connecting Marqeable to a dot.
We have not yet run these on a dot. Dots had not reached our account on launch day. The tool sequences below are the ones we run today from ChatGPT, Claude and Codex through the same connector. We will update this post with results from a dot.
The five at a glance
| Workflow | Trigger | Ends in |
|---|---|---|
| Monday plan check | Weekly schedule | A message, and a draft if there is a gap |
| Launch drafts | A release in another app, or a daily check | A campaign draft with pieces |
| Calendar keeper | Weekly schedule | Proposed calendar changes, for approval |
| Draft review | Daily schedule | Suggestions on drafts |
| Automation readiness | Daily schedule while a draft exists | A checklist of what is blocking |
None of them ends in a send.
1. The Monday plan check
| Part | Setting |
|---|---|
| Goal | Tell me whether the next four weeks of marketing match our current focus, and draft what is missing |
| Tools | get_business_context, get_content_calendar, list_campaigns, create_campaign_draft |
| Trigger | Mondays, start of the working day |
| Rules | Reads allowed. One campaign draft allowed per run |
The dot reads your focus and offers, reads the calendar, and compares. A product date with nothing planned around it is a gap. A quiet month with three campaigns stacked in one week is a different problem. It reports both and drafts only for the first.
This is the one to start with. It is almost entirely reads, and its output is a message you can judge in a minute. Marqeable’s own weekly plan does a version of this inside the product, described in the AI marketing agent weekly plan. The dot’s advantage is that it can also see your other connected apps.
2. Launch drafts when the product changes
| Part | Setting |
|---|---|
| Goal | When we ship something customers should hear about, draft the announcement campaign |
| Tools | Your release source, then get_business_context, get_automation_overview, create_campaign_draft, create_content_draft |
| Trigger | An event from the app where releases happen, or a daily check of it |
| Rules | Reads allowed. Drafts allowed. Image generation requires approval |
This is OpenAI’s own marketing example: a dot that revises launch material when the product changes. The trigger comes from the other app, such as your code repository or project board, if its plugin supports events. Otherwise the dot checks daily.
The step people skip is get_automation_overview. It shows which automations are live and how much contacts are already being messaged. A launch email that lands the same day as a nurture step is a worse launch. The dot should move the date and say why.
Content drafts are created copy first. The builder writes the copy and then waits for a person before any image work. Keep it that way.
3. The calendar keeper
| Part | Setting |
|---|---|
| Goal | Keep the marketing calendar accurate: our own dates, and the market dates worth planning around |
| Tools | get_content_calendar, add_calendar_entry, update_calendar_entry, adopt_suggested_date |
| Trigger | Weekly |
| Rules | Reads allowed. Every calendar change requires approval |
Calendar tools are different from the draft tools. A calendar change takes effect directly and feeds the next plan. So the dot proposes and you approve, one entry at a time.
A good run looks like this: “The conference moved to the 14th according to the organizer’s site. Update the entry? And there is a suggested date on the 30th that fits your audience. Add it?” Two questions, two taps in Slack.
4. The draft review pass
| Part | Setting |
|---|---|
| Goal | Before I review a draft, check it against its brief and our voice, and leave suggestions |
| Tools | search_content, get_content, get_business_context, add_content_suggestion |
| Trigger | Daily |
| Rules | Reads allowed. Suggestions allowed |
The dot finds drafts, reads each against its brief, and attaches suggestions anchored to the exact text. Each suggestion carries a reason. You accept or dismiss them in Marqeable when you open the piece.
It cannot rewrite the piece, and it cannot touch content that is already live. A suggestion is the smallest write there is, which makes this a good second workflow.
5. The automation readiness check
| Part | Setting |
|---|---|
| Goal | For every automation in draft, tell me exactly what stops it from being turned on |
| Tools | get_automation_overview, check_automation_readiness, inspect_landing_pages, create_automation_step_content |
| Trigger | Daily, while any automation is in draft |
| Rules | Reads allowed. Step content drafts allowed. Page creation requires approval |
Automations stall in draft for small reasons: a step with no content, a form page that is not published, a form with no email or phone field. The readiness check returns each check with whether it passes and why not. The dot turns that into a short list, and drafts content for empty steps if you allow it.
Turning the automation on is done by a person, in the app. There is no tool for it.
What all five have in common
They read far more than they write. Most calls in every workflow are reads. That is what makes background work tolerable.
They end at a person. A message, a draft, a suggestion or a checklist. The decision is yours.
They run on a schedule. Marqeable’s connector does not send events as of September 2026, so none of these waits on Marqeable to wake the dot.
A workflow is ready to run unattended when its output has stopped surprising you. Until then, keep draft creation on approval and read the Activity View after each run.
For the limits of this approach, and what should stay with a purpose-built engine, see dots do not replace your marketing automation.
Frequently asked questions
What marketing work is a dot best at?
Recurring work that is mostly reading and checking, with a small amount of drafting at the end: comparing the calendar against the plan, drafting launch material when something ships, reviewing drafts against the brief, and checking that an automation is ready before a person launches it.
Can these workflows run without me?
The reading and drafting can. Every workflow here ends in a draft or a note for a person, and none ends in a send. That is by design: the dot does the preparation, and the decision to publish stays with you.
Do I need MCP Events for these workflows?
No. All five run on a schedule. Event triggers depend on the connected app supporting the proposed MCP Events specification. Marqeable’s connector does not send events today, so the examples use scheduled runs or an event from another app, such as a code repository.
Have you run these on a dot?
Not yet. As of launch day, September 29, 2026, dots had not reached our account. The tool sequences are the ones we run today from ChatGPT, Claude and Codex through the same connector. We will update this post with results from a dot.
The bottom line
Start with one workflow, not five. The Monday plan check is the safest: it is all reads, it ends in a message to you, and within two weeks you will know whether the dot’s judgment is worth trusting with drafts. Add the next one only when the first has stopped surprising you.
See what the drafts become: Marqeable’s agents build and run your campaigns, run behavior-triggered automations across text and email, answer and qualify every visitor with AI website chat, and tie revenue to the exact message with attribution.
Marqeable runs your campaigns, answers every visitor, text, and email in seconds, and turns them into booked jobs and meetings - even at 9pm on a Saturday. We’re in private beta with a small early cohort. Get early access
