Dock for marketing is the workspace where the brief lives, the agent drafts copy against it, brand approves variants, and the send platform receives a clean, signed payload. Teams already run AI across briefing, copywriting, subject-line testing, and performance pulls. The cascade breaks at handoff: the brief sits in a doc, drafts sit in Slack, the approval lives in someone's head, and the campaign ships with no record of who decided what. Dock closes that gap by giving the agent its own persistent table.
HubSpot Marketing Hub, Marketo, Salesforce Marketing Cloud, Mailchimp, and Iterable stay the system of record for the raw marketing data: contacts, lists, send events, opens, clicks, bounces, attribution. Dock is the system of record for what the agent interprets from that data: the prioritized brief, the chosen copy variant, the brand reviewer's sign-off, and the audit log of every change. Each Dock row carries a pointer back to the platform record, typically hubspot_campaign_id or marketo_program_id, alongside agent identity, decision, reviewer, and timestamp. The agent re-fetches platform data via fresh API reads when it needs current open rates or list counts. Dock holds the persistent interpretive layer that survives across sessions.
One surface: the Campaigns table
| campaign_id | brief_summary | agent_draft_variant | brand_status | reviewer | hubspot_campaign_id | decided_at |
|---|---|---|---|---|---|---|
| cmp_4471 | Q3 launch, ICP=RevOps leads, tone=technical | Subject A: "Your pipeline, audited." Body v3. | approved | ana@ | 18772341 | 2026-05-28 14:02 |
| cmp_4472 | Reactivation, dormant 90d trial users | Subject B: "Three things changed since you left." Body v2. | needs_revision | dan@ | 18772398 | 2026-05-29 09:41 |
| cmp_4473 | Field event follow-up, SF Summit attendees | Subject C: "We met at booth 14." Body v1. | pending | unassigned | 18772410 | 2026-05-30 08:15 |
A worked workflow
The marketing agent reads the brief row for cmp_4472, pulls the dormant-trial segment count from HubSpot via a fresh API read, drafts three subject lines and two body variants, and writes them back to the row. Brand reviewer Dan opens the row, flags the subject as too casual, and flips brand_status to needs_revision. The agent re-drafts against the note, Dan approves, and only then does the consent gate fire: the agent stages the approved variant in HubSpot as a draft campaign and returns the hubspot_campaign_id. Send is a human click. Dock keeps the trail of which brief produced which copy, who approved it, and when. That is the marketing version of the dangerous-ops contract: draft freely, stage on approval, send by hand.
Why it matters
HubSpot's 2026 State of Marketing report finds that 61 percent of marketers believe their function is in its biggest disruption in twenty years because of AI (HubSpot). Gartner's annual CMO Spend Survey has tracked marketing budgets compressing for three straight years, which means every dollar of agent leverage matters more (Gartner). Forrester argues the engagement metrics marketers used to justify spend are becoming untenable as AI search reshapes accountability (Forrester). The teams that win in this environment are not the ones with the most aggressive AI; they are the ones whose AI leaves a reviewable trail.
A Dock workspace gives that trail a home. The brief, the draft, the reviewer note, and the approval each get a row, a timestamp, and a signed agent identity. When the CMO asks why a subject line went out, the answer is a permalink. When compliance asks who approved a regulated claim, the answer is an agent audit row. The send platform still owns delivery. Dock owns the decision layer above it.
Start a Dock workspace for your marketing team.
FAQ
Does Dock replace HubSpot or Marketo? No. HubSpot, Marketo, Salesforce Marketing Cloud, Mailchimp, and Iterable remain the system of record for contacts, sends, and analytics. Dock holds the brief, the agent-drafted variants, and the brand approvals that produced the send.
How is this different from a Notion brief plus a Slack approval? Notion and Slack do not give the agent its own row with a pointer back to hubspot_campaign_id, a signed agent identity on every write, or an audit log of every revision. Dock does. The same workspace pattern shows up in Dock for SEO and Dock for sales.
What stops the agent from sending a campaign on its own? The consent gate. The agent can draft and stage in HubSpot, but the actual send is a human action behind a separate approval. This is the marketing version of the pattern outlined in AI sales prospecting that actually compounds.
Which marketing tasks fit this model first? Brief intake, copy variant generation, subject-line testing, segment definition, and post-send performance summaries. Brand-sensitive output stays in Dock until a human reviewer signs it.
