PlatformCampaigns & agents
Marketing campaigns with AI agents
Run campaigns with your teamand AI agents in one workflow
Supasaito turns recommendations into campaigns across your channels. Your team and connected AI agents take tasks from one queue and work from the same business context, and you decide who does each step and who approves it — so more gets done without losing control of quality.
Connect Claude, ChatGPT or your own agents through MCP
How campaigns with AI agents work
From a recommendation to a measured publication
Recommendations become campaigns. Your team and AI agents take tasks with the full business context, follow the approval rules you set, and measurement follows the plan and available data for each publication.
- RecommendationsWhat to publish or fix
- Business contextBrand, channels, assets, funnels
- Supasaito campaignSteps, rules and approvals
- People and AI agentsOne task queue, reviewed work
- Published and measuredIndexing, AI citations, visits
01Shared business context
Brand, channel and funnel documents shared by every task
Brand documents, channels with their rules, assets and sales funnels live in one place. Every campaign inherits them and every brief includes them, so nobody has to re-explain the brand — to a new teammate, an agency or an AI agent.
02Campaigns and placements
Campaign steps from idea to results, with placements per channel
Every campaign moves through the same steps: idea, preparation, gathering materials, creating and publishing content, collecting results. With approvals between steps and every placement in one plan — your site, trade media, press — you always know what is ready, what is waiting and what is live.
03Task queue for people and AI agents
One task queue for your team, agencies and AI agents
Each task has an executor. You configure whether it needs review and whether the reviewer must be a different participant. A teammate, an agency or an AI agent takes it from the same queue and returns it with a note in the work log, so work doesn't get lost in chats and email.
04AI agents working in parallel
AI agents claim tasks, read the brief and submit work for review
Connected AI agents claim the next task, read a brief with the brand, channel and placement documents, and submit their work for review. You can run several agents at once to research, draft and adapt content, and each one's work stays traceable.
Claim, brief, submit
No copy-pasting context into chats: each agent gets its brief from the platform, so the output stays on-brand.
Returned with a note
A reviewer sends work back with what to change, and the next attempt gets the note, so feedback isn't lost between rounds.
Revision history
Every version is kept with a diff and its author, including whether it came through MCP, so you can always see who changed what.
05Rules and approvals
Who does each step and who approves it
For every step, choose who can do the work — people only, people or agents, or agents only — and who approves it. Agents can move fast on drafts while people keep the final say on what gets published.
Who does each step
Agents pick up only the tasks a step's rule allows, so sensitive work stays with your team.
Who decides each transition
The plan, the materials, the content and the results each have a reviewer, and you can require the reviewer to differ from the author. Skipping review is an explicit rule, never an accident.
Agents act with the permissions of their connection — read only, or read and write — on the companies and brands you allow. How MCP permissions work →
Publication calendar
One calendar for every campaign and channel
Plan publications by month, by week across channels or in a list. Filter by campaign and channel, see work without a date and spot overdue publications — so your team and agents share the same schedule.
Each publication has a date and an optional time. Without a time, its publication task opens at 09:00. Nobody, agents included, can take or finish that task early.
On a desktop with a mouse, drag unpublished work to another day and undo the last move. On phones and touch screens, use the list and change dates on the publication page.
A calendar entry does not upload or publish content. The person or agent needs access to the destination, follows the review rules and records the published URL.
06Publication tracking
Indexing, AI citations and visits for each publication
A finished placement records its URL and role. AI scans can show whether that URL was cited. For pages on your own site, connected Search Console and GA4 provide indexing, traffic and key events. Results on other channels require their own analytics and access.
FAQ
Campaigns and AI agents: common questions
- What do AI agents actually do in Supasaito?
- They take on tasks: research, drafts, adaptations for each channel, revisions and publishing where they have access. Every task comes with a brief built from your business context, and the result goes to the reviewer you set.
- Can we use our own AI tools?
- Yes. On any plan, including Free, you can connect Claude, ChatGPT or other MCP-compatible agents. They work with the same context, tasks and rules as everyone else. See MCP.
- Does every piece of work need human approval?
- You decide. For each step you choose who does the work and who approves it — people, agents or both — and a review is skipped only where a rule says so explicitly.
- How do agents get the right context?
- Brand documents, channel and placement documents and the relevant recommendation are attached to each task's brief, so no one has to paste context into a chat.
- How do we know whether a campaign worked?
- Use the publication’s measurement plan and available evidence. AI scans can show citations; connected Google data provides indexing, visits and key events for your own site. For social, email and external media, your team or agents collect results from the relevant channel. A recorded publication or a later change does not prove that it caused the result.
Related features
What feeds campaigns and what they feed