
How to build an AI social media management suite
Build a social media workflow with AI employees, automation flows and shared apps for planning, content review, publishing and performance reporting.
KyoubeAI · · 11 min read
Use cases
Campaigns planned, produced, published and measured from one place

This suite is built from parts KyoubeAI runs today: AI employees, data tables, apps, connections and routines. Explore its applications and workflows, then talk to us about scope, connected systems and licensing for your team.
Marketing in a small company is one or two people doing the work of eight: planning, writing, designing, posting, emailing, running ads, reading the numbers, and reporting on all of it. The result is that the calendar slips, the numbers get read once a month, and nobody can say which campaign produced which customer.
The Marketing Supercube gives that team the eight. It seeds tables for the brand guide, audiences, campaigns, the content calendar, assets, channels, posts, emails, ad spend and attribution. It declares agents that plan the calendar from the campaign brief, draft and adapt content for each channel, check everything against the brand guide, schedule publication, pull performance from every channel into one table, and attribute leads back to what produced them. It publishes the apps a marketing team lives in: the calendar, the campaign page, the brand review queue and the performance dashboard.
Once it is running, a product launch starts as a brief in the campaign app. The Planner agent turns it into a calendar of blog posts, emails, social posts and ad sets with dates and owners. The Writer agent drafts each piece in the brand voice and the Brand Reviewer agent marks anything off-voice or making an unsupported claim before a person sees it. The marketing lead approves the pieces, the Publisher schedules them, and two weeks later the Analyst reports which piece drove sign-ups and what it cost per lead, with the leads themselves handed to the Sales Supercube.
Companies with a marketing function of one to ten people, from a founder doing marketing on Fridays to a small team with a content lead, a performance lead and a designer. Roles that use it are the marketing lead, content writers, designers, whoever runs ads, and the sales lead who receives the leads.
| Agent | Role | Owns | Harness preference |
|---|---|---|---|
| Planner | general | Campaign briefs into calendars; channel mix; owner and date assignment | Claude Code |
| Writer | general | Drafts for blog, email, social and landing pages, adapted per channel, in the brand voice | Claude Code, then pi |
| Brand Reviewer | general | Voice, terminology, claims and disclaimers checked before anything is approved | Claude Code |
| Publisher | general | Scheduling and publishing to connected channels; UTM discipline; asset handling | Claude Code, then Hermes |
| Analyst | analyst | Channel performance, attribution, cost per lead, the weekly and monthly report | Claude Code |
| Table | Purpose | Key fields |
|---|---|---|
brand_guide | Voice, tone, terminology, banned phrases, claim rules, disclaimers | section, rule, examples, severity |
audiences | Segments and personas with the messaging pillar for each | name, definition, pillar, channels |
campaigns | Each campaign's brief, objective, budget and dates | name, objective, audience, budget, start, end, status, owner |
calendar_items | Every planned piece of content or send | campaign, channel, title, owner, due, publish_at, status |
content | Drafts and final copy, versioned | calendar_item, version, body, review_status, reviewer_notes |
assets | Images, video, documents, with rights and usage | name, file, source, rights, used_in |
channels | Connected channels and their posting rules | name, kind, account, cadence, limits |
posts | Published social posts and their metrics | channel, content, published_at, url, impressions, engagement |
emails | Sends and their metrics | campaign, audience, subject, sent_at, delivered, opens, clicks, unsubscribes |
ad_spend | Daily spend and results per ad set | platform, campaign, ad_set, date, spend, impressions, clicks, conversions |
leads | Leads captured, shared with Sales | source, campaign, content, contact, captured_at, status |
attribution | Touchpoints per lead and the model applied | lead, touchpoints, model, credited_campaign |
| Group | Target systems | Flows in | Flows out |
|---|---|---|---|
| Email marketing | HubSpot, Mailchimp, Klaviyo | Audiences, send metrics | Campaigns and templates, scheduled sends |
| Social | LinkedIn, X, Instagram, Meta Pages | Post metrics, comments | Scheduled posts with assets |
| Advertising | Google Ads, Meta Ads, LinkedIn Ads | Spend, impressions, clicks, conversions | Budget change proposals, never applied without approval |
| Analytics | Google Analytics, Google Search Console | Sessions, conversions, queries, rankings | Nothing |
| Web | WordPress, Webflow | Published pages | Drafted posts and landing pages, published on approval |
| Design | Canva, Figma | Assets | Nothing |
| Chat | Slack, Microsoft Teams | Feedback on drafts | Review requests, weekly report |
| Sales | Via the Sales Supercube | Deal outcomes | Qualified leads |
Brief to calendar Trigger: a campaign is created with a brief. Steps: the Planner reads the brief, the audience and the channel rules, proposes the content mix and cadence, and creates calendar items with owners and dates. Checkpoint: the marketing lead edits and approves the calendar. Output: a campaign with every piece scheduled.
Draft, review, approve Trigger: a calendar item's due date is five working days away. Steps: the Writer drafts the piece for its channel with the brand voice and the messaging pillar; the Brand Reviewer checks voice, terminology, claims and required disclaimers and annotates the draft; the piece lands in the review queue. Checkpoint: a person approves every piece. Output: approved content, ready to publish.
Publish Trigger: an approved piece reaches its publish time. Steps: the Publisher attaches assets, adds UTM parameters from the campaign, posts through the channel connector, and records the URL. Checkpoint: none for pre-approved pieces; an admin can require a final confirmation per channel. Output: published, tracked.
Weekly performance report Trigger: routine, Monday 07:30. Steps: the Analyst pulls every channel's metrics and ad spend, updates the attribution table from new leads and touchpoints, computes cost per lead by campaign, and writes the wins, misses and three recommendations. Checkpoint: none. Output: report in the dashboard and the marketing channel; lead counts to General Management.
Lead handoff Trigger: a lead is captured. Steps: the Analyst records source, campaign and content; the lead is handed to the Sales Supercube's SDR with its touchpoint history. Checkpoint: none. Output: Sales sees where every lead came from, Marketing sees what every lead became.
SEO refresh Trigger: routine, monthly. Steps: the Analyst reads Search Console for pages losing position and queries gaining volume, proposes which posts to update and which new topics to plan; the Writer drafts the updates into the review queue. Checkpoint: the marketing lead approves. Output: calendar items for refreshes and new content.
SaaS company. Campaigns are feature launches and comparison content; the Brand Reviewer enforces claim rules against competitors. Attribution runs to trial sign-ups and then to paid conversions from the Sales Supercube. Product usage events from the app are a channel in their own right.
E-commerce brand. Klaviyo flows and Meta Ads dominate; the Analyst reports return on ad spend by product from the Supply Chain Supercube's order data and flags products to stop advertising when stock is low. The calendar follows the promotional calendar and drops.
Professional services firm. Thought leadership on LinkedIn and email newsletters; the Writer works from partner interviews recorded as notes. Leads are event registrations and downloads, handed to the partners through Sales. The Brand Reviewer enforces regulatory wording where the firm is regulated.
Hospitality group. One brand guide per venue; local social channels per site; the Publisher posts menus, events and offers on a weekly cadence. The Analyst ties bookings from the booking system to campaigns.
Education provider. Campaigns follow the enrolment cycle; audiences are prospective students and parents by programme; the Brand Reviewer enforces accuracy on fees and outcomes claims. Leads are enquiries handed to admissions in Sales.
Local services business. Google Business Profile posts, review responses drafted by the Writer for the owner to approve, and a monthly email. The Analyst reports cost per booked job from Google Ads.
blocker finding, for example an unsupported claim
or a missing disclaimer, stops the piece reaching the approve button.write on the suite's tables; the Analyst has read-only SQL for reporting; channel connectors
run under tool profiles listing exactly the operations allowed.Step-by-step guides that set up part of this work on KyoubeAI.

Build a social media workflow with AI employees, automation flows and shared apps for planning, content review, publishing and performance reporting.
KyoubeAI · · 11 min read

Start an AI marketing team with three agents for blog and search, social video and outbound, using shared briefs and tracked work in KyoubeAI.
KyoubeAI · · 8 min read
Suites share tables and compose, so the ones next to this one are the ones it works with.

The operating rhythm of the company, run by agents, owned by the leadership team
Goals, KPIs, meeting cadence and decisions in one place, with agents that prepare every review and chase every action so leaders spend their time deciding, not collecting.

Books that close themselves, cash you can see, approvals that stay with people
Accounts payable, receivable, month-end close, cash forecasting and management reporting run by agents against your own ledger data, with every payment and every journal approved by a person.

A pipeline that works itself while your people talk to customers
Lead qualification, CRM hygiene, quoting, follow-up and forecasting handled by agents in your own customer tables, so salespeople sell and managers see a forecast they can trust.