Skip to content
Tasklet

Page7 · AI social media operations platform

A week of social posts, checked and waiting for approval

An AI social media platform that researches a brand, drafts Instagram and Facebook posts natively in Arabic and English, runs risk checks, and never publishes without a human sign-off.

Full-stack product engineering
Page7
Page7 demonstration turning a business description into a week of social posts

What we did, plainly

We engineered the marketing site, the signed-in product and the FastAPI backend: brand research, content AI, cultural calendar checks, approval queues, review links and publishing integrations. Page7 is pre-launch; screens below are public product demonstrations with sample data, not customer accounts.

23
backend feature modules
30
frontend application routes
2
languages drafted natively today
0
customers claimed — pre-launch

Module and route counts from the codebase in November 2026. Customer count is intentionally zero: Page7’s own site states it is pre-launch.

The problem

Busy local businesses need a steady stream of social posts without hiring a full marketing team. Generic AI writers invent a brand voice, skip local calendar risks and publish too easily.

Agencies managing several clients also need one place to review drafts — including read-only links for clients — without sharing logins or losing the audit trail.

Why off-the-shelf was not enough

Scheduler tools post what you give them. Chatbots draft without brand research, bilingual native writing or market-specific risk checks. Page7 combines research, drafting, risk and mandatory human approval in one pipeline.

What the platform does

Page7 reads a website and existing posts to draft a brand profile the customer confirms. It then plans and writes captions natively in each language, generates visuals, and runs timing, sensitivity and ad-disclosure checks.

Nothing goes live until a person approves. Agencies can share read-only review links. A plain-language weekly report follows — and the product is honest that it is still pre-launch.

How it works

From brand research to human approval

  1. 01 / Week plan

    From a short brief to seven days of posts

    Describe the business or point at a website. Page7 plans the week, writes captions and chooses visuals, then hands the result to a human for review.

    BuiltOur work
    Page7
    Week plan demonstration with sample coffee-shop posts
  2. 02 / Brand Brain

    Learn the brand before writing

    A brand profile — tone, audience, offers, cadence — is drafted from the real site and posts, then confirmed before any caption is written.

    BuiltOur work
    Page7
    Brand Brain profile card for a sample bakery
  3. 03 / Bilingual writer

    Native Arabic and English, not translated

    Captions are written directly in each language the business posts in, with alt text generated for every image.

    BuiltOur work
    Page7
    Side-by-side English and Arabic caption drafts
  4. 04 / Risk engine

    Checks before the queue

    Drafts are screened for calendar conflicts, sensitive topics and missing ad disclosures before they reach a person.

    BuiltOur work
    Page7
    Risk engine sample checks for timing, disclosure and sensitivity
  5. 05 / Approval

    Nothing publishes without sign-off

    Every draft waits in one queue. Agencies can send clients a read-only review link without handing out a login.

    BuiltOur work
    Page7
    Approval queue with bilingual draft awaiting human sign-off

Screens are public product demonstrations from page7.io with sample brands and captions. Page7 is pre-launch and does not claim customers yet.

Capabilities

What exists, and what does not.

Every capability carries a status. Nothing is listed as live unless it is running in production.

Live
Running in the client's production environment.
Built
Implemented in the codebase, but switched off for production users or waiting on configuration.
Could build
Not built. Something we could build for you, shown so you can see where the work could go.
  • Brand research and Brand Brain profile

    Website and post analysis before drafting.

    Built
  • Native bilingual caption drafting

    Arabic and English written directly, not machine-translated.

    Built
  • Cultural calendar and risk checks

    Timing, sensitivity and ad-disclosure screening.

    Built
  • Human approval queue

    Mandatory sign-off; edit, approve or reject.

    Built
  • Agency review links

    Read-only client review without sharing logins.

    Built
  • Instagram and Facebook publishing

    Via social integrations; other networks on the roadmap.

    Built
  • Studio creative pipeline

    Templates, image generation and QC gates.

    Built
  • Waitlist and marketing site

    Public bilingual site with honest pre-launch positioning.

    Live
  • Additional networks (TikTok, LinkedIn, etc.)

    On the product roadmap; no dates promised.

    Could build

Our work Everything not marked as “Could build” was engineered by us.

Operational value

What it changes day to day.

Described from how the system works. We do not quote savings or percentages we cannot evidence.

Brand first, drafts second
The system learns tone and offers before it writes, so posts start from the business rather than a generic prompt.
Risk before publish
Calendar, sensitivity and disclosure checks reduce avoidable mistakes before a human ever sees the queue.
Human control is the product
Approval is a fixed rule, not a setting — the right posture for businesses that cannot risk an unsupervised post.
Honest about stage
Pre-launch positioning means no invented logos or results. What you see is the product demonstration, labelled as such.

Built with

  • Next.js and React
  • TypeScript and Tailwind CSS
  • next-intl
  • FastAPI on Python
  • PostgreSQL
  • OpenAI and Anthropic models
  • Social publishing integrations
  • Google Cloud Run

Next

Glasshouse Realty

Ohio real estate brokerage

Read the Glasshouse Realty case study

Need social content without unsupervised AI posting?

Tell us how you plan, draft and approve posts today. We will tell you where an approval-gated pipeline would help.