Operator's Guide

About this guide

This is the working manual for Turquoise Tree Studio — the AI‑assisted, human‑led marketing console your agency runs its clients on. It's written for the person at the controls.

Read it front‑to‑back once, then keep it open as a reference. Every screen in the console maps to a section here.

• New to the console? Read Part 1 and Part 2, then run the Quick Start.
• Looking for one screen? Jump to Part 3 — it follows the sidebar top‑to‑bottom.
• Doing a job end‑to‑end? Use the Playbooks in Part 4.
• Running the platform itself? See Part 5.

Part 1 · The Operating Model

Five ideas explain why the console behaves the way it does. Understand these and the rest is intuitive.

1 · Strategy is the single source of truth

Every brand begins with a Strategy Brief. Industry, audience, voice and market all live there, and every downstream tool inherits from it. You never re‑type a brand's positioning — you set it once and the system carries it.

2 · Brand Memory grounds every generation

When a brief exists, the console silently prepends a Brand Context block to every AI call for that client. This is why two clients asking for "a launch post" get completely different, on‑brand output. A 🧠 Brand Memory on badge shows when it's active. It's enforced server‑side — a stray prompt can't bypass it.

3 · Research first, then write ("grounding")

For anything that should reflect the real world, the console runs a Perplexity → Claude sequence: Perplexity Sonar researches current, factual market data; Claude then writes grounded in those facts, with citations. This is what stops the AI inventing figures.

4 · Industry‑agnostic by design

The engine never hard‑codes a region, sector or house style. A fintech, a coffee roaster and a Dubai developer each get output shaped by their brand context. When you see market‑specific language for a specific brand, that's the system deferring to the client's real brief — which is correct, not a leak.

5 · Nothing publishes without a human

Every generated asset lands in Approvals. Autonomous mode, batch generation, campaigns — all stop at the approval gate. Publishing is a deliberate, separate act. The AI does the heavy lifting; you sign off.

TT Studio as a Marketing Operating System

TT Studio is not an AI content generator with extra features. It's an AI Marketing Operating System: one place where a brand's marketing is researched, planned, produced, published, and learned from — on a persistent memory of that brand. A point tool answers "write me a post." An OS answers "run this brand's marketing."

That reframes what the console is for. Every screen is a stage in one continuous loop:

      Research ─▶ Business understanding ─▶ Competitive intel
         ▲                                          │
         │                                          ▼
  Strategy update                              Brand Memory
         ▲                                          │
      Learning                                   Strategy
         ▲                                          │
         │                                          ▼
  Analytics & attribution                  Campaign planning
         ▲                                          │
         │                                          ▼
     Publishing ◀─ Video ◀─ Image ◀─ Content creation

The console is built in layers, each answering a different question:

The five principles above are the OS's operating rules; the layers are how it's organised; the loop is what it does.

The loop today vs. the vision

Be clear with clients about where the loop is closed and where a human still drives:

• Closed today (research → publish). The console takes a brand from blank slate → signed strategy → a month of grounded, on‑brand, human‑approved content and video, scheduled to channels. This front two‑thirds of the loop runs now.
• The edge still being closed (learn → adapt). Real performance data flowing back into strategy — analytics, attribution, autonomous optimization — is the roadmap. Today that feedback is human‑led; the platform is being wired to close it automatically as native platform data comes online.
• Deliberately human, not missing. The approval gate and strategic sign‑off are design choices, not gaps. TT Studio is a human‑led marketing OS — the AI runs the loop, a strategist owns the outcome.

Part 2 · Orientation

Roles & access

Admin‑only areas are marked admin throughout. Access is enforced on the server, not just hidden in the UI.

The navigation map

The sidebar is organised into six buckets. Click a bucket to expand it.

🏠 Home            Command centre for the active client
🏢 Agency [admin]  Cross-client portfolio — all brands at a glance
✨ Studio          ✦ Content · 🎬 Video · 🎨 Design & Assets
📅 Calendar        📆 Schedule · 📣 Campaigns · ✓ Approvals
📈 Performance     📊 Analytics · 🗂 Agency Dashboard [admin] · 🧮 Ad-Spend Calc
👥 Clients         🎯 Strategy · 💎 Brand IQ · 📄 Proposal · ⭐ Influencer
⚙️ Settings        🔗 Channels · 🔑 API Console [admin]

The active‑client pill (top‑left, under the logo) shows whose workspace you're in — click it to switch. Everything you generate applies to the active client. The live balance strip (bottom‑left) shows real provider spend, so you always know your runway.

Quick Start — your first 15 minutes

1. Create a client — top‑right + New Client, or open the onboarding wizard. Give it a name and industry.
2. Build the Strategy Brief — Clients → 🎯 Strategy. Answer the guided questions (or let AI research and draft them) and save. This unlocks Brand Memory.
3. Generate a post — Studio → ✦ Content. Enter what to post; the engine writes copy + a matching image, grounded in the brief.
4. Review it — the asset appears in Calendar → ✓ Approvals. Open it, tweak the image if needed, Approve.
5. Schedule it — connect a channel in Settings → 🔗 Channels, then push the approved post to the queue.

That's the whole loop in miniature.

Part 3 · Feature Reference

Follows the sidebar top‑to‑bottom.

🏠 Home

The active client's command centre.

• Headline tiles — Content Score, Content Generated, Engagement, Leads. Populate as real data flows in.
• AI Recommendations — living, brand‑aware suggestions for what to post next.
• Content Pipeline — counts of posts / images / videos in play.
• Plan Usage · This Month — image and video‑second consumption against the client's plan (visibility only — it informs, it doesn't hard‑block).
• Content Plans admin — the four‑tier pricing table (TT Lite / Green / Blue / Aqua), USD with AED in brackets.

🏢 Agency admin

The cross‑client portfolio dashboard — your whole book of business on one screen.

• A card per client showing pending approvals, posts this month, total posts, plan and last activity.
• The header sums the portfolio (N clients · X pending · Y posts this month).
• Click any card to jump straight into that client and land on their Home.

✨ Studio

✦ Content

The core post factory.

1. Brief — say what to post (a topic, an angle, or paste a source URL to ground on). Optionally Set up & Fill to let AI infer the settings, or dictate by voice.
2. Settings — Content Type, Platform, Tone, Length, plus the image model tier. All inherit from the brief but stay editable.
3. Generate — a structured post (hook, caption, hashtags, CTA) and a matching on‑brand photograph, grounded in Brand Memory.
4. Per‑channel captions — expand Customise per channel for a native‑style caption per platform.
5. Save Draft / Preview / Approve — send it toward Approvals.

🎬 Video

The hero‑video studio — video is TT Studio's showpiece.

1. Video Prompt — describe the video, or Set up & Fill to infer aspect / length / style / on‑screen title.
2. Settings — Duration, Aspect ratio, Video style and the hero model. Hero models are curated to Kling, Veo and Seedance; 🔊 marks native‑audio models.
3. AI Script — a scene‑by‑scene brief with a consistency anchor so characters, palette and lighting stay coherent across cuts.
4. Generate Video — multi‑scene videos are stitched and captioned automatically (lower‑third burn‑in).

Captions are burned in server‑side; no text is baked into the raw plate — the overlay layer owns text.

🎨 Design & Assets

The client's media library, plus the light image editor.

• Browse every generated asset for the client.
• Open any image to edit: overlay text, reposition (top / middle / lower‑third), recolour to brand, resize (S/M/L/XL), or Swap plate to regenerate the underlying image from the post's brief.
• Saves render the caption server‑side and host the result durably. Also reachable from an image's card in Approvals.

📅 Calendar

📆 Schedule

A read‑only month view of what's scheduled, plus the 🚀 Autonomous Campaign card — the flagship one‑click workflow.

1. Enter an optional Campaign Goal (it shapes the narrative arc) and a few themes; pick a cadence.
2. ✨ Plan only — builds a phased 3–4 stage narrative arc and a full month of posts, each mapped to a phase. Free (no asset spend). Review before committing.
3. 🚀 Launch autonomous campaign — builds the arc, plans the month and auto‑generates every asset into Approvals. Nothing publishes.
4. 🎨 Generate all images (background) — durable batch generation; keeps running even if you close the tab.

The plan is grounded on the whole goal + theme set (not just the first line) and every post carries its phase chip.

📣 Campaigns

Groups related content under a named campaign so you can track and report on it as a unit. Assets generated from a campaign carry its ID.

✓ Approvals

The sign‑off queue — the gate everything passes through.

• Toggle Pending / Approved. A live count badge in the sidebar shows how many items await you.
• Open any item to review copy + media. For images, ✎ Edit image opens the light editor.
• Approve or Reject. Approved items become eligible for scheduling.

📈 Performance

• 📊 Analytics — engagement, reach and performance over a selectable period (week / month / quarter / year), with trend and channel breakdowns.
• 🗂 Agency Dashboard admin — operator‑level oversight across the book.
• 🧮 Ad‑Spend Calc — model ad budgets and expected returns for planning and proposals.

👥 Clients

🎯 Strategy

The onboarding wizard and Strategy Brief — the front door for every brand.

• A guided flow: brand questions → optional website build → a finished Strategy Brief the client approves and signs.
• Free‑text audience and theme fields with Generate Input with AI (web‑grounded research).
• Paste a source URL to auto‑draft a brief from a brand's existing site.

💎 Brand IQ

The editable brand profile — voice, palette, do's and don'ts — that enriches Brand Memory beyond the brief.

📄 Proposal admin

Generate a client‑ready proposal from a saved template, scoped to the active client's industry and brief. Hidden from client logins.

⭐ Influencer

Shortlist and manage influencer / creator partners for the client.

⚙️ Settings

🔗 Channels

Connect the client's social channels (via Buffer) so approved content can be scheduled. Channel selection at publish time is honoured per post.

🔑 API Console admin

Provider health and model usage — see which engines ran, spend and status. Your window into the AI supply chain.

Part 4 · Playbooks

End‑to‑end workflows. Each is a repeatable job.

Part 5 · Administration

Providers & credits

TT Studio orchestrates several AI providers: media generation on Fal (draft + hero image and video); text and reasoning on Claude; live research on Perplexity; durable hosting on Vercel Blob; scheduling via Buffer. The live balance strip and 🔑 API Console show real spend and health. Keep Fal and Perplexity funded — grounding fails open (degrades gracefully) if research credit runs low, but media generation needs Fal balance.

Metering

Plan usage is visibility‑only — the console shows consumption against a client's plan to inform decisions, but does not hard‑block generation.

Roles & security

Client accounts are fully authenticated and scoped to their own workspace, server‑enforced. Admin sees everything. Admin‑only tools are never reachable by a client login.

Deployment

The console runs as a single app on Vercel, backed by a shared cloud database. Production is aliased to the live URL; preview deploys are used to review changes before promoting. Media rendering runs server‑side on the deploy.

Part 6 · FAQ & Troubleshooting

Frequently asked questions

Troubleshooting

A generation came back generic / off‑brand.
Check the client has a saved Strategy Brief and the 🧠 Brand Memory on badge is showing. No brief → no brand grounding.

Numbers or claims look invented.
Grounding may have been off or research credit low. Re‑run with grounding, or paste a source URL.

An image tile is blank / broken.
The provider's temporary URL may have expired. Open the asset and Swap plate in the editor to regenerate, then Save (which hosts it durably).

The campaign plan looks cut off.
Very large plans are token‑budgeted automatically; if a plan ever looks truncated, reduce the cadence and re‑plan.

A client can't see a feature I can.
By design — Agency, Proposal, Agency Dashboard and API Console are admin‑only.

Nothing published after I approved.
Approval and publishing are separate. Connect the channel in 🔗 Channels and push the approved item to the queue. The platform never auto‑publishes.

I want to work across all clients at once.
Use 🏢 Agency for the portfolio view. Bulk actions across clients are a planned extension.

Appendix · Glossary