Meridian

The operating system for creator-driven growth.
A decision engine that turns a product, a budget, and a revenue goal into the exact creators who will hit it.

Live demo → meridian-eight-cyan.vercel.app

---

Meridian is not an influencer marketplace. It is the intelligence layer that decides which creators generate revenue, forecasts the return before a rupee is spent, and measures what actually happened afterward. Brands stop asking "which influencer should I hire?" and start asking "I need this much revenue next quarter, how?"

Open the live demo

Brand → Meridian data core → creator → attributed revenue. Every connection backed by data.

---

Highlights

• AI command console. Describe a product, budget, and goal in one sentence. An LLM tool-use loop runs the engines and returns a ranked lineup, budget allocation, and a revenue forecast.
• Five intelligence engines. Recommendation, prediction, budget planning, attribution, and trend scoring, all reading from one creator graph.
• Real creator profiles. Creators are real public figures linked by their genuine Instagram and YouTube handles, with live profile avatars.
• The data flywheel. Every completed campaign feeds realized ROAS back into the rankings, so each forecast is sharper than the last.
• A landing experience to match. Motion-rich marketing site with an animated brand-to-revenue flow and an interactive WebGL creator graph.

---

Demo and screenshots

The console: a brief becomes a plan

Creator intelligence and real social profiles

Campaign attribution and trend radar

Landing: connection flow and the WebGL creator graph

---

The intelligence layer

The moat is not the UI. It is the engines underneath it.

Engine	What it does	Output
Recommendation	Scores creators on category fit, audience fit, realized performance, fraud-adjusted quality, and value.	Ranked creators with match and confidence scores
Prediction	Forecasts the funnel: reach, clicks, conversions, revenue, ROAS.	Outcome forecast with confidence bands
Planner	Allocates a budget across creators against a revenue goal, with diversification limits.	Lineup, allocation, goal coverage, risks
Attribution	Connects creator activity to realized outcomes, predicted vs actual.	Per-creator track records, forecast accuracy
Trends	Ranks where consumer attention is accelerating and who is positioned to ride it.	Opportunity-scored trends

---

Architecture

flowchart TD
  A["Brand brief<br/>product, budget, goal"] --> CONSOLE["AI Command Console<br/>(LLM tool-use loop)"]
  CONSOLE -->|tool calls| ENGINES{{Intelligence Engines}}

  ENGINES --> REC[Recommendation]
  ENGINES --> PRED[Prediction]
  ENGINES --> PLAN[Planner]
  ENGINES --> ATTR[Attribution]
  ENGINES --> TREND[Trends]

  REC --> GRAPH[("Creator Graph<br/>Drizzle ORM + PGlite")]
  PRED --> GRAPH
  PLAN --> GRAPH
  ATTR --> GRAPH
  TREND --> GRAPH

  ENGINES --> OUT["Ranked creators · budget allocation<br/>revenue forecast · risks"]
  OUT --> CONSOLE
  ATTR -. "realized ROAS retrains rankings" .-> REC

Loading

The data flywheel

flowchart LR
  CD[Campaign data] --> CP[Creator performance] --> SP[Sharper predictions] --> BR[Better recommendations] --> BO[Better outcomes] --> MC[More customers] --> MD[More data] --> CD

Loading

Each turn of the loop widens the gap a competitor starting today would have to close.

---

Tech stack

Layer	Choice
Framework	Next.js 16 (App Router, RSC), React 19
Language	TypeScript
Styling	Tailwind CSS v4, custom design tokens
Data	Drizzle ORM over PGlite (embedded Postgres, zero setup)
AI console	Groq (OpenAI-compatible) tool-use loop, free tier
Motion	Motion (motion/react) and GSAP
3D	Three.js (WebGL creator graph)

---

Project structure

src/
├─ app/
│  ├─ (marketing)/        Landing, About, What we do
│  ├─ (app)/              Console, Creators, Campaigns, Trends
│  └─ api/chat/           AI tool-use endpoint
├─ components/
│  ├─ marketing/          Hero, connection flow, 3D graph, pipeline, bento
│  └─ ...                 Console, artifacts, social previews, primitives
└─ lib/
   ├─ db/                 Drizzle schema, PGlite client, seed
   ├─ engines/            recommendation, prediction, planner, attribution, trends
   └─ ai/                 engines exposed as LLM tools

---

How a brief becomes a plan

1. The brand states a goal. Product, budget, geography, and the revenue it needs.
2. The data scores the graph. Audience quality, purchase intent, and realized outcomes per creator.
3. The engines match and forecast. The creators who maximize expected outcome, ranked with confidence.
4. Revenue is attributed back. Realized results feed the flywheel and sharpen the next forecast.

---

_{Built as a decision engine for creator-driven growth. Not a marketplace.}