Sentra AI

Sentra AI is a production-style enterprise AI intelligence platform built with Next.js 15, TypeScript, Tailwind CSS, ShadCN-style UI primitives, Framer Motion, GSAP, Three.js / React Three Fiber, AI/ML API, Bright Data, Speechmatics, Lucide icons, and Recharts.

The app presents a futuristic operating system for autonomous business intelligence: landing page, authentication screens, AI dashboard, live alert center, enterprise chat, voice playback, charts, command palette, and typed API routes.

Features

• Premium dark glassmorphism UI with animated gradients, particles, glowing AI orb, smooth reveal animations, and responsive layouts.
• Dashboard for competitor monitoring, market intelligence, live signals, sentiment/risk analysis, company tracking, trend visualization, and AI recommendations.
• AI chat interface with markdown responses, typing animation, prompt shortcuts, thinking state, contextual follow-up questions, retrieval labels, and voice playback controls.
• AI World Engine with a WebGL intelligence globe, adaptive visualizations, relationship graph, predictive timelines, scenario simulations, narrated briefings, and cinematic World Pulse presentation mode.
• Real-time AI Activity Console backed by server-sent events, with measured collection/API lifecycle logs, virtualized terminal output, live source tracking, system health telemetry, and safe summarized thought-stream cards.
• Visual-forensics analyst workspace for prompt-directed image investigations, comparison review, verdict scoring, report history, and PDF export.
• Backend API routes for Bright Data intelligence collection (SERP + Web Unlocker), AI/ML API analysis, and Speechmatics voice synthesis.
• Demo-safe fallbacks in local dev when provider keys are absent; production requires live Bright Data for GTM collection paths.

Deploy on Vercel

1. Import the repo at vercel.com/new (Next.js auto-detected).
2. Set env vars from .env.example (Vercel block) — run npm run env:check:vercel for the list.
3. Add Supabase redirect URL: https://<your-app>.vercel.app/auth/callback
4. Run npm run secrets:sync locally so AIML / Bright Data keys live in the Supabase vault.
5. Deploy — vercel.json uses a daily cron (Hobby-safe). Full guide: docs/DEPLOY_VERCEL.md.

Getting Started

npm install
npm run dev

Open http://localhost:3001 (dev server default port).

Collaborators (missing .env.local)

.env.local is gitignored — it is not pushed to GitHub. New developers:

npm run env:setup    # creates .env.local from .env.example
# add Supabase keys (see docs/COLLABORATOR_SETUP.md)
npm run env:check
npm run dev

Get keys from the project owner (secure message) or a Supabase team invite. Full guide: docs/COLLABORATOR_SETUP.md and docs/SUPABASE_SETUP.md.

Local development (GitHub collab, no deploy yet)

You can run Sentra without Supabase. Leave NEXT_PUBLIC_SUPABASE_URL and NEXT_PUBLIC_SUPABASE_ANON_KEY empty in .env.local.

1. npm run env:setup then sync provider keys to Supabase vault (npm run secrets:sync) or add AIML_API_KEY, Bright Data, and SPEECHMATICS_API_KEY locally.
2. npm run dev → open http://localhost:3001 → Enter workspace (local mode).
3. Monitors are stored in browser localStorage per machine; chat history is session-only until you add Supabase for deploy.

When you deploy, create a Supabase project, run supabase/migrations/001_initial_schema.sql, add the three Supabase env vars, and real auth + cloud persistence turn on automatically.

Environment Variables

Copy .env.example to .env.local and fill in the provider keys you want to enable:

# Optional until production deploy
NEXT_PUBLIC_SUPABASE_URL=
NEXT_PUBLIC_SUPABASE_ANON_KEY=
SUPABASE_SERVICE_ROLE_KEY=

AIML_API_KEY=
AIML_BASE_URL=https://api.aimlapi.com/v1
AIML_MODEL_ANALYSIS=gpt-4o-mini
AIML_MODEL_CHAT=gpt-4o
AIML_MODEL_SEARCH=gpt-4o-search-preview
AIML_MODEL_INTENT=gpt-4o-mini
AIML_MODEL_WORLD=gpt-4o
AIML_MODEL_VISION=gpt-4o
AIML_MODEL_TRANSCRIBE=#g1_whisper-medium
SENTRA_TIMEZONE=Asia/Colombo

BRIGHT_DATA_API_KEY=
BRIGHT_DATA_CACHE_TTL_SECONDS=900
BRIGHT_DATA_SERP_ENDPOINT=https://api.brightdata.com/request
BRIGHT_DATA_SERP_ZONE=
BRIGHT_DATA_WEB_UNLOCKER_ENDPOINT=https://api.brightdata.com/request
BRIGHT_DATA_WEB_UNLOCKER_ZONE=
BRIGHT_DATA_SCRAPER_ENDPOINT=

SPEECHMATICS_API_KEY=

Without keys, Sentra AI uses polished demo intelligence so the UI and API routes remain functional.

Routes

• / - cinematic landing page
• /sign-in - Supabase auth (email, magic link, Google, GitHub)
• /sign-up - create workspace
• /onboarding - first-run setup checklist
• /settings - integration status (Bright Data zones, AI/ML API, Speechmatics)
• /dashboard - intelligence operating system with live briefing refresh
• /chat - AI analyst chat with microphone transcription and voice playback
• /analyst - AI World Engine and visual-forensics investigation workspaces
• /alerts - enterprise alert center

API Routes

• POST /api/chat - answers general chat questions using AI/ML API live web search; monitoring, competitor, pricing, and URL prompts use Bright Data collection first, then AIML analysis with citations
• POST /api/transcribe - transcribes microphone recordings (AIML Whisper)
• GET|POST /api/intelligence - returns risks, opportunities, recommendations, confidence score, and live signals
• POST /api/world-engine - creates visualization-ready global intelligence models with live-search verification and Bright Data evidence when configured
• POST /api/world-engine/stream - streams real collection, web-search, synthesis, visualization, and completion activity events over SSE while returning the completed World Engine model
• POST /api/image-analysis - analyzes uploaded visual evidence after an investigator submits a prompt
• POST /api/bright-data - reusable Bright Data collection endpoint (auth required)
• POST /api/voice - Speechmatics text-to-speech (WAV)
• GET /api/cron/monitors - scheduled GTM monitor checks (Vercel Cron + CRON_SECRET)
• GET /api/signals - latest signals for the signed-in user
• GET|POST /api/monitors - CRUD custom monitors
• POST /api/monitors/[id]/check - Bright Data + AIML monitor check (manual or cron)
• GET /api/health/integrations - integration readiness for settings/onboarding

Supabase setup

1. Create a project at supabase.com.
2. Run the SQL in supabase/migrations/001_initial_schema.sql in the SQL editor.
3. Enable Email, Magic Link, Google, and GitHub providers in Authentication.
4. Set redirect URL: http://localhost:3001/auth/callback (and your production URL).

Hackathon — GTM Intelligence (Web Data UNLOCKED)

Track: GTM Intelligence
Full submission guide: docs/HACKATHON_GTM_SUBMISSION.md
Operator setup (zones, Vercel, vault): docs/USER_SETUP.md

Stack: Bright Data (SERP + Web Unlocker) → AI/ML API → Supabase → Vercel. Voice via Speechmatics.

Judge demo (5 min)

1. npm run secrets:sync — AIML, Bright Data zones, Speechmatics in Supabase vault.
2. Deploy to Vercel with Supabase env vars + CRON_SECRET (see hackathon doc).
3. Settings → AI/ML API + Bright Data + Speechmatics Ready.
4. Dashboard → Refresh briefing → Live · Bright Data (not Sample).
5. Chat → paste competitor https:// URL → badge Bright Data + AIML.
6. Alerts → monitor → Check now → executive report.
7. Optional: World Engine, document upload (Featherless), voice play.

Bright Data promo: unlocked on brightdata.com.

Production GTM rules

• On Vercel/production, sample Bright Data fallback is disabled unless SENTRA_ALLOW_DEMO_FALLBACK=true.
• Active monitors are checked on a schedule via /api/cron/monitors (daily on Vercel Hobby; see vercel.json and docs/DEPLOY_VERCEL.md).

Production Notes

• Deploys cleanly on Vercel.
• Provider calls run in Node.js API routes and keep secrets server-side.
• Chat uses Bright Data only when configured and a collection-oriented prompt is detected; otherwise it uses AI/ML API search-capable models.
• Demo fallbacks are intentionally typed and isolated in src/data and src/services so they can be swapped for real data pipelines.