# Nivetix Technologies

> Nivetix Technologies is a premier AI automation agency and full-stack software development company. We build custom SaaS platforms, autonomous AI agents, conversational chatbots, and high-performance, conversion-focused websites that scale businesses.

## Core Capabilities

*   **SaaS Development:** Custom Next.js architectures, multi-tenant database designs, isolated row-level security (RLS), and stripe/razorpay billing integrations.
*   **AI Automation & Agents:** Decoupled AI microservice pipelines (Python/Flask), autonomous workflow automation (Make, Zapier, n8n), and custom LLM integrations.
*   **Conversational Voice AI:** Real-time speech recognition (STT) and low-latency acoustic emotion classification engines.
*   **High-Conversion Web Development:** Ultra-fast, SEO-optimized web applications with dynamic edge rendering and pixel-perfect design systems.

---

## Site Navigation & Documentation Links

*   [Homepage](https://nivetix.software/) - Overview of agency capabilities, cost estimator, process timeline, and testimonials.
*   [About Us](https://nivetix.software/about) - Agency philosophy, technical founders (Vineet Verma & Priyanshu Shukla), and core values.
*   [Services Index](https://nivetix.software/services) - Comprehensive detail of development, design, and automation solutions.
    *   [SaaS Platform Engineering](https://nivetix.software/services/saas-development) - Scalable B2B platforms, API engines, and subscription models.
    *   [AI Automation Services](https://nivetix.software/services/ai-automation) - Process automation, cognitive workflows, and script execution.
    *   [AI Chatbot Development](https://nivetix.software/services/ai-chatbot-development) - Custom-trained intelligent client support assistants.
*   [Pricing & Plans](https://nivetix.software/pricing) - Structured retainer tiers, comparison charts, and project estimations.
*   [Project Portfolio](https://nivetix.software/portfolio) - Production case studies, client reviews, and custom-built applications.
*   [Technical Blog](https://nivetix.software/blog) - Dynamic index of engineering guides, tutorials, and industry insights.
*   [AI Innovation Hub](https://nivetix.software/ai-tools) - Interactive web tools including the AI Project Advisor, SEO Web Grader, and Copywriter Sandbox.
*   [Contact & Booking](https://nivetix.software/contact) - Consultation booking scheduler and project brief submission interface.

---

## Selected Technical Documentation

### 1. Multi-Tenant SaaS Row-Level Security (RLS) Schema
Below is the PostgreSQL schema pattern Nivetix utilizes to guarantee absolute database isolation for multi-tenant SaaS structures:

```sql
-- Join table for Multi-Tenant Membership
create table public.organization_members (
  id uuid default gen_random_uuid() primary key,
  organization_id uuid references public.organizations(id) on delete cascade not null,
  user_id uuid references public.profiles(id) on delete cascade not null,
  role text check (role in ('owner', 'admin', 'member')) default 'member' not null,
  created_at timestamp with time zone default timezone('utc'::text, now()) not null,
  unique (organization_id, user_id)
);

-- RLS check helper avoiding nested query recursion
create or replace function auth.jwt_belongs_to_org(org_id uuid)
returns boolean as $$
  select exists (
    select 1 
    from public.organization_members 
    where organization_id = org_id 
    and user_id = auth.uid()
  );
$$ language sql security definer;

-- Apply policy to tenant projects
create policy "Users can view projects belonging to their organization"
  on public.projects
  for select
  using (auth.jwt_belongs_to_org(organization_id));
```

### 2. Hybrid Flask AI Backend & Next.js Proxy Integration
To keep heavy machine learning compute workloads out of the JavaScript main thread, we separate frontends from Python inference backends:

*   **Next.js Server Proxy Handler:** Captures client requests securely, processes tokens, and routes prompts over private networks to avoid exposing underlying model endpoints:
    ```typescript
    // app/api/ai/analyze/route.ts
    const FLASK_BACKEND_URL = process.env.FLASK_BACKEND_URL || 'http://localhost:5000'
    const response = await fetch(`${FLASK_BACKEND_URL}/api/analyze`, {
      method: 'POST',
      headers: { 'Content-Type': 'application/json' },
      body: JSON.stringify({ text: body.text }),
    })
    ```
*   **Flask Model Inference Service:** Loads ML weights into memory on startup (singleton pattern) and handles raw input pipelines:
    ```python
    # app.py (Flask AI Server)
    ai_classifier = pipeline("sentiment-analysis", model="distilbert-base-uncased-finetuned-sst-2-english")
    @app.route('/api/analyze', methods=['POST'])
    def analyze_text():
        data = request.get_json()
        model_output = ai_classifier(data['text'])
        return jsonify(model_output[0]), 200
    ```

### 3. Voice AI & Real-Time Emotion Classification
Our dual-pipeline voice assistants stream audio in binary chunks (PCM 16-bit, 16kHz) to orchestrators over WebSockets, splitting processing into parallel paths:
*   **Linguistic:** Handles live automatic speech recognition (ASR) to obtain transcripts.
*   **Acoustic:** Bypasses semantic text, feeding Mel-Frequency Cepstral Coefficients (MFCCs) to specialized SER networks to deduce client sentiment (Calm, Happy, Frustrated, Neutral) in real time.