- Home>
- Ai Marketing Automation Platform Growstack
GrowStack AI: Intelligent Growth and Automation Platform
Bitontree built GrowStack AI as a modular, microservice-based SaaS platform that unifies AI content creation, lead generation, workflow automation, campaign orchestration, and integrations in one product. We delivered a multi-model AI agent framework, a drag-and-drop workflow builder, an AI lead generator, and custom MCP infrastructure on an event-driven backend. The result is a single platform where growth teams run marketing, sales, and communication workflows end to end.
Team:
AI engineers + backend engineers + frontend engineers + Project Manager
Location:
USA
Industry:SaaS
Project Type:AI SaaS Platform
Duration:15 months
Services Used:AI Agent Development, AI Automation Development, Custom Software Development
Team:AI engineers + backend engineers + frontend engineers + Project Manager
About the Client
GrowStack AI is an enterprise-grade growth automation platform built for marketing teams, sales professionals, and business operators. The client set out to build a scalable SaaS product that unifies AI content creation, lead generation, workflow automation, campaign orchestration, subscription management, and third-party integrations in a single interface.
The objective was a powerful automation platform that lets businesses run marketing, sales, lead generation, content, and communication workflows from one place, with multi-channel automation, AI content across multiple models, lead prospecting and scoring, multi-tenant subscriptions, and extensible AI tooling.
The Challenge
Growth teams usually stitch together separate tools for lead research, outreach, social posting, CRM updates, content generation, automation, billing, and analytics. GrowStack AI needed to bring all of it into one platform.
Many AI providers, one framework
The platform had to support multiple AI model providers, including OpenAI, Anthropic Claude, Google Gemini, AWS Bedrock, and Perplexity, behind a flexible agent architecture capable of running more than 130 specialized agents.
Visual automation for non-technical users
Teams needed a visual workflow builder with 50+ action nodes, scheduled and event-based triggers, conditional branching, and approval gates, so non-technical operators could build automations without writing code.
Large-scale lead data processing
The system had to process large raw lead datasets and pinpoint specific prospects from them, with enrichment, scoring, and analytics on top.
High-throughput orchestration
Workflow orchestration had to handle high-throughput automation through event-driven services, workers, schedulers, and queues, with real-time state management.
Deep third-party integration
GrowStack AI needed to connect with the tools growth teams already use, spanning CRMs, calendars, social and ad platforms, communication tools, storage systems, and productivity apps, with 30+ integrations in scope.
Subscriptions, usage, and observability
The product required multi-tenant subscription plans, credit-based usage metering, and end-to-end observability across frontend and backend services.
What Was at Stake
Running growth on a stack of disconnected tools came with a real cost. Lead lists lived in one product, outreach in another, social posting somewhere else, and the CRM only learned about any of it after someone exported a file and pasted it across. Operators lost hours every week to copy-paste handoffs and reconciling numbers that never quite matched. Each new model provider or channel meant another subscription, another login, and another integration to babysit. Worst of all, no single view showed what was actually working, so decisions were made on partial data while the most repetitive work stayed manual. GrowStack AI set out to collapse that sprawl into one platform where the data, the automation, and the AI all live together.
Our Approach and What We Built
Bitontree operated as an embedded AI engineering team, building dedicated services for each major capability and connecting them through an event-driven architecture. The frontend runs on Next.js 16, React 19, Tailwind CSS, Radix UI, Shadcn, and React Flow / XY Flow. The backend is a NestJS NX monorepo with orchestrator, worker, scheduler, and workflow cache services, alongside an Express user microservice, an Express AI agents service on the Vercel AI SDK and LlamaIndex, and Python FastAPI services for lead generation (LangChain, LangGraph, ClickHouse) and custom MCP tooling. MongoDB, Redis, and ClickHouse handle storage, Kafka, RabbitMQ, and Socket.IO handle messaging and real-time state, and the platform runs on AWS (S3, ECR, ECS, EC2, Route 53, CloudWatch, SES, Secrets Manager) with Docker and Kubernetes, instrumented with OpenTelemetry and Sentry.
How We Built It
Multi-Model AI Agent Framework
We built an AI agent service supporting OpenAI, Anthropic Claude, Google Gemini, AWS Bedrock, and Perplexity through the Vercel AI SDK and LlamaIndex. The architecture lets GrowStack AI create small and large agents for specific user tasks, and we delivered more than 130 agents across content, automation, lead generation, and tool-based workflows.
Visual Workflow Builder
We developed a drag-and-drop workflow automation builder using React Flow / XY Flow. It supports 50+ pre-built action nodes, scheduled triggers, event-based execution, conditional branching, approval gates, and Redis-backed workflow state management.
AI Lead Generator
We built an AI-powered lead research and enrichment system with Python, FastAPI, LangChain, LangGraph, and ClickHouse. It handles prospect research, email enrichment, scoring, CSV and Excel import and export, and analytics dashboards, searching and filtering a large Apollo dataset to identify specific leads from user requirements.
Custom MCP Infrastructure
We developed custom Model Context Protocol servers and clients to extend the platform's AI tooling, letting agents connect with external tools and services in a flexible, tool-augmented architecture.
Event-Driven Backend Architecture
We built the backend as a NestJS NX monorepo with dedicated orchestrator, worker, scheduler, and workflow cache services. Kafka, RabbitMQ, Redis, and Socket.IO support workflow execution, queueing, real-time state, and high-throughput automation.
Multi-Channel Campaign Orchestration
We integrated campaign execution across email, social, messaging, and ad channels, including Gmail, Microsoft, SendGrid, LinkedIn, Facebook, Twitter/X, Instagram, SMS, WhatsApp, Google Ads, and Facebook Marketing.
Enterprise Integrations
We delivered 30+ integrations across business tools, including CRMs (Salesforce, HubSpot, Zoho, GoHighLevel), calendars (Google Calendar, Outlook, Cal.com, Calendly), storage and data (Google Drive, Google Sheets, Airtable, Snowflake), communication (Slack, Telegram, Zoom, Fireflies), marketing (Instantly, Apollo, WordPress, YouTube), and productivity apps.
Subscription, Billing, and Usage Metering
We implemented Stripe-powered subscription and plan management with credit-based usage tracking, enabling GrowStack AI to manage multi-tenant subscriptions, usage limits, and billing workflows inside the platform.
Admin, Analytics, and Observability
We built admin dashboards for user management, credit usage tracking, plan analytics, report generation, tag-based filtering, world map analytics, and chart-based reporting. We also implemented OpenTelemetry instrumentation and Sentry error tracking across the frontend and backend.
The Results
Bitontree helped GrowStack AI ship a large-scale growth automation SaaS platform with multi-model AI, visual workflow automation, lead generation, integrations, billing, analytics, and custom MCP infrastructure.
130+
AI agents delivered across content, workflow, research, and automation use cases
50+
Action nodes in the drag-and-drop visual workflow builder
30+
Third-party platform integrations across CRM, calendars, social, and more
5
AI model providers supported behind a single agent framework
Stitched-Together Growth Stack vs GrowStack AI
| Dimension | Stitched-Together Growth Stack | GrowStack AI |
|---|---|---|
| AI model choice | Locked to whatever single model your point tool ships with | OpenAI, Anthropic Claude, Google Gemini, AWS Bedrock, and Perplexity behind one agent framework |
| Automation building | Built with code or glued together through Zapier handoffs | A drag-and-drop builder with 50+ action nodes, triggers, branching, and approval gates |
| Lead research | A separate prospecting tool plus manual CSV exports and imports | Built-in AI lead research with enrichment and scoring over a large Apollo dataset |
| Tool extensibility | Fixed connectors, so AI tooling stops where the vendor decided | Custom MCP servers and clients that let agents reach new tools and services |
| Real-time orchestration | Batch syncs between tools, with state scattered across products | Event-driven services with Kafka, RabbitMQ, Redis, and Socket.IO for real-time state |
| Usage metering | Each tool bills and tracks usage on its own terms | Credit-based usage metering across the whole platform in one place |
| Observability | Logs trapped inside each separate tool, no shared view | OpenTelemetry and Sentry instrumentation across frontend and backend |
Technologies Used
GrowStack AI was engineered as an event-driven, microservice-based SaaS platform, with the stack chosen for multi-model AI execution, high-throughput automation, and integration reliability at scale.
Looking For More Case Studies?
Don’t just take our word for it - our track record reflects our expertise and success.
LogisticsSingapore: 
Smart AI Invoice Processing System
AI-powered invoice processing for a Singapore-based logistics enterprise. OCR and ML automate data extraction, validate against business rules, and process invoices end-to-end across multiple formats and currencies.
PythonLangGraphCrewaiStreamlitAzure
Frequently Asked Questions
What is GrowStack AI?
GrowStack AI is an enterprise-grade growth automation platform that unifies AI content creation, lead generation, workflow automation, campaign orchestration, subscription management, and third-party integrations in a single product for marketing and sales teams.
Which AI model providers does the platform support?
The multi-model AI agent framework supports OpenAI, Anthropic Claude, Google Gemini, AWS Bedrock, and Perplexity through the Vercel AI SDK and LlamaIndex, so agents can run on the most suitable model for each task.
How does the workflow automation work?
GrowStack AI includes a drag-and-drop workflow builder with 50+ pre-built action nodes, scheduled and event-based triggers, conditional branching, and approval gates, backed by Redis state management so non-technical users can build automations visually.
How many integrations and AI agents were delivered?
Bitontree delivered more than 130 specialized AI agents and 30+ third-party integrations spanning CRMs, calendars, social and ad platforms, communication tools, storage systems, and productivity apps.
What is the MCP infrastructure used for?
We built custom Model Context Protocol servers and clients so GrowStack AI can connect its AI agents with external tools and services in a flexible, tool-augmented architecture, making the platform's AI capabilities extensible over time.
What is a multi-model AI agent framework?
It is the architecture that lets one platform run AI agents on several model providers instead of being tied to a single vendor. In GrowStack AI, the framework supports OpenAI, Anthropic Claude, Google Gemini, AWS Bedrock, and Perplexity through the Vercel AI SDK and LlamaIndex, so each of the 130+ agents can run on the model best suited to its task across content, automation, lead generation, and tool-based workflows.
How does credit-based usage metering work?
Usage across the platform is tracked as credits rather than counted per tool. We built Stripe-powered subscription and plan management with credit-based usage tracking, so GrowStack AI can run multi-tenant subscriptions, apply usage limits per plan, and handle billing workflows inside the product. Admins see credit usage and plan analytics through the built-in dashboards.
What integrations does GrowStack AI support?
The platform ships with 30+ integrations across the tools growth teams already use: CRMs such as Salesforce, HubSpot, Zoho, and GoHighLevel, calendars including Google Calendar, Outlook, Cal.com, and Calendly, storage and data tools like Google Drive, Google Sheets, Airtable, and Snowflake, communication apps such as Slack, Telegram, Zoom, and Fireflies, and marketing platforms including Instantly, Apollo, WordPress, and YouTube. Campaign execution also reaches email, social, messaging, and ad channels.