AI Innovation Lab

Our AI Projects

Real-world AI implementations spanning healthcare, medical data interoperability, CRM automation, and intelligent clinical assistants — built on cutting-edge cloud platforms.

01
Healthcare AI

Surgical Video Intelligence

An AI pipeline that analyzes live operating room footage to classify whether a surgical procedure was successful. Video is streamed to GCP Video Intelligence, which labels each timeframe with confidence scores — enabling an ensemble model to render a final success/failure classification.

KEY CAPABILITIES
  • Frame-level surgical event labelling via Video Intelligence API
  • Temporal confidence aggregation across procedure phases
  • Binary classification: Successful / Needs Review
  • Secure video storage with GCS lifecycle policies
TECH STACK
GCP Vertex AI Video Intelligence API Cloud Storage (GCS) Cloud Functions BigQuery Cloud IAM
Architecture Flow
flowchart TD A([🎥 OR Video Feed]) --> B[GCS Video Bucket] B --> C{Video Intelligence API} C --> D[Frame-Level Labels\nTimestamp: 0s-30s\nLabel: Incision Initiated] C --> E[Frame-Level Labels\nTimestamp: 31s-120s\nLabel: Organ Exposure] C --> F[Frame-Level Labels\nTimestamp: 121s-END\nLabel: Closure Complete] D & E & F --> G[Temporal Confidence\nAggregator] G --> H{Vertex AI\nClassification Model} H --> I([✅ SUCCESS]) H --> J([⚠️ NEEDS REVIEW]) I & J --> K[(BigQuery Results\nAudit Table)] K --> L[Dashboard /\nAlert System] classDef default fill:#111,stroke:#00a86b,stroke-width:2px,color:#fff; classDef bucket fill:#0a2a0a,stroke:#00a86b,stroke-width:2px,color:#fff; classDef result fill:#0a2a0a,stroke:#4caf50,stroke-width:3px,color:#4caf50; classDef warn fill:#2a1a00,stroke:#FF9900,stroke-width:3px,color:#FF9900; class B bucket; class I result; class J warn;
02
Code Generation AI

FHIR → Whistle Code Generator

A Gemini-powered pipeline that ingests FHIR health data from medical devices (ECG, monitors etc.) and automatically generates Whistle transformation functions. When new metadata arrives without an existing mapping, the system falls back to Gemini to synthesize a new function on-the-fly and integrates it into the pipeline.

KEY CAPABILITIES
  • FHIR R4 resource parsing (Observation, Device, Patient)
  • Whistle code lookup from GCS function registry
  • Gemini fallback: text-bison → code-bison → gemini-pro chain
  • Auto-deployed via Docker + GCP Load Balancer
TECH STACK
Vertex AI Gemini Pro Code Bison Text Bison Docker GCS Bucket Load Balancer FHIR R4
FHIR → Whistle Pipeline
flowchart TD A([🏥 ECG / Medical Device]) --> B[FHIR R4 Payload\nObservation Resource] B --> C{GCS Function\nRegistry Lookup} C -- Whistle fn exists --> D[Load Existing\nWhistle Function] C -- New metadata / no mapping --> E[FALLBACK CHAIN] E --> F[text-bison\nContext Analysis] F --> G[code-bison\nWhistle Synthesis] G -- Low confidence --> H[gemini-pro\nAdvanced Generation] H --> I[Validate & Test\nGenerated Code] I --> J[(Store New fn\nin GCS Registry)] D & J --> K[Execute Whistle\nTransformation] K --> L([✅ Transformed\nHealth Data]) L --> M[Downstream Systems\nEHR / Analytics] classDef default fill:#111,stroke:#4285F4,stroke-width:2px,color:#fff; classDef fallback fill:#0a0a2a,stroke:#4285F4,stroke-width:2px,color:#fff; classDef store fill:#001a2a,stroke:#4285F4,stroke-width:2px,color:#4285F4; class E,F,G,H fallback; class J store;
03
CRM Automation

Zoho CRM MCP Chatbot

A conversational AI system integrating OpenAI GPT-4.1 with a custom-built Model Context Protocol (MCP) layer for Zoho CRM. When a user asks a question, the MCP decides which CRM action to invoke — creating leads, fetching deal pipelines, updating contacts, or generating reports — and returns structured results via Flutter UI.

KEY CAPABILITIES
  • Custom MCP with 20+ Zoho CRM tool definitions
  • Intent classification and action routing via GPT-4.1
  • Real-time CRM data fetch, create & update operations
  • Flutter cross-platform mobile & web UI
  • Deployed on AWS EC2 with Node.js + Python backend
TECH STACK
OpenAI GPT-4.1 Custom MCP Zoho CRM AWS EC2 Node.js Python Flutter
MCP Decision Flow
flowchart TD A([👤 User Query\nFlutter App]) --> B[Node.js API Gateway\nAWS EC2] B --> C{GPT-4.1\nIntent Analyzer} C --> D[Custom MCP\nAction Router] D -- Get Leads --> E[Zoho CRM\nLeads API] D -- Update Deal --> F[Zoho CRM\nDeals API] D -- Fetch Report --> G[Zoho CRM\nAnalytics API] D -- Create Contact --> H[Zoho CRM\nContacts API] E & F & G & H --> I[Python Response\nFormatter] I --> J[GPT-4.1\nNatural Language Reply] J --> K([📱 Flutter UI\nStructured Result]) classDef default fill:#111,stroke:#FF9900,stroke-width:2px,color:#fff; classDef mcp fill:#1a0f00,stroke:#FF9900,stroke-width:3px,color:#FF9900; classDef zoho fill:#0a1a00,stroke:#FF9900,stroke-width:2px,color:#fff; class D mcp; class E,F,G,H zoho;
04
Agentic AI

Doctor's Agentic AI Assistant

A clinical-grade agentic AI chatbot that acts as an intelligent intake assistant for doctors. The system collects comprehensive patient symptom data through multi-turn conversation, synthesizes a structured clinical summary, and presents it to the physician who then decides on prescriptions and recommendations. OpenAI GPT-4o is primary; Gemini text-bison serves as an automatic fallback.

KEY CAPABILITIES
  • Multi-turn agentic conversation for symptom collection
  • Automatic OpenAI → Gemini failover with zero downtime
  • Structured patient summary generation for physicians
  • DynamoDB for persistent patient session & history
  • HIPAA-conscious data flow on AWS EC2
TECH STACK
OpenAI GPT-4o/4.1/5.1 Gemini text-bison AWS EC2 DynamoDB Clinical NLP HIPAA Design
Agentic AI Clinical Flow
flowchart TD A([🧑 Patient\nStarts Chat]) --> B[Agentic AI\nIntake Agent] B --> C{LLM Router\nHealth Check} C -- Available --> D[OpenAI GPT-4o/4.1/5.1\nPrimary] C -- Unavailable / Error --> E[Gemini text-bison\nFallback] D & E --> F[Multi-turn Q&A\nSymptom Collection] F --> G{All Required\nData Collected?} G -- No --> F G -- Yes --> H[Clinical Summary\nSynthesis Engine] H --> I[(DynamoDB\nSession Store)] H --> J[📋 Structured\nClinical Report] J --> K([👨‍⚕️ Doctor\nReviews Summary]) K --> L[Prescription /\nRecommendations] L --> M[(DynamoDB\nPatient Record)] classDef default fill:#111,stroke:#D3212D,stroke-width:2px,color:#fff; classDef primary fill:#1a0505,stroke:#D3212D,stroke-width:3px,color:#D3212D; classDef fallback fill:#0a0a1a,stroke:#4285F4,stroke-width:2px,color:#fff; classDef db fill:#0a0505,stroke:#D3212D,stroke-width:2px,color:#fff; class D primary; class E fallback; class I,M db;

Ready to Build Your AI Solution?

Let's architect your next enterprise AI project — from proof of concept to production deployment on GCP, AWS, or Azure.

Start a Conversation
AI & Agentic Hubs
Sydney Melbourne Bangalore Mumbai
Zoho Centers
Coimbatore Chennai Trichy India Hub
Cloud Partners
Australia USA Singapore UK
Global Service Network

As an elite digital transformation agency, Leskode serves as a premier Zoho Implementation partner in Coimbatore, Zoho Implementation partner in Trichy, Zoho Implementation partner in Chennai, Zoho Implementation partner in Bangalore, Zoho Implementation partner in Australia, Zoho Implementation partner in China, Zoho Implementation partner in Hyderabad, Zoho Implementation partner in Karnataka, Zoho Implementation partner in Tamil Nadu, Zoho Implementation partner in United States, Zoho Implementation partner in Canada, Zoho Implementation partner in Mexico, Zoho Implementation partner in Germany, AI Implementation partner in Coimbatore, AI Implementation partner in Trichy, AI Implementation partner in Chennai, AI Implementation partner in Bangalore, AI Implementation partner in Australia, AI Implementation partner in China, AI Implementation partner in Hyderabad, AI Implementation partner in Karnataka, AI Implementation partner in Tamil Nadu, AI Implementation partner in United States, AI Implementation partner in Canada, AI Implementation partner in Mexico, AI Implementation partner in Germany, Google Cloud Implementation partner in Coimbatore, Google Cloud Implementation partner in Trichy, Google Cloud Implementation partner in Chennai, Google Cloud Implementation partner in Bangalore, Google Cloud Implementation partner in Australia, Google Cloud Implementation partner in China, Google Cloud Implementation partner in Hyderabad, Google Cloud Implementation partner in Karnataka, Google Cloud Implementation partner in Tamil Nadu, Google Cloud Implementation partner in United States, Google Cloud Implementation partner in Canada, Google Cloud Implementation partner in Mexico, Google Cloud Implementation partner in Germany, Azure Implementation partner in Coimbatore, Azure Implementation partner in Trichy, Azure Implementation partner in Chennai, Azure Implementation partner in Bangalore, Azure Implementation partner in Australia, Azure Implementation partner in China, Azure Implementation partner in Hyderabad, Azure Implementation partner in Karnataka, Azure Implementation partner in Tamil Nadu, Azure Implementation partner in United States, Azure Implementation partner in Canada, Azure Implementation partner in Mexico, Azure Implementation partner in Germany, AWS Implementation partner in Coimbatore, AWS Implementation partner in Trichy, AWS Implementation partner in Chennai, AWS Implementation partner in Bangalore, AWS Implementation partner in Australia, AWS Implementation partner in China, AWS Implementation partner in Hyderabad, AWS Implementation partner in Karnataka, AWS Implementation partner in Tamil Nadu, AWS Implementation partner in United States, AWS Implementation partner in Canada, AWS Implementation partner in Mexico, AWS Implementation partner in Germany.