TMA AI Newsletter Quarter II of 2025

A set of AI Agents to increase productivity in software development life cycle.

• Multi-agent systems
• No-code/low-code architecture
• Cloud-native infrastructure
• DevOps & CI/CD automation
• AI-based software design & code generation
• Security & privacy technologies

A flexible framework for rapidly developing object detection solutions, enabling real-time alerts and automated responses.

• Dynamic response system: Detect issues and activates the right automated response based on urgency
• Self-customizable system: Preloaded with cases that you can tailor or expand using natural language 

●    AI frameworks: Vision-Language Model, GPT-4v 
●    Video streaming: NVIDIA DeepStream, NVIDIA NIM, NVIDIA NeMo 
●    Edge AI: Jetson Orin

A no-code AI framework for developing intelligent, agent-driven workflows with autonomous AI agents that understand context, make decisions, and act across systems.

• Integration hub:  Support Model Context Protocol (MCP) with built-in tools for flexible integration
• No-code workflow builder: Enable creating AI workflows with drag-and-drop
• Knowledge management: Manage and test knowledge bases with built-in tools
• Agent & framework SDK: Build and customize agents for enterprise needs

• Multi-agent framework: Model Context Protocol (MCP)
• AI & LLM models: DeepSeek, LLaMA (Meta), Qwen (Alibaba)
• Embedding models: Hugging Face
• Backend layer: Python, FastAPI, Spring Boot
• Communication: Redis, gRPC, REST API

This Agent can create test cases from user stories, simulate manual testing steps, and writing automation scripts. This Agent can adapt to system changes with self-healing capabilities, reducing the need for constant test maintenance.

• Generate test cases from story: Automatically create test cases from user stories
• Mimic manual testing execution: Perform clicks, data entry, and navigation like a real tester
• Simulate automation scripting process: Write test scripts without manual coding
• Self-healing/maintenance: Adapt to UI changes automatically

• Front-end: React, Next.js, TailwindCSS
• Back-end: Spring Boot, FastAPI
• Infrastructure: MinIO, PostgreSQL, Redis, Kafka, Keycloak, Kubernetes
• AI/LLMs: LangChain, LlamaIndex, Claude, OpenAI

This solution simplifies patient check-ins and provides clear guidance. It also integrates with HIS, EMR, and scheduling systems for better coordination.

• Symptom input via text or image: Patients can submit symptoms by typing or uploading photos
• Medical AI understanding: Analyze inputs using trusted medical datasets for accurate assessment
• Clear health guidance: Deliver easy-to-follow advice based on symptom analysis
• System integration: Connect with HIS, EMR, and scheduling systems for updates and bookings

• TMA Data Platform
• TMA Agentic AI Framework
• Gemini

An AI assistant that responds to client questions in real-time. It learns from your documents, is easy to set up with a drag-and-drop interface, and works seamlessly across your website and social media channels

• Ingest knowledge from multiple data sources (web, text, doc, PDF, internal systems)
• Route complex cases to the right human agents with full conversation history and client data
• Track client satisfaction and emotional tone for continuous improvement
• Analyze queries and preferences to suggest relevant products or services in real time
• Provide instant translation for multilingual, global client support
• Integrate across websites and social media channels (Facebook, Telegram, TikTok, Zalo, etc.)

• Front-end: Next.js, CopilotKit, Tailwind CSS
• Back-end: FastAPI, LangGraph, PostgreSQL, LightRAG
• Infrastructure: Docker
• LLM models: Qwen3, OpenAI 

An AI agent chatbot integrated across web and social channels to automate product queries, confirm purchases, and track orders - enhancing the full client journey.

• Sales assistant: Help clients find products, order, and get answers
• Auto order creation: Create orders instantly via system integration
• Auto handoff: Switch to a human agent when needed
• Easy integration: Integrate to websites and social platforms

• Multi-agent framework: ADK, A2A, MCP
• AI & LLM models: OpenAI, Google AI
• Embedding models: OpenAI embeddings model, Google AI embedding model
• Backend layer: Python, FastAPI
• Database & storage: MongoDB, MinIO