Certificate Programme · Dubai / Online
AI Agents for Business Applications: Practitioner Programme
Build practical AI agents that work with documents, databases, APIs and real business workflows.
This hands-on programme guides you through the design, development and deployment of controlled AI agents. You will build retrieval systems, create controlled tools, connect agents to approved data sources, design multi-step workflows and evaluate their reliability.
Based on practical systems developed at Innosoft Gulf, the programme uses managed development environments, local language models and private AI infrastructure.
Register now
Quick facts
Duration
15–16 hours live
Format
8 sessions × 2 hours, or 6 sessions × 2.5 hours
Delivery
In-person or Zoom
Location
Dubai Knowledge Park
Level
Intermediate · Practitioner
Certificate
Innosoft Gulf certificate · KHDA attestation available
Next cohorts
Morning Project Track
21 July – 13 August 2026
Tuesday and Thursday · 10:00 AM–12:00 PM GST · 8 sessions
Attend in person at Dubai Knowledge Park or live online via Zoom.
USD 1,250
VAT inclusive
The fee includes live instruction, managed practical labs, course materials and an Innosoft Gulf certificate of completion. KHDA attestation is available.
Evening Practitioner Track
27 July – 7 August 2026
Monday, Wednesday and Friday · 6:30 PM–9:00 PM GST · 6 sessions
Attend in person at Dubai Knowledge Park or live online via Zoom.
USD 1,250
VAT inclusive
The fee includes 15 hours of live instruction, managed practical labs, course materials and an Innosoft Gulf certificate of completion. KHDA attestation is available.
What you will build
During the programme, you will develop an AI agent that can:
Work with approved documents and business knowledge
Connect to databases, APIs and controlled tools
Follow structured multi-step workflows
Validate inputs and handle errors safely
Produce grounded responses with clear sources
Operate using local or private language models
Curriculum
6 core modules · delivered across 8 sessions (Morning Track) or 6 sessions (Evening Track)
The programme draws on Innosoft Gulf’s Market Intelligence Agent as an instructor case study, demonstrating request classification, parameter validation, controlled database tools, local language models and Telegram deployment.
Session 1
Structured LLM Applications
- Chatbot versus workflow versus agent
- Cloud versus local language models — Ollama and Qwen
- Structured output and Pydantic schemas
- Semantic request classification — rule-based versus LLM routing
- Managed Python development environment and guided implementation workflow
Practical outcome
Build a controlled request classifier and extract structured parameters from natural language.
Session 2
RAG Foundations
- Document ingestion and PDF processing
- Chunking and embeddings
- Vector databases and semantic search
- Weaviate integration
Practical outcome
Build a document-based question-answering agent.
Session 3
Advanced RAG and Evaluation
- Metadata filtering and hybrid search
- Reranking and retrieval quality
- Citations and grounded generation
- Evaluation and reducing hallucinations
Practical outcome
Improve and evaluate the document agent from Session 2.
Session 4
Tools, APIs and Business Data
- Python tool design and function calling
- PostgreSQL access and REST APIs
- Read-only database users and safe SQL practices
- Input validation, deterministic outputs and error handling
Practical outcome
Build an agent that queries a controlled database or API.
Session 5
LangGraph Workflows
- Nodes, edges and state
- Routing, semantic classification and parameter extraction
- Clarification branches and conversation persistence
- Human approval, retry handling and auditability
Practical outcome
Build a controlled multi-step agent workflow.
Session 6
Integration and Deployment
- Connecting agents to Telegram using long polling — no public IP required
- Webhooks and FastAPI as an alternative deployment pattern
- Ollama integration and local model deployment
- systemd, Docker overview, logging and health checks
- Security, rate limiting and final project demonstration
Practical outcome
Deploy a working agent and present your final project.
Practical learning environment
Participants work in managed development environments running on Innosoft Gulf’s private AI and big-data infrastructure. The practical labs include:
Local language models through Ollama and Qwen
Document retrieval and vector search using Weaviate
Controlled PostgreSQL database access
Python tools and REST API integrations
Selected HDFS and Spark data examples
Telegram integration and local application deployment
Each participant works in an isolated training environment with approved datasets, read-only accounts and controlled system access.
Who this programme is for
Python developers and software engineers
Data analysts and data scientists
Technical consultants and solution architects
AI practitioners building business applications
Technical managers and founders
Participants should be familiar with basic Python, APIs or data workflows. Advanced software engineering experience is not required, but this is a technical, hands-on practitioner programme.
Ready to build AI agents?
Limited to 12 participants per track. The Morning Project Track starts Tuesday 21 July 2026; the Evening Practitioner Track starts Monday 27 July 2026.
Register now
