I'm excited to join this amazing community as the Product Manager for watsonx.ai, specifically focused on Model Customization and Advanced RAG. I'll be here regularly to share insights, gather your feedback, and help shape our roadmap based on your real-world needs.

Quick Overview: What watsonx.ai Brings to Model Customization

🎯 Fine-Tuning at Scale

• Domain-specific adaptation of foundation models (Llama 2, Granite and more)
• Parameter-efficient tuning methods (LoRA, QLoRA) to reduce costs and training time
• Enterprise-grade security with your data never leaving your environment
• Automated hyperparameter optimization to get the best results with minimal effort

🔍 Advanced RAG Capabilities

• Basic RAG - Easy to use RAG with "Chat with documents" in Prompt Lab
• Auto AI for RAG - Automate and optimize RAG pipelines by taking guesswork away
• Agentic RAG - Template to kick off advanced Agentic RAG workflows (intelligently search multiple grounding documents, intelligently decide which tool to call

🛠️ What Makes watsonx.ai Different

• Unified platform - No need to stitch together multiple tools
• Production-ready scaling from prototype to enterprise deployment
• Governance and compliance built-in for regulated industries
• Cost optimization with efficient training and inference options

I want to hear from you! What is your biggest challenge when you want to build use cases which need SLMs/LLMs and AI Agents to understand enterprise proprietary data?

• Data preparation and quality for training
• Cost management during fine-tuning and inference
• Performance optimization and latency requirements
• RAG accuracy and relevance issues
• Integration complexity with existing systems
• Governance and compliance requirements
• Scaling from prototype to production
• Something else (please comment!)

🗣️ Discussion Starters:

1. What use cases are you most excited to tackle with customized models?
2. Which RAG challenges keep you up at night?
3. What would make model customization 10x easier for your team?

📈 Share Your Journey:

• Are you currently fine-tuning models? What's your approach?
• How are you handling retrieval and knowledge management today?
• What would convince you to switch or upgrade your current solution?

Wow, great set of capabilities.  Are there link you can share for partners to start learning more about everything?

------------------------------
ANTHONY STEVENS
Principal AI Engineer, IBM Build Engineering
San Francisco
------------------------------

Original Message:
Sent: Tue July 01, 2025 03:19 PM
From: Suhas Kashyap
Subject: Introducing watsonx.ai Model Customization: What's Your Biggest Challenge?

I'm excited to join this amazing community as the Product Manager for watsonx.ai, specifically focused on Model Customization and Advanced RAG. I'll be here regularly to share insights, gather your feedback, and help shape our roadmap based on your real-world needs.

Quick Overview: What watsonx.ai Brings to Model Customization

🎯 Fine-Tuning at Scale

• Domain-specific adaptation of foundation models (Llama 2, Granite and more)
• Parameter-efficient tuning methods (LoRA, QLoRA) to reduce costs and training time
• Enterprise-grade security with your data never leaving your environment
• Automated hyperparameter optimization to get the best results with minimal effort

🔍 Advanced RAG Capabilities

• Basic RAG - Easy to use RAG with "Chat with documents" in Prompt Lab
• Auto AI for RAG - Automate and optimize RAG pipelines by taking guesswork away
• Agentic RAG - Template to kick off advanced Agentic RAG workflows (intelligently search multiple grounding documents, intelligently decide which tool to call

🛠️ What Makes watsonx.ai Different

• Unified platform - No need to stitch together multiple tools
• Production-ready scaling from prototype to enterprise deployment
• Governance and compliance built-in for regulated industries
• Cost optimization with efficient training and inference options

I want to hear from you! What is your biggest challenge when you want to build use cases which need SLMs/LLMs and AI Agents to understand enterprise proprietary data?

• Data preparation and quality for training
• Cost management during fine-tuning and inference
• Performance optimization and latency requirements
• RAG accuracy and relevance issues
• Integration complexity with existing systems
• Governance and compliance requirements
• Scaling from prototype to production
• Something else (please comment!)

🗣️ Discussion Starters:

1. What use cases are you most excited to tackle with customized models?
2. Which RAG challenges keep you up at night?
3. What would make model customization 10x easier for your team?

📈 Share Your Journey:

• Are you currently fine-tuning models? What's your approach?
• How are you handling retrieval and knowledge management today?
• What would convince you to switch or upgrade your current solution?

I am intending to host my application on IBM Cloud. What cost should a gear my self towards. My application has a number of complex functionalities including AI assistance and an agent capable of deployment of code. The idea is to use Watson AI in further agents development 

Please advise 

------------------------------
Dr Terry Ramabulana
Founder & MD
TISholdings Pty Ltd
Sandton
+27834558115
------------------------------

Original Message:
Sent: Tue July 01, 2025 03:19 PM
From: Suhas Kashyap
Subject: Introducing watsonx.ai Model Customization: What's Your Biggest Challenge?

I'm excited to join this amazing community as the Product Manager for watsonx.ai, specifically focused on Model Customization and Advanced RAG. I'll be here regularly to share insights, gather your feedback, and help shape our roadmap based on your real-world needs.

Quick Overview: What watsonx.ai Brings to Model Customization

🎯 Fine-Tuning at Scale

• Domain-specific adaptation of foundation models (Llama 2, Granite and more)
• Parameter-efficient tuning methods (LoRA, QLoRA) to reduce costs and training time
• Enterprise-grade security with your data never leaving your environment
• Automated hyperparameter optimization to get the best results with minimal effort

🔍 Advanced RAG Capabilities

• Basic RAG - Easy to use RAG with "Chat with documents" in Prompt Lab
• Auto AI for RAG - Automate and optimize RAG pipelines by taking guesswork away
• Agentic RAG - Template to kick off advanced Agentic RAG workflows (intelligently search multiple grounding documents, intelligently decide which tool to call

🛠️ What Makes watsonx.ai Different

• Unified platform - No need to stitch together multiple tools
• Production-ready scaling from prototype to enterprise deployment
• Governance and compliance built-in for regulated industries
• Cost optimization with efficient training and inference options

I want to hear from you! What is your biggest challenge when you want to build use cases which need SLMs/LLMs and AI Agents to understand enterprise proprietary data?

• Data preparation and quality for training
• Cost management during fine-tuning and inference
• Performance optimization and latency requirements
• RAG accuracy and relevance issues
• Integration complexity with existing systems
• Governance and compliance requirements
• Scaling from prototype to production
• Something else (please comment!)

🗣️ Discussion Starters:

1. What use cases are you most excited to tackle with customized models?
2. Which RAG challenges keep you up at night?
3. What would make model customization 10x easier for your team?

📈 Share Your Journey:

• Are you currently fine-tuning models? What's your approach?
• How are you handling retrieval and knowledge management today?
• What would convince you to switch or upgrade your current solution?