watsonx.ai

Hey everyone 👋 - I've been digging into the Enterprise AI Development Survey IBM published in January (link). There's a lot in there that aligns with what we've been seeing too - especially around tool sprawl, integration challenges, and how much friction teams still face even with all the GenAI momentum.

A few data points really jumped out:

• 72% of devs use 5–15 tools to build an AI app. (13% use more than 15!)
• Integration with existing tools was one of the most essential - and also most lacking - traits in current solutions.
• And 99% are exploring or developing AI agents, but it's not clear how many are using RAG in production yet.

This got me wondering - and I'd love to hear how others are thinking about this:

1. Which tools or data sources have been the hardest for you to integrate into your GenAI stack - and what made it difficult? (e.g., weird APIs, security policies, semi-structured content, etc.)

2. Is your team using RAG in production today? If so, how are you managing things like content drift, retrieval quality, or keeping knowledge aligned with changing LLMs?

My company is building some tools to support retrieval-based systems and would love to learn from others working through these same challenges - especially anything surprising you've learned or workarounds you've come up with.

Thanks in advance for any insights!

P.S. I wasn't sure which community/group would be the most appropriate to post about this, so I guessed this one.

#GenerativeAI

------------------------------
Matt Genovese
CEO, Planorama
"Time to market, accelerated by DESIGN"
Contact: https://planorama.design/matt
------------------------------

Thanks for sharing your insights from the report Matt.

Nick

Hey everyone 👋 - I've been digging into the Enterprise AI Development Survey IBM published in January (link). There's a lot in there that aligns with what we've been seeing too - especially around tool sprawl, integration challenges, and how much friction teams still face even with all the GenAI momentum.

A few data points really jumped out:

• 72% of devs use 5–15 tools to build an AI app. (13% use more than 15!)
• Integration with existing tools was one of the most essential - and also most lacking - traits in current solutions.
• And 99% are exploring or developing AI agents, but it's not clear how many are using RAG in production yet.

This got me wondering - and I'd love to hear how others are thinking about this:

1. Which tools or data sources have been the hardest for you to integrate into your GenAI stack - and what made it difficult? (e.g., weird APIs, security policies, semi-structured content, etc.)

2. Is your team using RAG in production today? If so, how are you managing things like content drift, retrieval quality, or keeping knowledge aligned with changing LLMs?

My company is building some tools to support retrieval-based systems and would love to learn from others working through these same challenges - especially anything surprising you've learned or workarounds you've come up with.

Thanks in advance for any insights!

P.S. I wasn't sure which community/group would be the most appropriate to post about this, so I guessed this one.

#GenerativeAI

------------------------------
Matt Genovese
CEO, Planorama
"Time to market, accelerated by DESIGN"
Contact: https://planorama.design/matt
------------------------------