Abstract

I will cover a GenAI framework to more quickly build enterprise applications, to accelerate the application development process.

 

 

 

Features to differentiate the framework includes:

* medium term memory between multiple agents and users, at a work-task level.  Personalization also works at this level. 

 

* Meaning Markers for auto-tagging, extracting structured data from user text input such as from chatbot or emails.  This supports proprietary detailed jargon.

 

* Design of Experiments (DoE) is not a user feature, but a feature to accelerate the development process of a new vertical application, to enable testing of any architecture decisions, such as:  chunking, embedding models, context engineering, LLM’s

 

* Defense against Scheming.  First I will describe how frontier models can scheme, to define the problem. If you don’t monitor for scheming, you won’t know when your LLM’s / agents may decide to lie to you.  Defenses include Guardrails monitoring the thought processes of an agent (where possible).  Other defenses include architectural solutions, such as separating

 

 

I will review the Ccube accelerator architecture and how that supports vertical applications.  Components include IntelliConnect (i.e. RAG features) and Guardiance (with Guardrails features).  Both support adaption over time, which includes updating few-shot learning.

 

Who Is this presentation for?

 

People or organizations that want to develop and deploy enterprise AI applications more reliably and faster.  You can leverage ideas from the design principles and architecture design.

 

Greg Makowski has been training and deploying AI since 1992.  He has been growing Data Science Teams since 2010, has exited 4 startups and has 6+ AI patents granted or under review. He is currently the Chief of Data Science at Ccube, leading GenAI/AI consulting and framework development for enterprise applications. See also:

 

https://www.ccube.com/luminai    (framework details)

 

https://www.ccube.com/genai  (consulting services)

 

https://www.linkedin.com/in/GregMakowski  (connect and contact)