GenerativeAI and Automation: Shifting Hard Left to Autonomous Systems

Over the last decade IT has increasingly made data accessible and systems, infrastructure and devices #programmable via #APIs giving developers more power (and responsibility) - it shifted effort left, but didn't materially reduce the effort or complexity.

Now those same APIs, particularly when exposed through a high-level abstraction, can integrate and train #AIML, expose #NLP interfaces, and interact directly with #GenerativeAI. This is the logical and inevitable trajectory to #AutonomousSystems which can observe their own behavior, trouble-shoot and mitigate issues in a continuous #ClosedLoop. The demands for real-time data-driven platforms that optimize actions and processes is hurdling IT towards autonomous, self-organizing systems.

All forms of AI/ML inherently favor models over hard-code and dynamic non-linear processes over static workflows, because #declarative and #dynamicsystems fully separate domain logic from implementation, allowing for the greatest runtime flexibility and agility, which is the whole point of being real-time.

It will start with low-hanging fruit. #RPA was really a workaround for challenges in formally integrating data across silos so it's prime for informal automation methods like #ChatBots. Complex, enterprise-grade, end-to-end automation is not going away anytime soon. However, Large Language Models (LLMs) will expedite the death of static, tightly-coupled, vertically-integrated stacks and compiled applications because they are not responsive or agile by design.

All forms of AI/ML inherently favor models over hard-code and dynamic non-linear processes over static workflows, because #declarative and #dynamicsystems fully separate domain logic from implementation, allowing for the greatest runtime flexibility and agility, which is the whole point of being real-time.

A high-level abstraction provides a unified interface for an LLM to semantically and syntactically understand a domain. The breadth of the model effectively sets the scope for end-to-end automation across business silos, ecosystem partners and SaaS services. The depth of the domain model defines how it can work across layers of technology protocols (OSI 1-7).

#LLMs are naturally aligned with high-level #ModelDriven #IntelligentAutomation platforms that are dynamic, loosely-coupled, horizontally-architected platforms like EnterpriseWeb , because their #eventdriven #integration , #orchestration and #automation "middleware" capabilities are designed to dispatched on-demand (#FaaS) and to be dynamically configured by models that provide #context and #policies.

EnterpriseWeb, which is #APIfirst, can expose its models and capabilities to LLMs through a unified interface, just as it does for developers. Over time, LLMs take on increasingly complex design and management decisions.

The Future is Here. If you want to see a #Telecom example of autonomous networks in action, please watch the replay of our award-winning #Intel #5G #Edge demo where, in conjunction with our AI/ML partner KX we are demonstrating a #selfscaling, #selfhealing, #selfoptimizing network that delivers consistent, predictable #LowLatency and #EnergyEfficiency at scale. The demo features partners 英特尔 , 红帽 , Fortinet , Keysight Technologies and Tech Mahindra

Copyright 2023, EnterpriseWeb LLC