Sign In

Kence Anderson

Industrial-Strength Intelligent Autonomous Agents

Kence Anderson is a seasoned technical translator, storyteller, and business builder, specializing in the introduction of cutting-edge technologies to the market. Currently, Kence is actively engaged in bringing Deep Reinforcement Learning and Autonomous AI to the forefront.

Highlights

Aug 29 · linkedin

I once consulted for a client that had deployed thousands of robotic process automation (#RPA) bots in production but asked me to help design Autonomous #ai for some of their process decisions that were too complex for the rules-based decision-making in RPA.

This consultation happened before transformers, but #generativeai could not have made this decision well either.

But, the engineering mindset tells us that each of these automation techniques are valuable for specific parts of the enterprise autonomy. So, instead of pushing one technique aside as "obsolete," let's match them to their contributions to the Autonomous Enterprise.

📏 📐 Rules-Based Automation for Simple Repetitive Tasks There are millions of tasks in every enterprise that are repetitive and well automated. This layer of the Autonomous Enterprise helps reduce human boredom and human error. These tasks tend to be tedious, but lower value. Automate many tasks to generate ROI from the automation.

Examples: Document Processing, List Gathering, Form Processing, Factory Automation

💬 📥 Generative AI for Autonomy in Language Tasks Large Language Models (#LLM) leverage the statistical arrangement of letters to generate text and multi-modal responses to natural-language queries. This use of language is not a decision-making paradigm, but can provide autonomy for tasks that are uniquely limited to language.

Examples: Customer Support (Chat), Sales Automation (Email), Document Summarization & Analysis

🧠 🥅 Intelligent Agents for Complex, Nuanced, Decisions But the most valuable decisions in the enterprise are too complex for rules-based automation and live outside the realm of language manipulation. They require intelligence characteristics like perception, strategy, adaptatbility, and deduction. These tasks always have goals (often competing goals that trade-off) and take humans about a decade to master. Each of these decisions generates millions of dollars of ROI when made well.

Examples: Stock Trading, Financial Allocation, Supply Chain Optimization, Industrial Process & Machine Control

David C Williams (He/Him), Andreas Welsch, Andreas Welsch, Brian Evergreen, Doug Shannon 🪢, Jeff Winter, Kary Bheemaiah

#intelligentagents #engineeredintelligence #machineteaching

Aug 19 · linkedin

It was a pleasure to present the keynote at the Procter & Gamble Intelligent Controls and Automation Symposium near Cincinnati last week.

Manufacturing R&D groups like this one are crucial to making #intelligentagents work in real life, and this group in particular has a very sophisticated take on how to drive technology adoption into plants and how to hire and retain key tech innovators.

Here's a few key points that we discussed:

🧠 Intelligent Controls and Automation will move forward as we effectively stitch together enabling #ml / #ai technologies with domain expertise under the supervision of #engineers.

📄 Joshua Carroll, brought up a key point that we need a methodology, analogous to Six Sigma for lean manufacturing to educate millions of engineers on intelligent controls in a standardized way. I hope that #machineteaching will be the genesis of such a movement.

⚠⚙ Roland Joseph, raised a warning about what will happen if we pursue #datascience apart from engineering subject matter expertise. He told a hilarious story about a data scientist that "just wanted the data" then returned with insights about how a pump worked, insights that are defined by fluid dynamics.

Thank you Steve Ziegler, for such an engaging, well organized symposium.

It was also fun to hang out with the Rockwell Automation crew at the exhibition booth and it was great to see 50+ engineers come out for the Composabl platform demo!

Michael Hanavan, David Miracle, Jim Sell, Brad Kerr, Jordan Reynolds, Paul Thomas

Kence Anderson
Add to my network

Location

San Francisco Bay Area