Setting up for agentic AI

One of the best supervillain perks is an endless supply of labor-saving henchmen. So, while it’s terrific that artificial intelligence (AI) can streamline data gathering, organization and analysis, why stop there? Maybe it can perform a few more preselected chores while it’s out there, including carrying out preprogrammed responses to conditions it encounters?

That’s what agentic AI is. It’s your minion that does things for you, and it’s getting more capable with increasingly complex goals, plans, tools and actions.

This shouldn’t be a shock for process control engineers, who’s whole gig is predicated on using automation to do more with less, and save time, money and labor. It’s just that now it’s software that’s automating itself, and achieving increasing autonomy. As usual, while consumer and business users charge ahead with developing AI and agentic AI tools and solutions, the process industries hang back, and experiment with them in similarly non-critical settings with help from simulation, Internet search, virtualization, analytics and other forms of software-based digitalization.

Installments

1. Coordinate convergence and calm complexity: HighByte uses AI and AWS to help glassmaker streamline data infrastructure.
2. Data-source links let LLMs understand and deliver: Eli Lilly relies on HiveMQ as core MQTT hub for secure, real-time data exchange.
3. AI algorithm helps optimize AGR, and save amine, steam and power: Yokogawa deployed AI agents at Aramco’s Fadhili gas plant.
4. Agentic environment uses AI to accelerate decision-making: Emerson’s AspenTech Subsurface Intelligence (ASI) segments AI for machine learning (ML) and model context protocol (MCP).
5. Gen AI and agentic AI development tools: Some of the best known and reportedly most widely applied solutions.