This is the time of year when every board is going through their annual planning cycle. The big topic in every board conversation has been how we should think about using AI in our company this coming year.

The priority is how we are going to apply AI, Large Language Models, and “Agentic” AI to drive efficiency in our businesses and take out costs.

At a high level many of my recent board discussions are concluding that they can target to take 10% of our head count out of our total employee count by the end of 2026.They believe Agentic AI is the enabler of these savings.

The areas of priority focus are on the low risk, most proven use cases. Targeted use cases are from the headquarter staff functions such as HR, Finance, Legal, Compliance and IT. From the front office customer facing groups such as call center and customer support functions are clear use cases. Start up companies are solving these groups’ business flows. These areas are being rethought, redesigned, and redeployed with empathetic humanoid chatbots, yielding higher user /  customer satisfaction.

Next year is the first phase proving out these use cases for many companies. Most companies are still deploying “proof of concept” use cases.

Once the companies have tested, derisked and concluded these Agentic solutions are working to replace specific roles, they will be scaled up, later in 2026

One of the consistent themes across my public and private corporate boards is the focus on deployment of Agentic AI.

The tech terms are confusing. Here’s a refresh on the difference between AI, LLMs and Agentic AI.

In the early 2020’s AI became embedded in most SaaS Enterprise Software. In late ‘22/’23  LLMs “Generative AI” entered the scene with Open AI, Chat GPT hitting both business and consumers at scale in late ‘22. By early ‘23 generative AI was becoming widely adopted.

2025 is when “Agentic” becomes broadly discussed.

Agentic basically operationalizes AI to perform a job. The new Agentic empathetic humanoid intelligent chatboat is going to replace call center / customer service personnel.

That same Agentic tech is going to replace many other “persona” roles throughout corporate America.

For example – in human resources, your benefits manager may be replaceable with an Agentic benefits agent. In finance, you may have an Agentic collections / Agentic payables agent. In legal, you may have an Agentic compliance agent and in manufacturing you may have procurement augmented with Agentic buying agents obtaining competitive quotes. Lastly, software coders are rapidly being augmented and many IT organizations have clear slowdowns or headcount reductions planned.

When you look at the back office organization of finance, legal, HR and IT, many of these organizations are the first use cases that have been most de-risked and proven.

For example, as of 2025, Meta generates up to 30% of its internal SW developed codebase with Agentic developers and plans to reach 50% by 2026.

Additionally, United Health Care had 130K call center workforce, down 40K in 2025 and down another 30K in 2026, with increased customer satisfaction scores.

Every company I have met with is looking at possibly taking 10% headcount out of their organization in the coming year.

I believe that every engaged board member should be thinking about asking management how they are looking at Agentic deployment.

My guess is that most companies will be doing “Proof Of Concept” (P.O.C.) small scale deployment of these types of functional or other proven use cases.

I expect that in Q4 of 2026 we will start to see the potential headcount reductions flow through our organization as these “POC”s bear fruit becoming deployed more widely.

May I suggest we consider asking management how they plan to stay competitive? We may want to ask our leadership teams and our fellow board colleagues: “Should we be a leader in Agentic deployment? Should we be a fast follower, or should we be deploying later? Is Agentic a fit for our organization, or is our business model so different that for some reason there is no efficiency gained?”

This is a conversation I recommend to be on our board agenda for Q4, Q1.

Boards must be “change adaptive” looking at forward opportunities and risks. Coming up the AI, LLM and Agentic learning curve is important so that we are informed, comfortable. We need the frame of reference to be able to have these thought provoking discussions, looking at the big opportunities, transformations, and challenges going into our 2026 planning cycle.

Read original article