By function
The first agent win is usually in the work between departments, not in a new front-end.
21%
scaled agent use in service operations inside the technology sector. Elsewhere, scaled use is still in the single digits.
Stanford HAI, AI Index 20261Requests arrive by email. Documents need interpretation. Four systems need an update. Someone has to follow up. That is the operating cost.
How I see it
Operations agents
Operations agents fit a company big enough to have expensive workflows and not big enough to have an internal agentic team. The work is intake, routing, document reading, status, vendor coordination, and case completion.
This is where the $540K surface-area math shows up. Twenty people spending 30% of their time on a workflow is a real number. A conservative improvement funds a serious diagnostic.
The redesign should keep the operating model intact. Do not stand up a transformation office. Pick one flow. Draw the current handoffs. Put the agent on the outcome. Leave authority where it is.
Distribution, real estate, hospitality, specialty finance, and business services in Miami all have versions of this: packets, exceptions, and follow-up.
Common mistakes
What teams usually get wrong.
Starting with a company-wide operations agent
A general assistant has no baseline and no owner.
Automating the intake form only
If the work after intake is unchanged, you moved the queue.
Leaving follow-up human by default
Follow-up is often the majority of the cost.
A useful diagnostic
Five questions before you fund the work.
Which operational workflow has the most people touching it?
That is the first place to look.How much of the time is research, re-entry, and follow-up?
If it is a third or more, the economics are usually there.Is there a case object, or only an inbox?
An inbox is a process problem the agent will expose.What does done mean in one sentence?
If you cannot say it, you cannot agent it.Who feels the pain enough to own a 90-day path?
No owner, no win.
Economic model
Operations surface area20 people × $90,000 loaded cost × 30% workflow time = $540,000
A 25% improvement is $135,000. That is the kind of number that justifies a focused engagement.
Three credible paths
How far should you go?
Do not force one solution. Choose the path the economics, the risk, and the organization can support.
Improve the work you already have
Keep the process mostly intact and use AI on the bottlenecks that create delay, rework, or follow-up.
The workflow is already sound and a few steps create most of the friction.
Gains are usually incremental. The operating economics do not change much.
Redesign the workflow around AI
Question every handoff, queue, and duplicate step, then rebuild the process around what AI can now do.
The process grew over years and coordination now costs more than the work itself.
Requires process change, clearer ownership, and a willingness to retire old steps.
Put an agent on a high-value outcome
Give an agent responsibility for one valuable result across systems, with humans at the control points that require judgment, authority, or risk acceptance.
The workflow is high-value, variable, multi-step, and worth engineering for production.
Needs stronger architecture, evaluation, controls, and monitoring. This is not a prompt project.
When this is the wrong next step
Do not fund an agent here.
- The delays are caused by capacity you refuse to staff and a process you refuse to change.
- There is no volume. The work is a few special cases a month.
- Leadership wants a transformation roadmap more than one completed workflow.
A useful next step
Bring one workflow. Get guided into production.
We guide the implementation, go deep on the technical path, and stay hands-on through operations — or tell you when a simpler answer is better.
Discuss an AI opportunity