Our Services / AI Services / Workflow Automation / What Is an AI Agent
- Plain Language, No Hype
What Is an AI Agent, Really?
The word is doing a lot of work in a lot of sales decks. Here is what an agent actually is, how it differs from a chatbot and from ordinary automation, and when a small business genuinely needs one.
The Definition
An Agent Takes Actions, Not Just Turns
A chatbot responds to you. You ask, it answers, the exchange ends, and any action that follows is taken by a human. An agent is given a goal and then takes a series of steps toward it on its own, reading a system, deciding what to do next, doing it, checking the result, and continuing until the goal is met or it gets stuck.
That difference sounds technical and is actually a governance difference. A chatbot’s mistake produces a bad sentence someone reads before acting on it. An agent’s mistake produces a bad action that has already happened, an email sent, a record updated, an appointment moved. Everything that makes agents useful is the same thing that makes them require more careful design.
This is why we scope agent work more conservatively than most vendors do, and why our first question is never “what could this do” but “what should it be allowed to do without asking.”
Every agent we have put into production has been for a business inside the Dallas area, run from our office on Legacy Drive in Plano. That proximity matters more than it sounds. When an agent takes a wrong turn on a Tuesday afternoon, someone who already knows your systems can be at your office instead of booking a call for Thursday.
Telling them apart
Four Things People Call the Same Thing
Rule-Based Automation
“When this happens, do that.” Utterly predictable, cheap, and auditable. For most of what small businesses want automated, this is the correct answer and no AI is involved at all.
Use when: the rules can be written down completely.
A Chatbot or Assistant
Answers questions and drafts text on request. A person stays in the loop for every output. Low risk, immediate value, and where the large majority of businesses should start.
Use when: the input is messy and a human will review the output anyway.
An AI Step Inside a Workflow
Ordinary automation with one Al-powered decision in the middle, classify this email, extract these fields, route this request. The most under-rated pattern and the one we build most often.
Use when: the flow is fixed but one judgment call resists a rule.
A True Agent
Given a goal, chooses its own sequence of actions across systems. Genuinely capable, and the only one of the four where a mistake happens before a person sees it.
Use when: the path varies every time and the actions are reversible.
What we usually recommend
The third option, for most businesses, most of the time. An AI step inside a controlled workflow captures the majority of the benefit with a fraction of the risk, and it is far easier to explain to a client, a regulator, or an insurer. We build true agents where the case is strong, and we say so when it is not.
Guardrails
Four Rules We Apply to Any Agent We Build
Reversible actions only
Drafting, tagging, scheduling, yes. Sending externally, deleting, or moving money, a person confirms.
Narrow permissions
The agent gets its own account with the least access it needs, never a shared administrative login.
Everything logged
Every action attributable and reviewable after the fact. Without this you cannot investigate anything.
A stop condition
A defined point at which it escalates to a human rather than trying harder. Agents without one loop expensively.
A vendor is selling us an "agent." How do we tell what it is?
Ask one question: what actions can it take without a person clicking approve? If the answer is none, it is an assistant with better marketing, which may still be exactly what you want, at a chatbot price rather than an agent price.
Are agents safe for a medical practice or law firm?
They can be, on internal work with reversible actions and full logging. What we would not do is let an agent act unsupervised on anything client- or patient-facing. See assessments for regulated businesses.
Do we need an agent to get value from Al?
No, and most businesses our clients’ size get their first real return from an assistant or a single AI step inside an existing workflow. Agents are a later- stage tool for problems where the path genuinely varies each time.
BK
Written and reviewed by Bob Kehr, President & Founder of Kehr Technologies, Plano City Council member, and Plano Chamber Small Business Person of the Year 2022. He vets vendor claims about agents on behalf of clients, and has read a lot of decks that meant something simpler.
Last reviewed August 2026 · Questions we have not answered here? Call 214-444-3583. Kehr Technologies is based in Plano, Texas, and works with businesses throughout the Dallas area.