Should I hire or automate?
Hire or automate? It depends on the tasks, not the position. Automation and AI, up to agents that handle a task from start to finish, can take over the repetitive work of a position, rarely the whole position. Judgment, relationships and decisions remain human.
:quality(80))
The work is piling up and the reflex is to post a job. Then the question comes up: could software, or artificial intelligence, do part of that work?
The honest answer: it depends. AI assistants and agents now do things we thought only humans could do: read an email, fill out a form, prepare a follow-up. But no machine fully replaces a person's role. Depending on your industry and the tasks of the position, though, a good part of the work can already be done differently.
Why did some industries automate faster?
Technology has transformed industries unevenly. A company's level of technological maturity depends mostly on the nature of its tasks and its sector.
Manufacturing is the best example. Assembly-line work is repetitive, and every productivity gain counts. Robotics quickly became a necessity.
The more repetitive a task, the earlier it was handed over to a machine.
What computers have already changed in the office.
Management and services changed later, with digital tools. First, software made it possible to organize and circulate information, and paper declined.
Then came enterprise resource planning systems (ERP): a single piece of software that brings together sales, production, purchasing and accounting. Managers gained access to information from every department at the same time, and they could make better decisions faster.
Before, compiling financial or production data took a lot of time and created little value for the effort involved. Today, work has refocused on what matters.
Hire or automate: look at the tasks, not the position.
A position is a set of tasks. Some are monotonous and time-consuming: data entry, compiling, filing. Those are the ones automation takes over.
With AI, the list gets longer. Tasks that require reading and writing, like sorting emails, filling out a form from a document or preparing a reply, can now be handed over to software.
Other tasks require human skills that no machine can replace: creativity, complex decision-making, relationships with people and understanding context.
Yes, computers have made some tasks disappear. But those were mostly tasks that created little value, and it's largely thanks to technology that one person produces much more today than before.
What AI changes for an SMB.
Most teams already use an everyday assistant like Claude, ChatGPT, Copilot or Gemini. It helps write, summarize information, find answers and prepare content. Decisions are better informed.
The next step: connecting the assistant to the company's tools, like email, the calendar, customer files and internal documents. It no longer just answers. It finds the right information in your documents, prepares the quote, creates the task and updates the file.
In operations, AI can also spot what slows management down and suggest adjustments based on the context. AI handles the handoffs. People make the important decisions.
Examples
- Preparing a first draft of a text, an offer or a report
- Summarizing a meeting and creating the follow-ups in your tools
- Finding an answer in your internal documents, with the source
- Spotting the step that slows a process down and suggesting an adjustment
The Ville du Lac project is moving along well. Approval for the equipment budget is still missing.
AI agents: a task from start to finish, a person who approves.
An AI agent goes further than an assistant: you hand it a complete task, and it works through the steps. To follow up on overdue invoices, for example, it finds the invoices in question, writes a follow-up for each customer, waits for a person to approve it, sends it and logs the follow-up in the file.
That's what makes the question "hire or automate?" more concrete than before. Entire tasks of a position can now be handed over to an agent. Not the position: the tasks.
An agent doesn't work alone. It follows clear rules, it flags what's out of the ordinary and a person approves anything that goes out to a customer or commits the company. The goal isn't to make people work faster. It's to remove the steps that use their time without using their judgment.
Hello, our invoice from last month is still open. Could you confirm the payment date? Thank you!
In customer service, AI handles the simple questions.
AI understands language and intent better and better. So it has found its place in customer service and support.
An assistant connected to your documents (your policies, your guides, a file's history) answers simple questions on your website or by email, at any hour. When a request is out of the ordinary, it hands off to someone on your team, with all the context in hand. Your team keeps the cases that require judgment.
The quality of the answers depends on the quality of your documents. An up-to-date guide is worth more than one more assistant.
What AI doesn't change.
The tools have changed a lot. What makes an automation project succeed, much less so. When AI speeds up the work, choosing what to automate and doing it well becomes even more important.
The right problem
Not everything deserves to be automated, and not everything needs AI. Start with a task that comes up often, takes time and has a result that's easy to check.
Data in order
An agent that reads incomplete or contradictory files makes mistakes, and faster than a person. Before automating, make sure the information is up to date and stored in one place.
Security
Who has access to what, which data can leave the company, what the agent is allowed to do on its own: decide this before connecting AI to your tools.
Team adoption
A tool the team doesn't use solves nothing. Explain what AI will take on and what stays with people, then train them.
A person who decides
AI prepares, suggests and executes. A person keeps the decision on what matters: a price, an exception, a sensitive reply.
Where to start?
You may not be ready to hand an entire position over to a machine, and that isn't the goal. But using technology to support the work you do today is within reach. It has increased business productivity, given rise to new business models and opened global markets to small businesses.
Between the assistant your team already opens in a window and AI connected throughout your operations, there is a technological middle ground: automating one specific process, with a person who approves. That's often where your head start lies.
A good starting point: identify the leaders in your market and try to understand what got them there. Then pick one task from the position you want to fill and see whether off-the-shelf software already handles it or whether you need software built for you. We covered the question in this article on custom software. Good news: a simple app, built with the help of AI, now costs from $5,000. And a prototype takes a few days instead of a few months.
Frequently asked questions.
Can automation replace an employee?
Rarely entirely. It takes over tasks of a position, such as data entry, compiling data or follow-ups. Creativity, complex decisions, relationships with people and understanding context remain human.
Which tasks should I automate first?
Repetitive tasks that take a lot of time, create little value and have a result that's easy to check: compiling financial or production data, circulating information, answering simple customer questions, following up on overdue invoices.
How does AI help an SMB day to day?
With an assistant like Claude, ChatGPT, Copilot or Gemini, the team writes, summarizes and finds information faster. Connected to the company's tools, the assistant can also act: create a task, update a file, find an answer in your internal documents. Important decisions remain human.
What is an AI agent, and can it do an employee's job?
An AI agent handles a task from start to finish by working through the steps, for example following up on overdue invoices. It can take over entire tasks of a position, not the whole position. It follows clear rules, flags what's out of the ordinary and a person approves anything that commits the company.
Do you need a big AI project to get started?
No. Between the assistant your team already uses and AI connected throughout your operations, there is a middle ground: automating one specific process, with a person who approves. A simple app, built with the help of AI, costs from $5,000.
Before you post the next job, look at what that person would spend their days doing.
If a good part of their time would go to data entry, compiling or follow-ups, those tasks can probably be handed over to software, an assistant or an agent. The person you hire will then spend their time on what really requires a human.
Read next
An idea in mind?
Let’s chat!
:quality(80))
AI and automationCan you put confidential data into ChatGPT?
AI and automationHow do you find your company's knowledge with an AI assistant?
AI and automationWill AI replace my employees?