AI, automation and the future of work

The problem it solves

Faced with artificial intelligence and automation, a small company tends to make one of two equally expensive mistakes: deciding it does not apply, or buying a tool before designing the process it was meant to bring order to. Both mistakes often grow from the same confusion: AI and automation are talked about as if they were one thing, when automation carries out rules somebody wrote and artificial intelligence produces plausible answers somebody has to check.

The second mistake is the costlier one, because it feels productive. There is an investment, an implementation and a go-live date. A few months later the team uses the system for what it already did and keeps a separate file for what actually matters. Data gets entered twice, handoffs between people still lose information, and every question from leadership is still answered by hand. Meanwhile, the idea of replacing people does its own damage: the ones who know the exceptions in the work stop explaining them.

Left alone, the same thing happens again with the next tool. The capacity that was supposed to be freed goes into maintaining two systems, the people who know the operation best stay tied to tasks that need none of their judgement, and the organization concludes that technology is not for it, when what it never had was a way of deciding.

What it covers

An executive talk on where the return on automation actually sits in a small or mid-sized organization: repetitive work that adds no value, data capture that today depends on memory, and the handoffs between people where information gets lost. The angle is deliberately about capacity rather than substitution. The useful question for whoever runs the business is not which roles might change in the abstract, but how much of their people's time goes on tasks that ask nothing of their judgement, and what the organization would do with that time if it came back.

It is explained with real processes and flows rather than product promises, and it rests on three distinctions. The first separates repetitive work from work that requires judgement. The second separates automation, which applies a written rule the same way every time, from artificial intelligence, which drafts, classifies or summarizes with a margin of error that calls for human review; confuse the two and you end up asking a rule for judgement or a draft for reliability. The third separates the process from the tool. And the talk names just as plainly what automation does not solve: a process that does not exist, a criterion nobody has written, or a decision nobody wants to make.

There is one short, deliberate stretch of individual work. Each participant notes the tasks in their own operation that eat time without calling for judgement, and the points where information changes hands, then checks that list against the signals that a process is not yet ready for a tool. It is brief, but enough for the criteria to be tested on real cases rather than borrowed examples.

The close is a prioritization criterion: what to automate first, what to leave alone, and which signals indicate the organization is not ready yet. Participants leave with a first ranked list of candidates from their own operation and an explicit rule for recognizing when buying nothing is the right call, which is the decision companies practice least. Operations and Technology are examined together in ALUD's work because of the very pattern this talk describes: a tool chosen without the process in view inherits every flaw the process has.

What participants learn

• Distinguish automatable work from work that requires judgement.
• Identify where information is lost in handoffs between people.
• Prioritize what to automate first and what to leave alone.
• Recognize the signals that a process is not ready for a tool.

Agenda

1. 01What automation is and is not in a small company
2. 02Repetitive work versus work requiring judgement
3. 03Capturing data at the moment the event occurs
4. 04Handoffs between people and information loss
5. 05Process first, tool second
6. 06Prioritization criteria and signals of not being ready

What the organization leaves with

• A prioritized list of automation candidates in your operation.
• An explicit criterion for deciding when NOT to buy a tool.