AI is making it easier to see what is happening inside a business.
It can summarise thousands of customer conversations, spot a drop in conversion, compare team performance, identify a new pattern and recommend what to do next.
That should help leaders make better decisions.
But it can also create a new kind of instability.
Monday's dashboard says customers want more choice, so the offer expands. Wednesday's analysis says too much choice is hurting conversion, so the offer contracts. Friday's sales summary says a large prospect wants an exception, so the process changes again.
Each decision sounds reasonable in isolation. Each has data behind it. Each may even produce a small improvement.
But nobody can explain the operating logic that connects them.
The organisation is not learning. It is reacting.
I call this vibe management.
What is vibe management?
Vibe management is allowing AI recommendations and short-term signals to keep changing how the organisation works without deliberate human ownership of the decisions or their consequences.
It is the management version of accepting the first plausible answer from a chatbot.
The leader is still technically making the decision. They may click approve, announce the change or ask the team to act. But the reasoning has been outsourced.
Ask why the organisation changed and the answer is not a clear strategic choice. It is some version of:
- the data suggested it;
- the AI found a pattern;
- this customer asked for it;
- the last campaign performed better; or
- it felt like the right response at the time.
Those can all be useful inputs. None of them is ownership.
AI can recommend a change. It cannot take responsibility for what the organisation becomes.
More data does not automatically create more direction
There is an appealing story about AI-assisted management.
The organisation captures more data. AI turns it into insight. Leaders follow the insight. The feedback loop gets faster, so the business improves continuously.
Sometimes that is exactly what happens.
But a signal only tells you what happened under a particular set of conditions. It does not decide what the business values, which trade-offs it is willing to make or whether a short-term gain supports the longer-term strategy.
Imagine an AI analysis finds that proposals with larger discounts close more often.
That may be true. It does not tell you whether the organisation should discount more.
Perhaps those projects are less profitable. Perhaps they create more delivery stress. Perhaps the discount attracts customers who are harder to retain. Perhaps the business has deliberately chosen to protect margin while building a premium position.
The conversion result is evidence. The commercial decision still belongs to someone.
When those two things are confused, a business can become highly responsive and poorly directed at the same time.
The people can fall behind the machine
An AI agent can adapt to a changed instruction immediately.
People cannot.
They need to understand what changed, why it changed and what the decision means for their work. Other teams may have commitments based on the previous direction. Systems, templates, measures and customer expectations may all depend on it.
If the organisation changes constantly, people stop trying to internalise the rules.
They wait for the latest instruction. They check before acting. They keep local workarounds because they do not trust the official direction to last. Experienced people spend more time interpreting change and less time improving the work.
The business may be perfectly machine-readable while becoming impossible for its own people to understand.
Agents can keep up with a specification that changes a thousand times a day. The people responsible for the outcome cannot.
A spec-driven organisation is not one where the specification changes whenever AI finds something interesting.
It is one where people remain meaningfully connected to what the organisation has decided and the outcomes those decisions produce.
Every change carries a tax
Leaders often experience a change as a single decision.
Change the approval threshold. Adjust the pricing rule. Add a new customer segment. Rewrite the handoff. Ask the agent to use a different measure.
Downstream, that one decision becomes many pieces of work.
Someone has to:
- understand what the new direction means;
- identify which teams and systems depend on the old one;
- update instructions, agents, templates and measures;
- communicate the change;
- handle work already in progress;
- resolve contradictions and edge cases; and
- rebuild confidence that the new decision will last.
That is the change tax.
AI can reduce it. An agent can trace dependencies, update supporting material, prepare communications and identify affected work.
It cannot eliminate the human cost of changing what other people rely on.
The larger and more connected the organisation becomes, the more expensive an impulsive decision can be.
The ability to change faster does not remove the need to change deliberately.
Stability is not the enemy of learning
The alternative to vibe management is not ignoring evidence or freezing the business in place.
It is separating learning from commitment.
An agent might notice that a particular proposal structure appears to perform better. That is a signal worth examining.
The team might choose to test that structure on suitable opportunities. That is an experiment.
Only after considering the evidence, customer context, delivery consequences and commercial strategy might the owner decide to make it the standard approach. That is a change to how the organisation operates.
Those events should not collapse into one automatic action.
A mature business can move quickly around a stable direction. It can collect signals continuously and run experiments without rewriting its operating model every afternoon.
The specification gives that learning somewhere to land when the organisation is ready to commit to it.
A living specification is a strategic memory
A living specification records the current organisational commitment for a product, service or repeated piece of work.
It makes clear:
- what outcome matters;
- which goals are being pursued;
- which trade-offs are acceptable;
- which boundaries must continue to hold;
- who owns the result;
- what remains undecided; and
- which changes have deliberately been adopted.
It does not replace dashboards, customer records or experimental data.
Those sources tell the organisation what is happening. The specification records what the organisation has decided to do about it.
This distinction protects the business from having its strategy quietly rewritten by whichever signal arrived most recently.
It also protects people from being asked to follow rules nobody can explain.
How to tell whether you are vibe-managing
Look at an important process that has changed several times in the last six months.
Then ask five questions.
1. Can anyone explain why the current approach exists?
Not only what the latest rule says, but the goal, evidence and trade-off behind it.
If the explanation is “the system recommended it,” the decision has no real owner.
2. Do changes have a named decision-maker?
A committee, AI tool or general sense of consensus is not the same as an accountable person.
Someone should be able to say: I considered the evidence, I made this call and I own the outcome.
3. Can you distinguish a test from a new rule?
If every experiment immediately becomes the process, the organisation will accumulate half-proven practices and contradictory instructions.
People should know when they are testing something and when the business has committed to it.
4. Does the organisation understand what else the change affects?
A pricing decision may affect delivery. A sales target may affect customer quality. A faster approval may move risk into finance or legal.
A local improvement is not an organisational improvement if another team silently absorbs the cost.
5. Could your people describe the current direction without consulting the AI?
The whole specification does not need to be memorised word for word.
But the people who own the outcome should understand its essential logic. If the machine is the only participant capable of explaining how the business now operates, human ownership has already been lost.
A simple discipline before changing the business
When AI recommends a meaningful change, require a short decision record:
- What did we observe? Separate the evidence from the interpretation.
- What do we believe it means? Include uncertainty and alternative explanations.
- What are we changing? Be explicit about the organisational commitment.
- What are we not changing? Protect the surrounding strategy from accidental drift.
- Who owns the decision and outcome? Name the accountable person.
- What else is affected? Identify the people, agents, systems and work that depend on it.
- When will we review it? A deliberate decision can still be revisited.
This does not need to become a seven-page form.
The point is to slow the decision down just enough that the organisation can move quickly afterwards.
AI can prepare the evidence and map the consequences. The human owner makes the commitment.
How BusyWork Dispatch fits
BusyWork Dispatch separates the durable truth about the business from the work and signals moving around it.
The Workbook is where the important, owned context belongs. Dispatch is where current work moves. Evidence can arrive from customer conversations, systems and completed work without automatically becoming a new organisational rule.
When something should change, the learning can be brought back to the person who owns that part of the business and deliberately reflected in the Workbook.
That is the practical defence against vibe management: AI helps the organisation see and act, while people remain responsible for what it decides to become.
Bottom line
Vibe management is what happens when AI recommendations and short-term signals keep changing the organisation without deliberate human ownership. The business becomes faster at reacting but less able to explain its direction, absorb change or remain accountable for the result.
Use AI to surface evidence, prepare options and reveal consequences.
Then ask a person to make the decision, own the trade-off and update the durable specification when the organisation has genuinely learned something.
The aim is not to move slowly.
It is to create enough strategic stability that many people and agents can move quickly in the same direction.
BusyWork Dispatch helps organisations connect current work to the strategic context people own, so AI can support faster action without quietly rewriting the business. See how BusyWork works or book a call with Ben.
