Asynchronous Intelligence

Imagine asking a capable analyst to review an entire property portfolio, then pulling up a chair beside her and waiting. Each time she opens a report, compares two figures, or checks a calculation, you remain there. When she needs a document, you hand it to her. When she finishes one step, you tell her to begin the next.

No manager would deliberately organise work this way, yet conversations with AI can easily follow this pattern.

With a chatbot, you exchange messages with an AI assistant: you ask, it answers, and you guide the next step. Even when it does part of the work, you remain responsible for keeping it moving.

My first article looked at how intelligence is organised. This second one is about the relation between time and intelligence: when you delegate an assignment, how much of your time does it still occupy? Asynchronous intelligence means the work carries on while you turn your attention elsewhere.

Chatbot conversations turn assignments into meetings

Meetings are useful when a decision needs immediate discussion. Other work is better delegated: a manager explains what is needed, agrees on the boundaries, and returns when the result is ready.

Working with chatbots has trained us to turn both kinds of work into meetings. We ask for a first step, wait, provide the next instruction, and keep the window open. If the system needs another document or reaches an ambiguity, we become the mechanism that moves the process forward.

For a definition, a summary, or a quick calculation, this rhythm is natural. It becomes awkward when the answer depends on dozens of documents, several calculations, and an independent check. The user may have handed over the task, but not recovered her attention.

Here, asynchronous has a simple meaning: the assignment continues while the person who requested it does something else.

There are therefore two clocks: the time the assignment takes, and the time the person must remain involved. AI agents can work around the clock, carrying long assignments through evenings and weekends. Work that would stretch across several office days can finish sooner, while deeper investigations can continue during your absence.

Combine this with clear ownership and control, and one person can oversee several of these processes in parallel. That is the multiplier: days of background work across multiple assignments, with visibility over progress and the ability to intervene. Human involvement remains at the points your profession requires, whether to validate an assumption, approve a commitment, or make the final decision.

Léa delegates a week of analysis

Return to Léa, the portfolio analyst from the first article. This Monday, she needs a thorough portfolio review for Thursday's investment committee: what changed across the buildings, what deserves attention, and why? The answer requires gathering reports, comparing figures, tracing unusual movements, and checking the result, with some documents arriving during the week.

In a conversation with a chatbot, Léa asks for the latest figures, waits, uploads a missing report, asks for a comparison with last week, and then requests the source behind an unusual vacancy movement. She is doing less of the analysis, yet still managing every hand-off.

With delegation, Léa asks: “Review the portfolio for Thursday's committee, identify important changes since last week, and prepare the evidence.” The system confirms what it understood, what it will use, and which decisions remain with her.

Then Léa leaves for a tenant negotiation. Over the following days, the agents gather reports, compare buildings, and check the findings. AI agents can continue overnight, bringing in new documents and revisiting discrepancies; human colleagues contribute where their expertise or approval is required.

Léa receives a notification only when there is a reason to return. A source may be missing, an assumption may require approval, or the review may be complete. The system comes back with a specific request or result rather than asking her to discover where the work stopped.

Delegation must remain visible

Work that disappears into the background can feel less trustworthy than work we can watch. The answer is not a stream of artificial activity. It is a clear record of what the system accepted and what happened next.

Here, a parcel delivery offers a useful analogy. We are comfortable letting something valuable travel out of sight because we receive a receipt, a destination, a status, and notice when the route changes. We do not need to watch the van; we need to know that the parcel is under control.

A delegated assignment needs the same properties. At the beginning, the system should show what it understood. During the work, it should show meaningful stages rather than a theatrical percentage, while leaving Léa free to pause, redirect, or cancel the assignment. If something fails, it should say what remains usable and what needs attention. At the end, it should connect the result to the evidence behind it.

It must also preserve completed work. If one property report is unavailable, the other comparisons should not vanish. The system can mark the gap, retry that source, and ask Léa only if the missing information becomes consequential.

Visibility and recovery are not decorations around the AI. Without them, trust is impossible to build, and then asynchronous workflows merely replace a long wait with uncertainty.

Léa returns for judgment

Delegation does not mean removing Léa from the process. It means being more deliberate about where she returns.

She is needed at the beginning to define what matters, perhaps in the middle to approve an assumption, and at the end to decide what the result means. Between those moments, collecting numbers, preparing comparisons, and checking consistency do not need to occupy her attention.

Léa remains responsible for the conclusion, using the chat window to give direction and discuss the findings. Once she closes it, several assignments can continue for days, with her expertise guiding the work and her judgment called upon at the points she has defined.