FABCODE

Company Building

The work between the tools

A client sends a small request. Someone copies it into a task, asks for missing context in chat and forwards the answer to a specialist. The work may take an hour. Establishing what the work actually is can take much longer.

By FABCODE2 min read

Dark concrete corridor lit by a single shaft of warm light

Follow one request all the way through

This is an illustrative agency scenario, but it exposes a useful unit of analysis. Instead of counting applications or meetings, follow a request from arrival to acceptance. At each step, ask who owns it, what information is available and which decision allows it to move forward.

A project can appear fully digitised while depending on people to reconstruct its state. The brief sits in a document. The approved scope is buried in a conversation. The client sees one status and the delivery team uses another. A new tool can make one of those places better while adding another handoff between them.

McKinsey's October 2026 analysis of AI and organisational design argues for organising work around outcomes and reducing unnecessary handoffs. That is a useful external reference, not evidence that any particular platform will deliver a return. For an agency, the practical question is whether the operating process becomes easier to understand and manage.

Agree what the business needs to remember

An operational backbone should retain the information that would otherwise have to be requested again. That includes the original request, agreed scope, responsible person, relevant assets, approval decision and evidence of delivery. It does not require every specialist to use the same design or development tool.

The distinction matters. Replacing every application is expensive and often unnecessary. Establishing a consistent record of the work can connect specialist tools while leaving expertise where it belongs. An integration is useful when it preserves meaning across that boundary, rather than simply copying another notification.

AI can help classify a request, assemble context or prepare a first draft. Whether the result meets the brief still needs a decision. A workflow should make that decision visible, including who made it and which version was approved. Otherwise faster drafting can simply increase the amount of unresolved work waiting for review.

Measure the waiting as well as the working

A team considering a new operating platform should begin with one recurring service. Record how long work waits for clarification, how often a brief is reopened and how much effort goes into communicating status. Those observations provide a baseline for an improvement that can actually be checked.

Standardisation has a limit. A tightly specified maintenance request and an exploratory brand project need different degrees of structure. Forcing both through an identical checklist can replace uncertainty with bureaucracy. The backbone should standardise ownership and evidence while allowing the work itself to differ.

The investment case rests on continuity. As a business adds clients, specialists and automation, it should retain a reliable account of what was requested, decided and delivered. Software that protects that continuity can become more valuable as the organisation becomes more complex.

Sources

Autonomous Business

Autonomy begins with accountability

A useful AI agent needs more than capability. It needs a defined mandate, a way to prove its work and a clear point at which to stop.