Traditional software implementation teams are organised around specialist functions. Business analysts define requirements, solution architects design the system, configuration teams implement changes, testers verify functionality and project managers coordinate delivery.
Each function performs a necessary activity, but the organisational boundaries between them create additional work.
Requirements must be interpreted and transferred. Decisions lose context. Questions return to earlier stages. Different teams maintain separate records of the same implementation.
The customer, however, is interested in a single outcome: a working system that supports the intended business process.
AI creates an opportunity to reorganise delivery around that outcome.
Rather than assembling a separate team for every implementation activity, a smaller number of broadly capable delivery professionals could take responsibility for implementations from initial discovery through configuration, testing and handover.
AI would provide supporting capabilities that were previously distributed across specialist roles.
An implementation lead could use AI to analyse requirements, evaluate configuration options, prepare migration rules, generate test cases and maintain implementation documentation.
Specialists would still be needed for difficult technical decisions, security, complex integrations and areas requiring independent assurance. The distinction is that their expertise could be applied selectively, rather than every implementation moving through the same sequence of functional departments.
This suggests a shift from vertically organised functions towards a horizontal delivery capability: one accountable owner following an implementation across its full lifecycle, supported by shared expertise and specialised AI agents.
For this to work, context must remain consistent. Requirements, decisions, configurations and test evidence need to be accessible across activities without repeated interpretation or manual transfer.
Standardised products and implementation processes would make the model more practical. The greater the variation between customer deployments, the more specialist intervention would still be required.
There are also controls that should remain independent. Quality assurance, security approval and segregation of duties cannot simply be removed in pursuit of efficiency.
The objective is not to eliminate specialisation, but to reduce the organisational boundaries that make specialists necessary for routine work.
A horizontal delivery capability combines end-to-end accountability with access to specialist intelligence, rather than dividing accountability across a series of handovers.