From visibility to action
End-to-end process maps provided a shared view of delivery, but further detail was needed to identify where AI could assist and how individual activities should operate.
As Bound and Omnia explored AI-enabled consulting delivery, I helped translate end-to-end process knowledge into opportunities for AI support. I then led the design of four EXD workflows, reusable subagent capabilities and cross-disciplinary coordination rules, establishing foundations for a potential agentic delivery ecosystem.
A proposed agentic delivery framework comprising four EXD workflows, 20 subagent capabilities and seven cross-disciplinary collaboration use cases. Elements of the work informed a subsequent evidence-ledger experiment and were incorporated into Omnia's internal documentation.
I synthesised two delivery-process sources into end-to-end views and, building on architectural guidance from the Tech team, defined an operational workflow model. I led the design of four EXD workflows, identified 20 proposed subagent capabilities across them and developed seven cross-disciplinary interaction use cases.
Bound and Omnia were exploring how AI could support consulting delivery across the project lifecycle. Existing process documentation described activities and opportunities, but turning that knowledge into an agentic model required more explicit operating logic, reusable capabilities and coordination between disciplines.
End-to-end process maps provided a shared view of delivery, but further detail was needed to identify where AI could assist and how individual activities should operate.
Design practice draws on professional knowledge and contextual judgement that must be made explicit before tasks can be delegated to AI agents.
An initial hypothesis of ten role-agents needed to be examined against actual workflow activities to identify reusable capabilities rather than mirror existing job titles.
Research, Service Design, UX and UI share dependencies. Agents needed rules for exchanging evidence, escalating issues and retaining human decision authority.
I synthesised two existing sources into a top-level operating model and a detailed current/future-state workflow view, making delivery activities, opportunities and potential AI interventions easier to examine.
Building on Tech's architectural guidance, I defined an operable workflow model covering purpose, inputs, tasks, outputs, validation, human checkpoints and progression rules.
I mapped Research, Service Design, UX and UI activities using professional knowledge and established design practices, specifying how each workflow could operate with AI assistance.
I developed a cross-comparative analysis of the workflows to identify 20 specialised subagents, separating recurring capabilities from the phases and disciplines in which they are used.
I identified seven use cases where work in one discipline requires evidence, decisions or intervention from another, informing proposed handoff and escalation rules.
I consolidated the workflows, subagent inventory and collaboration use cases as specifications that could inform a future agentic harness and EXD–Tech integration.
Documented workflows for Research, Service Design, UX and UI, alongside a cross-workflow inventory of 20 specialised subagent capabilities. These specifications established a design-led foundation for potential agentic implementation.
Two team members used the conceptual model developed for the Research workflow to create an evidence ledger, providing an example of subsequent use of the proposed approach.
The identified interaction scenarios were subsequently documented by a Tech colleague as part of Omnia's agentic delivery work, making the proposed EXD dependencies available for further technical consideration.
The work contributed reusable design knowledge and informed early experimentation. The broader agentic ecosystem was not implemented or evaluated as part of my involvement.