Every element of the architecture is a proposal until a person with the authority to accept it does. This page walks the engagement end to end: the loop at the centre of it, the six phases it moves through, the gates between them, and how the work is closed.
The loop at the centre
Everything in eavantage comes back to one cycle.
Discovery material in. A consultant uploads what the client has — strategy documents, process maps, application inventories, spreadsheets, interview notes, and diagrams in Visio, BPMN, draw.io, or ArchiMate.
The AI proposes candidates. It extracts the capabilities, processes, applications, data objects, and the relationships between them. Business capabilities are anchored to the APQC Process Classification Framework, so the structure stays consistent from one engagement to the next. Every candidate carries a link back to the exact place in the source it came from.
Candidates are reviewed before they count. Each one is risk-tiered. High-impact elements — gaps, recommendations, objectives — and anything a reviewer flags or that sources disagree on go to individual review. Mid-confidence extractions go to a batch review queue. Only high-confidence inventory facts that three or more independent structured sources already agree on are committed automatically, and those are logged, flagged to the engagement lead, and can be rolled back. The AI never decides which findings matter.
Approved elements become authoritative. One software component writes the model of record — nothing else can, on any code path. Combined with per-client data isolation and a complete audit trail, that single-writer rule is what lets you stand behind the result.
The loop repeats as more material comes in and the picture fills out.
An engagement, phase by phase
An engagement moves through six phases. Each produces something concrete and closes with a gate.
1 · Platform setup. Before client work starts, the platform is configured — AI quality thresholds, the standards library, deliverable templates — and the engagement team is set up with roles.
2 · Client discovery. The problem is framed, an artifact-request checklist goes to the client, material comes in, the AI extracts candidates, the team reviews them, and the capability and process models take shape. This is where the loop above runs hardest.
3 · Gap and readiness assessment. eavantage assesses the current state across six dimensions — capability coverage, process maturity, data quality, team agility, AI readiness, and alignment to the client’s chosen framework. Each gap is classified by severity; the output is a readiness score, a migration-complexity read, and a process delta.
4 · Target-state design. For every gap, eavantage proposes a resolution and builds a closure chain (below). The team works the current and target states side by side on one canvas, drafts the target architecture, and records the client’s commitment to each change.
5 · Implementation handover. The design becomes deliverables an implementation team can act on — configuration specs, an implementation roadmap aligned to the client’s delivery framework, and one-way exports to EA tools (ArchiMate 3, LeanIX, Ardoq, Sparx).
6 · Governance and closure. After go-live, the design is validated against what was actually built, the engagement is signed off by both sides, anything worth keeping is promoted to the standards library, and the record is sealed.
Gates
Phases don’t advance because someone clicks “next”. Each transition has a gate — checks that run on the server and have to pass.
The gate that defines the product is the one on critical gaps. A gap classified as critical can’t be marked resolved until there’s a complete, traced line behind it:
gap → recommendation → target process → data objects → application → capability
If any link is missing, the gap stays open and the phase doesn’t close. You can’t declare a transformation done on the strength of a slide.
Current state and target state, together
Target-state design happens on a dual canvas — the current architecture and the proposed one, visible side by side, so every proposed change is anchored to something real. eavantage can draft target processes for the team to refine, and candidate recommendations are checked against the model and the standards library before they’re shown — but the direction is set by people. Each committed change is recorded against the client’s sign-off, so at closure there’s a clear line from “here’s what we found” to “here’s what we agreed to”.
Closure
Closing an engagement takes a dual sign-off — the client’s EA lead and the client’s executive sponsor both confirm, and eavantage checks each person’s role before accepting it. On closure, deliverables and templates are locked and the audit trail is sealed. Throughout the engagement, an on-demand graph integrity check is available — it flags orphaned, duplicated, or mis-nested elements so the model can be confirmed internally consistent before sign-off. Standards and playbook entries proven out during the engagement can be promoted into the library for the next one.
Two kinds of engagement
eavantage supports two scenarios as first-class:
- Migration — the client is moving off an existing tool or platform, and the job is to carry the architecture across cleanly.
- Greenfield — a fresh implementation, with the architecture built up from discovery.
Either way, the roadmap aligns to how the client delivers — SAFe, Scrum, Kanban, LeSS, or their own model.
Who does what
Roles carry named permissions, and the work is split so that no one person both drafts a deliverable and approves it. Each role’s permissions are a starting template an engagement lead can tighten per engagement.
- The engagement lead owns the engagement, approves the deliverables and the closure, manages the team, and moves a phase forward once its gate is met.
- Consultants do the delivery work — uploading material, curating the model, scoring maturity, drafting recommendations, reviewing candidates — but can’t approve deliverables or manage the team.
- The downstream implementation team gets a read-only view of the finished design and the artifacts to build from — through a dedicated role where a domain pack provides one, or a scoped account otherwise.
- On the client side, the EA lead contributes material and records commitments; the executive sponsor provides the second half of the closure sign-off.
See what each part does in more depth on theCapabilities page, how data is handled on theSecurity page, and what it takes to run on theDeployment page.