The strongest parts are target-state design and deliverable production, with the standards library and engagement management close behind. The thinnest is the client-facing portal — it covers the essentials, but the flows around it are lighter than the rest of the product. Where a capability has known limits, the Security page goes further.
Reading the discovery material
eavantage extracts architecture elements — capabilities, processes, applications, data objects, and the relationships between them — from the material a client already has. It reads text, and it reads diagrams: a three-stage pipeline pulls structure out of process maps and architecture drawings, not just prose.
Supported formats: PowerPoint, PDF, Excel, Visio, BPMN, draw.io, and ArchiMate XML.
Extracted capabilities are anchored to the APQC Process Classification Framework, so the same business function lands in the same place across engagements. Every extracted element keeps a link back to the source passage it came from.
The model of record
Approved elements go into a single authoritative model — a graph of architecture elements and the relationships between them, built on a defined ontology so the structure stays consistent and checkable. One software component writes it; nothing else can. You can navigate the graph directly, and model business capabilities and processes on top of it.
Shaping it to your method
The ontology isn’t fixed to what eavantage ships out of the box. A Platform Administrator can disable or rename built-in entity types, relationships, and scoring dimensions to match your firm’s language, and define entirely new entity types and relationships — a defence client’s “Mission Thread,” an insurer’s “Regulatory Obligation” — with plain attributes. A custom type is written through the same single-writer, approval-gated pipeline as a shipped one, and candidates for it never auto-commit — even a high-confidence extraction floors at batch review, since the AI has no priors for a type it’s never seen. An Engagement Lead can overlay further for a single engagement without touching the firm-wide rulebook, and every engagement is pinned to its rulebook at creation, so a later firm-wide change never shifts a project mid-delivery.
On top of any of this, you can build your own views — a matrix, a catalog, a decomposition tree, a network map, a flow, or a timeline — over any slice of the model, including your own types.
Two things this doesn’t reach: the LeanIX, Ardoq, and Sparx exporters map from a fixed type list and skip custom types (only the ArchiMate export carries them), and gap analysis and recommendations run against the core types, not custom ones.
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 framework. Each gap is classified by severity. It also flags common architecture anti-patterns, estimates migration complexity, and produces a delta between the current and target processes.
AI readiness is the dimension the flagship domain pack is built around — assessing whether an organisation is ready to put AI agents into production, and what stands in the way.
Target-state design
For each gap, eavantage proposes a resolution and assembles a closure chain — a traced line from the gap through to the capabilities that resolve it. Recommendations are generated against the standards library and the engagement’s own model, and checked before they’re shown. The team works the current and target states side by side on one canvas; eavantage can draft target processes for people to refine. Every committed change is tracked against the client’s sign-off.
The output includes an implementation roadmap with a Gantt view, checked for circular dependencies. This is one of the two most developed parts of the product.
Deliverables and export
eavantage produces around twenty templated deliverables spanning every phase — assessment reports, capability heat maps, target architectures, roadmaps, handover packs — rendered as Word, PowerPoint, and PDF. Each is checked for completeness against its template before it goes out. This is the other most developed part of the product.
For onward work in an EA tool, eavantage exports the model one wayto ArchiMate 3 (validated against the schema), LeanIX, Ardoq, and Sparx. It does not import from those tools or write back to them — files a client provides are treated as discovery material, never as an authoritative source.
The standards library
Every deployment has a three-tier standards library — framework standards, reusable patterns, and the firm’s own playbooks — each entry moving through a draft → approved → deprecated lifecycle. eavantage parses standards from whatever format they arrive in, extracts the usable knowledge, and retrieves the relevant pieces when generating recommendations. At the end of an engagement, anything worth keeping can be promoted into the library, so each engagement makes the next one sharper.
Running the engagement
eavantage manages clients, engagements, teams, phases, and closure, with gate checks between phases (see How it works). Access is role-based, with segregation of duties built in — the person who drafts a deliverable is not the person who approves it. Permissions are enforced per architecture layer and per action, with a few operations still on a coarser check; the Security page has the detail. When an engagement closes, it can seed a follow-on.
The AI layer
Every AI feature runs through one internal gateway that can talk to Anthropic, OpenAI, or Azure OpenAI — or, if no provider key is set, fall back to a manual mode in which the AI features degrade gracefully rather than break. The gateway meters usage per call, versions its prompts, and applies pseudonymisation before content leaves for the provider.
An in-app chat assistant answers questions grounded in the engagement’s own model and standards, and every exchange is logged separately. Where the AI suggests a maturity score, a person still sets it — scores are never written automatically.
Working with the client
The client portal gives the client’s team a handful of screens — contribute documents, see a capability heat map, review deliverables and recommendations, and answer an AI-led interview. This is the least developed part of eavantage: it covers the essentials, but the clarification and workflow flows around it are thinner than the rest of the product. If a polished client-facing experience is central to how you’d run engagements, raise it in a demo.
Domain packs
A domain pack bundles the standards, deliverable templates, and playbook entries for a particular kind of engagement. Two are active by default: a general IT baseline, and the flagship AI-readiness pack, which packages what a readiness engagement needs so a team isn’t assembling the framework from scratch each time. Which packs are active is a deployment setting.
How it works walks the engagement end to end.Security covers data handling and where capabilities have limits.