ATLAS, AI NATIVE EA PLATFORM
Strategy to architecture, evidence backed, MCP connected.
Atlas connects strategy, capabilities, applications and decisions in one data model. AI does the groundwork, you qualify. Two way integration to Miro, GitHub and more via MCP. Bring your own LLM.
12 CAPABILITIES
One platform. Many perspectives.
Strategy Parser
Upload PowerPoint or Word, AI structures as Playing to Win hierarchy with themes, objectives and initiatives.
Application Portfolio
CSV import or MCP sync. APP-NNN keys per workspace. AI classification on TIME, lifecycle and criticality.
Capability Map
AI suggests L1/L2/L3 based on industry. Wardley scoreable. Editable by EA team.
Technologies
Slug based reference to known tech (React, PostgreSQL, Snowflake). Linked to apps and dependencies.
Decisions (ADR)
Structured decisions with context, options, decision, consequences and evidence binding.
Alignment Detection
AI shows where portfolio supports strategy and where it blocks. Gap analysis in one view.
ArchiMate Full
6 viewpoints, 7 layers, AI generated diagrams. Complete ArchiMate implementation.
Wardley Mapping
AI scores capabilities on evolution axis (Genesis, Custom, Product, Commodity). Strategic moves panel.
Custom Workspace Lenses
Scale+ tier. Define your own lenses with custom DSL and workspace specific AI prompt.
MCP Server
Atlas exposes EA data as MCP server to Claude, Cursor, Miro Sidekick and future clients.
Multi LLM Platform
BYOL via OMICRON. Per feature model picker (Anthropic, OpenAI, Google, Azure). A/B test framework.
GitHub Repo Health
Daily sync 04:00 UTC. Health badges on applications. Orphan detection with digest email to admin.
USE CASES
Three scenarios we solve
Application rationalisation
Problem
Map 200 apps, identify overlap, generate retire/keep/evolve plan based on strategy.
Atlas
Upload portfolio as CSV, AI enriches with lifecycle and cost, Atlas generates TIME assessment and alignment gap per app.
Output
Executive summary with rationalisation recommendations ready for the board.
Post-M&A integration
Problem
Two portfolios merging. What's kept, what's retired, what is the shared target architecture?
Atlas
Atlas maps both portfolios, identifies overlap and generates target architecture across both.
Output
Target architecture report with consolidation roadmap and decision log.
AI readiness assessment
Problem
Where are the quick wins? Which apps have the data, ownership and regulatory maturity?
Atlas
Atlas scores apps against AI-readiness criteria: data quality, governance, compliance maturity.
Output
AI readiness score per app with prioritised shortlist for pilot projects.
TECHNICAL
Built on modern stack
Atlas runs on Next.js 16, Drizzle ORM and Neon serverless PostgreSQL. Hosted on Vercel. Authentication via NextAuth. AI pipeline via Anthropic API with streaming. No proprietary integration platforms. No agent overhead.
SECURITY AND GOVERNANCE
Built for sensitive architecture data.
EU hosting
Data hosted in Frankfurt (eu-central-1) via Vercel and Neon. No US sub-processors on the hot path.
Workspace isolation
Workspace scoped access plus role based permissions. Core data objects tied to workspace.
Bring your own model
Live for Scale+ customers. Your own keys, your own model choices via the OMICRON platform. Anthropic, OpenAI, Google, Azure.
Proposal first
External tools do not overwrite canonical Atlas data without human review. Changes flow through change proposal queue.
AI with evidence
Every critical AI output has model version, prompt version, confidence score, source evidence and review state.
Audit trail
Every decision and change traceable for 12 months. MCP key usage, external syncs and data exports logged.
OUR PRINCIPLE
“AI should not just answer. It should show its work. We do not commit AI output without model version, confidence score and citation to source.”
Founder, Spekir