LuminariaMemoBETA
DOC roadmap/v1UPDATED 2026-04-29SCOPE MEMO + KIRBY + CALENDAR + FORMSPUBLIC
PUBLIC ROADMAP

What we're building.
When it ships.

What's shipping today, what's funded next, and where the Luminaria family is going. Updated as the roadmap evolves — previous versions live in Git history. Honest framing throughout: we don't list anything we don't intend to ship.

§ 01 · SHIPPING TODAY

What's live.
Verifiable today.

TRL 7-8 · IN PRODUCTION
CAPTURE

Cross-source session import

Claude Code (JSONL), Claude.ai (ZIP), Claude Desktop / claude.ai (MCP). 4 importers in production, deduplicated and merge-idempotent.

RETRIEVE

Smart Search (semantic + keyword)

Mistral embeddings on Neo4j vector index, full-text fallback, cross-session graph depth, freshness weighting. 12 MCP tools live in Claude.

COMPLIANCE

Tenant isolation enforced by CI

Every Cypher query starts with MATCH (t:Tenant). Automated lint blocks any non-isolated query from merging. 1,800+ tests across the codebase.

COMPLIANCE

Append-only activity log

CREATE-only writes on Activity Atoms. No UPDATE paths in code. Foundation for hash-chained tamper-evidence (funded buildout).

COMPLIANCE

EU residency, by construction

Compute (Clever Cloud Paris), database (Neo4j Aura europe-west1), AI inference (Mistral Paris), auth (Supabase Frankfurt). Not policy — by code.

BILLING

Stripe billing — Free / Pro / Business tiers

Live subscription billing with grace periods, plan downgrades, plan-tiered retention (14d Pro, 90d Business), webhook-driven enforcement.

§ 02 · COMPLIANCE ROADMAP

From Q3 2026
to certified.

Funded buildout — every item below is dependent on Stellari Studio's funding pipeline closing. Dates are best estimates and will shift as funding lands.

Q3 2026

Funding secured · scale-up begins

Sub-processor migration to fully-EU stack (payments → Mollie, email → EU-resident provider). ISO 27001 gap analysis kicks off. Public sub-processor register deployed at /sub-processors.

Q4 2026

Hash-chained audit log

Tamper-evident hash chain over append-only log. Public verification endpoint. AI Act Article 12 mapping document published.

Q1 2027

ISO 27001 Type I

External audit. Statement of Applicability published. SOC 2 Type I prep begins.

Q3 2027

SOC 2 Type I

Trust services criteria audit. First enterprise contracts with full DPA + SCC pack. Customer-managed encryption keys (BYOK) shipped for Enterprise tier.

Q4 2027

AI Act conformity (notified body)

Article 12 conformity assessment certified by notified body. Public pen test programme launches. Annual external pen test cadence established.

Q2 2028

ISO 27001 Type II + SOC 2 Type II

Continuous monitoring audit. Bug bounty programme public. 99.95% uptime SLA contractual. Cross-region DR with RPO < 15 min.

§ 03 · MEMO PRODUCT ROADMAP

What's shipping next
in Memo specifically.

Nine quarters of product work, from Q2 2026 through Q2 2028. Items tagged [EIC] depend on EIC Accelerator funding closing; items tagged [ELIAS] depend on ELIAS Open Call funding. Untagged items are self-funded from Memo revenue.

Q2 2026

Cross-provider · phase 1

  • ChatGPT export importer (ZIP + per-conversation)
  • Cursor session importer (SQLite log parsing)
  • Generic JSONL / Markdown bring-your-own importer
  • VS Code extension scaffold
Q3 2026

Audit-tier MVP + ELIAS pillars

  • Hash-chained activity log (tamper-evident) [EIC]
  • Regulator-ready export format (PDF + JSON) [EIC]
  • Personalization engine v1 — adaptive context [ELIAS]
  • Sustainability dashboard v1 — token reuse metrics [ELIAS]
  • Enterprise tier launch
Q4 2026

Cross-provider · phase 2 + teams

  • Ollama importer (local LLM sessions)
  • Continue.dev importer (VS Code AI plugin)
  • Aider importer (terminal AI coding tool)
  • Team workspaces with RBAC
  • Browser extension v1
  • First EU enterprise pilots
Q1 2027

Compliance UX + analytics

  • DPIA contribution generator
  • Records of Processing Activities (Art. 30) auto-export
  • Customer-facing audit log viewer
  • Personal analytics dashboard v1
  • Mobile PWA
Q2 2027

Workflow integrations + cost reduction

  • GitHub PR description generator from sessions
  • Slack search slash command
  • Notion sync (decisions → docs)
  • Linear / Jira ticket generation from session decisions
  • Weekly email digests
  • Embedding cache + dedup (skip redundant API calls)
  • Query result memoization (TTL-based)
  • Per-user cost dashboard (€-denominated savings)
  • Smart context trimming v2 (drop low-information tokens)
Q3 2027

Personalization v2 + Memo's own agents

  • Adaptive context distillation (auto-summary of user patterns)
  • Persona auto-detection (engineer / lawyer / writer / researcher)
  • Tag suggestion agent (auto-tag from content)
  • Knowledge gap agent (flag repeated questions → docs)
  • Compliance flag agent (DPIA-trigger detection)
  • Provider routing optimization (cost-aware LLM selection)
  • Batch vs streaming optimization (50% off for non-time-sensitive ops)
Q4 2027

Multi-agent platform + cross-provider phase 3

  • 🌟 Multi-agent shared workspace (CCode + Claude Desktop + ChatGPT all share one Memo session)
  • Agent action queue (agents propose actions for human review before execution)
  • Agent decision audit (every autonomous decision logged with reasoning)
  • Agent memory persistence across sessions
  • Memo-as-platform marketing push (MCP ecosystem positioning)
  • Gemini Code Assist importer
  • GitHub Copilot Workspace importer
  • Windsurf importer
  • Decision graph visualization
  • Cross-session concept clustering
  • Auto-generated knowledge base from sessions
Q1 2028

Agent governance v1 + platform

  • Agent permission system (per-tenant, per-agent capability scoping)
  • Agent reputation tracking (decisions × outcomes over time)
  • Action replay ("show me what this agent did last week, in order")
  • Rollback (undo a sequence of agent actions if something went wrong)
  • Agent action approval workflows for high-stakes operations
  • Custom MCP tools marketplace
  • Public API + webhooks expansion
  • Voice search (mobile + desktop)
  • Custom ontology layer for domain-specific knowledge
Q2 2028

Agent governance v2 + federation

  • Sandbox mode (agents propose, humans approve in batch)
  • Policy engine (declarative rules for agent behaviour per tenant)
  • Federated search across Luminaria instances (multi-account)
  • Cross-account agent governance for teams
  • Agent kill switch (federated emergency halt across all instances)
  • Compliance dashboard for agent activity (audit-grade)
§ 04 · THE LUMINARIA FAMILY

One account.
Four EU-sovereign apps.

Memo is the anchor. Kirby, Calendar, and Forms are bootstrapped from Memo revenue post-funding. All four share the same EU stack: single Neo4j graph, Mollie payments, Supabase auth, Vercel EU edge.

Memo

LIVE
Memory for every conversation you have with Claude.

AI session capture and retrieval for Claude Code, Desktop, claude.ai. 12 MCP tools live in Claude. Live at luminaria.so.

When
TRL 7-8, in production today

Kirby

PLANNED
Link-in-bio with native payments.

EU-sovereign alternative to Linktree / Beacons. Build a single landing page with multiple links, sell digital products natively, embed Forms and Calendar without third-party integrations.

When
After Memo Enterprise tier ships

Calendar

PLANNED
Booking, EU-sovereign.

Cal.com / Calendly alternative. Connect Google / iCal / Outlook. Accept paid bookings via Mollie. Embed in Kirby pages. JSON export anytime.

When
After Kirby ships

Forms

PLANNED
Forms that respect data residency.

Tally / Typeform alternative. EU-resident submissions, conditional logic, paid forms via Mollie, embed in Kirby pages. GDPR-compliant by default.

When
Parallel with Calendar
§ 05 · NOTES

How to read this roadmap.

Dates are estimates

Quarters are realistic targets, not contractual commitments. We update this page as the roadmap evolves; previous versions live in Git history.

Funded means dependent

Anything in the funded buildout requires the funding pipeline to close. Until then, those items are intent, not commitment. We won't claim them as shipping.

Scope discipline

We commit to what's listed and resist scope creep on principle. If something gets cut, this page gets updated and the cut is noted in Git history.

Public + version-controlled

This roadmap lives in our public repo. Every change is timestamped via Git commits — the commit history IS the change log.

Honest framing throughout

We don't list things we don't intend to ship. We don't backfill commitments after they slip. If something gets cut, this page gets updated and the cut is noted.

Funder relevance

EIC / ELIAS / EU funder evaluators: this is the same roadmap referenced in our applications. Compliance dates align with Step 2 budgets.

BUILD ALONGSIDE US

See what's shipping today.
Try Memo free.

The fastest way to understand the roadmap is to use the live product. The future is everything you don't yet have.

Try free at luminaria.soFor funders & enterprises →