Provenance Compliance SaaS
Materiel DB
Your Art Collection, Protected Forever
The Challenge
Art provenance is one of the most complex problems in the commercial gallery world — and the tools haven't kept up.
Chapter 1 — The Problem
Why Provenance Is Broken
Decades of History
Galleries manage ownership chains, authentication records, and exhibition history spanning generations.
Legacy Tools Fail
Spreadsheets can't verify evidence. Traditional databases weren't designed for multi-decade reliability.
Trust Is Everything
Your reputation depends on unbreakable provenance chains and ironclad data security.
Chapter 2 — The Solution
A Database Built Like a Vault
Enterprise-grade architecture designed specifically for art world compliance and provenance tracking. Production-ready from day one.
Four Pillars of Protection
Decades-Scale Storage
Partitioned architecture ensures your 2050 data has a home today.
Multi-Tenant Security
Database-enforced isolation means your data stays yours — always.
Evidence Verification
Cryptographic hashing links proof to every provenance claim.
AI-Powered Insights
Automated extraction and smart linking reduce manual effort dramatically.
Chapter 3 — Fortress Security
Fortress-Grade Security
Enterprise SaaS security architecture that isolates every gallery at the database level — not the application layer.
Row-Level Security
Each gallery sees only their own artworks. Database-enforced isolation prevents any cross-tenant data leakage — no exceptions.
Unlike competitors who rely on application-layer filtering, our security is structural. Every query is automatically scoped to your tenant at the connection level.
Security Architecture in Depth
1
Composite Foreign Keys
All relationships include tenant ID, making cross-tenant references structurally impossible.
2
Session Context
Authentication sets tenant context at connection. Every query automatically filters to your gallery.
3
Database-Level Isolation
Security lives in PostgreSQL policies — not in application code that can be bypassed.
Chapter 4 — Built for Decades
Built for Decades
Art collections span generations. Your database should too.
Time-Series Architecture
Quarterly Partitioning
Audit tables automatically partition by quarter. Old data stays accessible, new data stays fast.
Composite Keys
Primary keys include partition columns. Queries hit only relevant quarters, not decades of history.
Auto-Maintenance
Future partitions create automatically. Your 2050 data already has a home.
Longevity by the Numbers
100+
Years of Capacity
Designed for true multi-generational storage
90%+
Query Optimization
Partition pruning eliminates irrelevant data scans
Zero
Data Loss Risk
Cryptographic hashing and immutable audit trails
Chapter 5 — Evidence-Based Truth
Evidence-Based Truth
Every provenance claim needs proof. We track both the claim and the evidence — down to the page and paragraph.
How Evidence Works
Digital Assets
PDFs, images, and certificates stored with content hashing. Verify file integrity at any point in time — forever.
Text Extraction
OCR and text hashing link specific document passages to provenance events. Span-level citations, not generic attachments.
Evidence Links
Polymorphic relationships connect evidence to artworks, transactions, exhibitions, and condition reports.
Chapter 6 — AI Intelligence
AI Intelligence
Let AI handle the tedious. You handle the art.
AI-Powered Workflow
Automatic Extraction
AI reads invoices, certificates, and correspondence to extract provenance events automatically.
Smart Linking
AI suggests relationships between documents, artworks, and constituents based on content similarity.
Confidence Scoring
Every AI extraction includes certainty levels. Review what needs review, trust what's solid.
Natural Language Queries
Query your collection in plain English. No SQL required.
Gap Detection
System flags missing dates, conflicting claims, and incomplete chains automatically.
Human Review
Every AI extraction goes to a review queue. You approve, reject, or edit before it's final.
Powered by Claude 4.5 Sonnet with structured output
Simple Power
Technical excellence you never think about. Gallery operations that just work.
PostgreSQL
Battle-tested reliability
Firebase
Real-time sync
GraphQL
Natural language queries

Composite keys ✓ Partitioned audit tables ✓ Row-level security ✓ Evidence tracking ✓ AI extraction ✓
Chapter 7 — Competitive Positioning
Competitive Positioning
Why our architectural moat is defensible — and why incumbents cannot easily replicate our evidence-first model.
The Architectural Moat
1
Schema Lock-In
Incumbent databases model provenance as text fields — not graphs with claims and evidence.
2
Breaking Changes
Adding an evidence layer requires schema rewrites that break thousands of existing integrations.
3
Product DNA Mismatch
They optimize for cataloging speed. We optimize for forensic defensibility.
4
AI Governance Gap
No framework for human-in-the-loop extraction or assertion lineage.
5
Multi-Tenancy Naivety
They use app-layer filtering. We use database-level isolation.
→ Competitive Window: 18–36 Months
The Schema Problem
Fundamental architectural divergence makes replication a 3-year project.
Incumbent Architecture
Artlogic, TMS
  • Provenance stored as a free text field
  • Evidence attached to artwork — not to specific claims
  • No structured ownership events, dates, or certainty levels
MaterielDB Architecture
Hardware Gallery
  • ProvenanceChain → ProvenanceEvent with certainty and supersession
  • EvidenceLink attaches to specific events with page/span citations
  • Multi-chain support for conflicting accounts

What incumbents must do: Migrate 3,000+ customers from text → events ($15–45M project)
Competitor #1: Artlogic
Leading Gallery CRM — 3,000 customers • $299–499/month • Strong in presentation and inventory
Why They Can't Pivot
  • 15-year technical debt — flat schema can't support a graph model without a full rewrite
  • Customer expectation trap — users expect simple, fast data entry
  • Small team (~30–40 employees) focused on feature parity, not R&D
  • Migration cost: $15–45M to move 3K customers
Our Advantage
  • Target compliance-driven galleries with high-value, export-heavy inventory
  • Migration tool ingests their CSV export seamlessly
  • "Artlogic stores provenance. We PROVE provenance."
Expected response: "Evidence Manager" add-on (generic tagging) by Month 12–24
Competitor #2: Gallery Systems (TMS)
Museum standard — MoMA, Getty, Met • $50–150K setup + $15–30K/year
Why They Can't Pivot
  • Wrong segment: Built for institutions with registrars, not commercial galleries
  • Wrong price: $50–150K setup = 10–50× our cost
  • Wrong speed: 12-month deployment vs. our 60 days
  • Legacy architecture: Linear provenance, SOAP APIs, on-prem bias
Our Advantage
  • 1/50th the price
  • 6× faster to value
  • Solves compliance museums don't face (AML, rapid export certification)
  • AI auto-extracts provenance vs. manual registrar entry
If they launch "TMS Lite": Captures top 5% (major dealers). We capture 80% mid-market.
The "Bolt-On" Trap
Why half-measures fail. Artlogic's most likely response — a generic "Evidence Tagging" feature — cannot compete.
Why Bolt-Ons Break Down
No Claim-Level Linking
Evidence attached to artwork, not to a specific ownership period.
No Multi-Chain Support
Can't model "seller says X, catalogue says Y" — forced to pick one.
No Supersession
Overwrites provenance when correcting errors. History lost.
No AI Governance
No extraction pipeline, review queue, or assertion lineage.
No Span-Level Citations
Can't cite "page 3, paragraph 2" as proof of a specific claim.
User Experience Delta
Three real-world scenarios illustrate why tagging isn't our product.
The Core Insight
Claim-level citations require the graph model — a 3-year rewrite for any incumbent. Tagging is a feature. Our provenance graph is infrastructure.
Chapter 8 — Switching Costs
Switching Cost Asymmetry
Migration friction creates natural lock-in — but only in one direction.
Migration: Easy In, Hard Out
FROM Incumbents → Us
Low Friction
  • Export CSV/XML from Artlogic/TMS
  • AI extracts structured provenance from text
  • Review queue for human approval
  • Timeline: 30–60 days
  • Cost: $0–5K (we absorb in pilot)
FROM Us → Incumbents
High Friction
  • Flatten events → text narrative (LOSSY)
  • Evidence links → generic attachments (lose claim context)
  • Multi-chain provenance → pick one, discard alternates
  • Supersession history → LOST ENTIRELY
  • Cost: $5–10K + audit re-validation risk
Our lock-in: Evidence vault + daily workflow + compliance dependency + network effects
Chapter 9 — Timing Window
Timing Window: 18–36 Months
Why the next 18 months are critical for category ownership.
Three Converging Forces
Regulatory Momentum
  • EU AMLD6 expands to art market (2024–25)
  • UK Economic Crime Bill in effect (2024)
  • US Treasury AML regs under consideration
Market Readiness
  • COVID forced gallery digitization (2020–23)
  • New generation expects software-first workflows
  • Replacement cycle: Excel → SaaS (happening now)
AI Window (2024–26)
  • Before 2023: Extraction quality too poor
  • 2023–24: GPT-4/Claude "good enough"
  • After 2025: Incumbents catch up on AI
Execution Trajectory
If we execute on schedule, the category is ours before incumbents can respond.
Chapter 10 — Competitive Summary
Competitive Summary Matrix
We're not "better Artlogic." We're a new category: Provenance Compliance SaaS.
Chapter 11 — Investor Q&A
Investor Soundbites
Prepared responses for the questions you'll hear in every meeting.
Anticipating Tough Questions
Won't Artlogic just add this?
"Our moat is the provenance graph with supersession — a schema rewrite that would break 3,000 customers. By the time they commit, we'll own the category."
Isn't TMS already solving this?
"TMS is a Rolls Royce for institutions. We're a Tesla for commercial galleries. Different segment, different speed, different job."
What if TMS launches TMS Lite?
"Their DNA is 12-month enterprise sales. We're 60-day product-led. 10× cheaper, 6× faster. Mid-market won't wait."
More Q&A Responses
Won't AI extraction commoditize?
"The moat isn't the LLM — it's the governed pipeline. Review queue, rules registry, assertion lineage. That's 18 months of workflow engineering."
What about PE-backed competition?
"If Artlogic gets PE for a rewrite, that's 36 months. We'll have 700 customers by then. We're the acquihire target."
The Moat Is Architectural
What looks like "features" is deep infrastructure divergence.
"We link evidence to claims"
→ Non-linear provenance graph with chain identity
"We extract with AI"
→ Governed pipeline: IR → LLM → resolution → review queue
"We track corrections"
→ Supersession without overwrite + assertion lineage
"We cite sources precisely"
→ Span-level citations with stable textHash verification
Incumbents see "evidence linking" as a feature. We know it's a 3-year schema rewrite. That asymmetry is the moat.
Let's Build the Moat Together
The competitive window is 18–36 months. The architecture is ready. The market is turning. The time is now.
MaterielDB • Confidential