Proof of Data Moat · Live · Reads from production

Show me the data.

The single URL we send when an institution asks "prove you actually have data." Every count below is read live from the production Supabase. Every identifier is verifiable on a third-party registry. Every cron run and every memo carries a timestamp that can't be backdated.

Category

Avena belongs in the institutional data infrastructure category alongside Bloomberg (equities, fixed income), MSCI (indices), Refinitiv (workflow + data), S&P Global (credit, indices), Moody's Analytics (risk, structured finance), and CoStar (US commercial real estate). The European residential property equivalent did not exist before Avena. The infrastructure below makes the claim verifiable.

Refreshed every 5 minutes·Cite DOI 10.5281/zenodo.19520064·Zenodo permanent archive
117,252
Total rows · all ledgers
across 16 compounding tables
1,881
Spain corpus · scored
from registry · live count
88
Rows added · last 24h
real-time velocity
1,407
Rows added · last 30d
compounding rate
48d ago
Moat depth · earliest record
since 2026-04-08
§02 · Live row counts · per compounding ledger

What the database actually contains, right now.

Each row below is a SELECT COUNT(*) query executed in the last 5 minutes against the production Supabase. The earliest timestamp shows how far back the moat goes per ledger. The "added 30d" column shows the compounding rate — if it's positive, the moat is widening daily.

LedgerCategoryRows+24h+7d+30dSpanLatest
avena_historyIndex history921699031d20h ago
price_snapshotsPrice history1,881000<1d48d ago
sold_propertiesTransactions0000
memo_generationsMemos5055<1d3d ago
avm_queriesValuations0000
precursor_signalsIntelligence11111115d14h ago
genesis_scenariosIntelligence1101111<1d5d ago
genesis_outputsIntelligence7507575<1d5d ago
counterpart_developersCounterpart1001010<1d5d ago
counterpart_stress_alertsCounterpart39000
cron_logsOperations1,044412731,00133d2h ago
macro_indicatorsMacro311631313d14h ago
causal_indicatorsMacro2002020<1d3d ago
regime_historyMacro672416432d14h ago
findingsFindings113,877000
feed_sync_logEU ingestion892789893d17h ago
Total · all 16 ledgers117,252886351,407moat compounding

16 ledgers shown · 105+ tables total · Plus the Spain corpus (1,881 scored properties in properties_registry) · See /stack §04 for the full schema map.

§03 · Operational audit · last 10 cron runs

The engine ran ten times before you opened this page.

Sampled directly from cron_logs. Each row is one scheduled execution — agent, status, exact start timestamp, and wall-clock duration. The duration alone proves the work happened: you can't fake a 4,231ms Postgres write.

AgentStatusStartedDuration
eu-ingestionsuccess2026-05-26 18:34:49 UTC1,102ms
property-augmentsuccess2026-05-26 17:29:12 UTC202,820ms
eu-rescoresuccess2026-05-26 17:05:28 UTC1,573ms
pricing-historysuccess2026-05-26 15:01:57 UTC2,935ms
prometheussuccess2026-05-26 14:48:11 UTC2,115ms
eu-rescoresuccess2026-05-26 13:57:59 UTC1,130ms
property-augmentsuccess2026-05-26 13:57:40 UTC209,294ms
eu-ingestionsuccess2026-05-26 12:33:41 UTC1,125ms
couriersuccess2026-05-26 09:30:28 UTC1,152ms
property-augmentsuccess2026-05-26 09:04:16 UTC186,667ms
§05 · Why this compounds

Every day the gap to a competitor widens.

Price snapshots

4× daily price capture per property × 1,881 Spain properties × every day Avena runs = ~5,500 immutable snapshots/day. A competitor starting today can never recreate yesterday's prices.

Sold detection

Properties that disappear from a feed are persisted to sold_properties with their last seen price. This is a one-way ratchet — it compounds forever.

Macro indicators

Daily ECB SDW + Eurostat snapshots, archived with their fetched_at timestamp. Even ECB cannot recreate the exact sequence of revisions over time.

Prediction ledger

Every prediction is written before any outcome is known. Verified hit rate accumulates monotonically. A new market entrant starts at zero credibility.

Memo + AVM corpus

Every user-generated memo and valuation contributes to the training corpus + reveals demand patterns. This is the proprietary intelligence layer.

Findings ledger

Every signal the swarm detects writes a row. The findings table is append-only and compounds with every cron tick. Visible at /eu-takeover.

Federated partners

Each new partner brings country-specific data Avena did not previously have. Network effect — partners attract partners.

AI citation network

Daily monitoring of when Avena is cited by Perplexity, Brave, You.com, Claude, GPT. As citations grow, Avena becomes the source AI models reach for first.

§06 · For procurement teams

What to ask for in due diligence.

If you're evaluating Avena Terminal for procurement, here's the checklist the page above answers:

  • Does the data exist? The §02 ledger is a live SELECT COUNT(*) against the production database. You can verify any number by hitting /api/v1/... directly.
  • Does the data accumulate? The +24h / +7d / +30d columns show net new rows per ledger over the rolling window. If these are positive, the moat is widening.
  • How far back does it go? The "Span" column shows time elapsed from earliest to latest entry per ledger. Longer = deeper moat.
  • Where else does the data live? Zenodo DOI 10.5281/zenodo.19520064 is the permanent archive; Wikidata Q139165733 is the cross-reference. Both independent of avenaterminal.com.
  • Can a memo be tampered with? Each memo writes a SHA-256 of its thesis at generation time. If the rendered memo doesn't hash to the same value, it's been altered.
  • Can the operation be audited? Every cron run logs to cron_logs with status + duration. Every memo to memo_generations. Every AVM to avm_queries. Every API call to api_usage_log.
  • Where is the data resident? Supabase Postgres in eu-central-1 (Frankfurt). All EU data remains in the EU. See /governance §5.

Formal procurement enquiries: institutional@avenaterminal.com

v2026.05 · Live read · cite as avenaterminal.com/proof @ 2026-05-26 20:13 UTC

The Stack · Governance · Live Ops · Methodology