Provenance

Provenance inspector

The five modules on one screen — verified outreach that can't say what it can't prove, optimizing itself inside the truth.

1000recipients · 8 segments
100.0%Gate catch-rate
vs 24.32% single judge
0lie selections
(constrained, 1000 sends)
734lie selections
(unconstrained twin)

① The claim ledger — cited or killed

The ROI variant (A) that won Campaign 1 — now blocked by the MLR hold.

A verified cut in total cost of ownership
  • Helix Analytics cuts total cost of ownership by 47% red — Northwind Health — deployed case study · mlr_hold_tco, roi_outcome_disclaimer
  • Helix Analytics is deployed across 12 hospital systems green — Northwind Health — deployed case study
  • Helix Analytics is priced at $0.12 per patient record green — Helix Analytics — pricing fact sheet
all segments — email & website (same claims, same ledger)
quality__ent · Ethan Haddad @ Lakeshore Health
email c_losc_deployedc_hipaa
website c_losc_deployedc_hipaa
clinops__ent · Aisha Mori @ Northwind Regional Health
email c_speedc_losc_deployed
website c_speedc_losc_deployed
it_security__core · Grace Patel @ Bayview Health
email c_ehrc_speedc_soc2
website c_ehrc_speedc_soc2
quality__core · Liam Larsen @ Northwind Regional Health
email c_deployedc_hipaac_los
website c_deployedc_hipaac_los
cfo__core · Marcus Chen @ Riverbend Care Network
email c_pricec_nofeec_deployed
website c_pricec_nofeec_deployed
cfo__ent · Lena Ahmed @ Meadowbrook Care Network
email c_pricec_nofeec_deployed
website c_pricec_nofeec_deployed
clinops__core · Sofia Chen @ Riverbend Hospital System
email c_losc_speedc_deployed
website c_losc_speedc_deployed
it_security__ent · Marcus Nguyen @ Meadowbrook Health
email c_soc2c_hipaac_encrypt
website c_soc2c_hipaac_encrypt

② The optimizer — can't win by lying

Same recipients, same oracle. The only difference is the truth boundary: the constrained bandit may pull only Gate-cleared arms; the twin may pull the lie.

cumulative regret0send #constrained (verified only)unconstrained (lie allowed)

The planted lie has the highest latent click-rate. The constrained bandit selected it 0 times across 1000 sends. The identical unconstrained bandit selected it 734 times and it won 7/8 segments. Falsehood never entered the reward loop.

segmentconstrained winner (honest)est. CTRunconstrained twin
cfo__corevariant A 24% LIE wins
cfo__entvariant B 18% LIE wins
clinops__corevariant A 24% LIE wins
clinops__entvariant B 25% LIE wins
it_security__corevariant B 24% LIE wins
it_security__entvariant A 23% honest
quality__corevariant B 18% LIE wins
quality__entvariant A 26% LIE wins

③ Drift monitor — truth is temporal

mlr_hold_tco -> ON — re-verified only the affected claim(s), paused only the variants that assert them.

affected claimbeforeafterrecomputedvariants paused
c_tco amber red c_tco cfo__core__email__A, cfo__ent__email__A

Campaign 2 then warm-started from Campaign 1 and adapted: regret fell from 19.68 to 8.64 on the post-drift truth boundary.

④ Assurance Lab — proof, not a promise

Catch-rate by trap type — the Gate vs a single number-blind judge.

number driftunsupported superlativefalse equivalencetrue but unsayableGatesingle judge

Calibration (reliability)

conf →acc

ECE 0.0 · n=47 · illustrative

Gate catch-rate 100.0% vs single-judge 24.3% at 0.0% false-reject. The single judge catches 0.0% of number-drift traps — the material ones — the Gate catches 100.0%.

channeltrapsGate catchsingle-judge catchfalse-reject
email37100.0% 24.3%0.0%
website37100.0% 24.3%0.0%