Every performance claim on louieauto.com has a source, a model, an attribution range, and the market adjustments we apply to NOT take full credit. This page discloses all of it. Including the results we don't attribute to ourselves.
This figure is retracted, not revised. It was previously described inconsistently across this site — as a real founder-operator 18-month pre/post measurement on /transparency, and as an 18-month simulation-engine output across 5 virtual dealership models here on this page. Both can't be true of the same number, and neither survives scrutiny: the real-deployment version requires an 18-month window that isn't possible (real development began January 2026), and a full investigation of the platform's own deals tables — deals, accounting_deals, profit_deals, dms_deals, crm_deals, consumer_deals, and store_pvr_stats — found zero real closed deals with real gross data anywhere in the platform to compute a real number from in the first place. Full findings are on /transparency.
The $312 figure and the tables below that were built on top of it (component breakdown, attribution %, market-factor backouts, sensitivity table) have been removed rather than kept up with more caveats attached. They will return, with a real sample size and real date range, once the founder's group is entering real deals through the platform.
Once a real PVR delta exists, we do not intend to claim 100% of it is caused by Louie's routing and coaching logic. The planned approach is a 50–80% attribution range with a 65% published midpoint — the same structure used for the routing-lift numbers on /transparency that are real today. External factors that would not be credited to Louie include Fed rate moves, regional used-vehicle price normalization, the operator's team gaining deal experience over time, and staffing changes at a given rooftop. We are not pre-populating attribution-percentage dollar tables against a retracted base figure — that would just be a more elaborate way of presenting the same unconfirmed number.
The market factors we intend to back out once real data exists are the same ones disclosed on /transparency: Fed rate moves, used-vehicle price stabilization, team tenure growth, and staffing changes at individual rooftops. No dollar table is shown here for the same reason as above — there is no confirmed real base figure to apply these adjustments to yet.
The sensitivity table that used to sit here (attribution % × unit volume → annual gross lift) was derived entirely from the retracted $312 figure and has been removed for the same reason. It will be republished, with the real numbers behind it, once a real PVR delta exists to run through it.
The brain's lender weight calibration is derived from 104,000+ closed-loop lender outcomes — first-look decisions, stip resolutions, funded deal terms, and declined deal patterns — accumulated since January 2026 from the simulation engine running synthetic deal archetypes through the 58-lender matrix.
| Data Layer | Source | Count |
|---|---|---|
| Simulated lender decisions | Brain simulation engine (800K+ deal runs × 58 lenders) | 800K+ simulations |
| Closed-loop lender weight records | lender_weight_cache table, louieauto.db | 26,948 weight records |
| Weighted active outcome calibration | sync-sims-to-closed-loop.js, nightly 1:30am | 104K+ active signals |
| Intelligence log entries | intelligence_log, louieauto.db | 1.07M+ entries |
Important distinction: The 104K outcome calibration is derived primarily from the simulation engine running against realistic dealer archetypes — not 104K independently funded real deals. The simulation is calibrated against national benchmark data (NADA/CFPB/Experian), not operator production data — the simulation outputs themselves are modeled. This distinction matters for external diligence. We disclose it here and in the diligence FAQ.
The 800K+ deal simulation runs are the output of Louie's AI simulation engine running deal archetypes through the 58-lender matrix under varied parameters (FICO tier, LTV, vehicle age, term, market region, lender box changes over time). These are not 800K+ independent customer transactions.
| Metric | Value | What It Means |
|---|---|---|
| Total simulation runs | 800K+ | Simulated deal scenarios run through the lender matrix |
| Scenario templates | 81 | Distinct credit and vehicle scenario templates used as generation seeds (9 roles × 9 scenarios each) |
| Lenders in matrix | 58 | Each scenario type × 58 lenders × parameter sweeps (credit score, LTV, term, region) = 800K+ total |
| Simulation DB size | 0.85GB | simulation.db — queryable on acquisition |
| Intelligence log entries | 1.07M+ | Patterns learned, signals captured, outcomes logged |
We believe this list matters as much as the proof claims. A platform that tells you what it doesn't own is more trustworthy than one that takes credit for everything that went up in the simulation window.
This is a pre-commercial simulation, not a live deployment. All metrics shown on louieauto.com come from the AI simulation engine run across 5 virtual dealership models — validated simulation data since January 2026 — not from external paying customers. LouieAuto is onboarding its first external dealer as of this page's last review (October 2026) — external validation is now underway, not purely a future milestone. Metrics labeled "simulation-based" or "modeled" use the simulation engine's outputs — not independently funded real deals from external dealers.
We disclose this on the diligence FAQ, the acquisition page, and the proof page. We do not hide it. The question for a potential acquirer or customer is not whether external validation exists — it's now underway with our first external dealer onboarding — but whether the platform's architecture, lender data, and IP are worth acquiring at the price, given what that validation would likely prove at scale.
All data tables, source files, and the full simulation database are available for review under NDA. Request access at the acquisition page.
Still have questions about the methodology?