OwnCRM
Case files & evidence

The evidence, on the record.

Real screenshots of the app the engine generates, compiler-proven conversion ledgers, and case files on the real, gnarly reference applications Salesforce ships to the world. Every number here is engine output — re-run the corpus and the verdict is identical.

§ 01Exhibits — same Opportunity, two worldsof 04

The record your team lives in every day — rendered by the meter you rent, and by the app you own.

Exhibit A — what you rent$175/user/mo · illustration

The meter runs while you work: per-seat license, per-seat AI add-on, storage overages — and the UI, the logic, and the data model all stay Salesforce's.

Exhibit B — what you own$0/user/mo · real screenshot
Real screenshot: the standard Opportunity record rendered by the engine-generated React app — layout-driven detail view with Edit and Delete, served from PostgreSQL behind JWT login
YOUR POSTGRESYOUR REACT CODEYOUR SERVERRLS = YOUR SHARING RULES

Real, unedited engine output — not a mockup: a vanilla org’s standard Opportunity, retrieved and transcribed by the engine, deployed to PostgreSQL + Fastify + React with JWT login on a real server, then screenshotted. Every object gets list, detail, and edit views; validation rules become Zod, sharing becomes Postgres RLS. See the edit form and login screen from the same app.

Convert a leadreal screenshot · live demo
Real screenshot: the Convert Lead dialog in the engine-generated CRM — one transaction creates the Account, Contact, and optional Opportunity, then marks the lead converted

Salesforce’s convert semantics, compiled: one transaction creates the Account, Contact, and optional Opportunity, then locks the lead. A second convert is refused. Unmapped fields are reported, never dropped silently.

Close a dealreal screenshot · live demo
Real screenshot: an opportunity closed won in the engine-generated CRM — the full sales stage path from the org's own picklist rendered as a chevron rail

The stage rail is your org’s own Opportunity picklist — Prospecting through Closed Won — not a template. Work the pipeline exactly the way your team already does. See the leads list and lead record from the same app.

§ 02Findings on the record — what a compiler provedof 04

Three claims a competitor can't make without a compiler. Every number below is engine output on six public benchmark orgs — re-run the corpus and the verdict is identical.

Finding 01 · Compiler-proven conversion

When we say converted, a compiler said so.

The LWC → React converter is a deterministic directive table — no LLM, no guesswork. A component earns the word converted only after a real strict-TypeScript compile returns zero diagnostics. Everything short of that is named with its exact blocking construct and line.

Finding 02 · Code, not a PDF

The audit ships as source you can open.

  • Converted React components — dropped into the generated app, not a slide.
  • Per-component rewrite worksheets — done template vs. the behavior our engineer wires, with construct + line.
  • Integration endpoint scaffolds — each citing the original Apex callout site, file:line.
  • Rebuild checklists — the flagged work, itemized and owned by our migration engineers, not left on your desk.
Finding 03 · The auditable estimate

Our quote shows its arithmetic.

Also on the record: a per-file-type verdict ledger (challenge any excluded file), a per-class Apex deep run bucketed by measured complexity, per-page flexipage explanations, and custom labels resolved from your org's own metadata. Every figure here is engine output — re-run it and check.

§ 03Case files — battle-tested on Salesforce's own appsof 04

Not toy orgs. The real, gnarly reference applications Salesforce ships to the world.

Case file 01 · code-heavy org · measured

Coral Cloud Resorts

Salesforce's flagship demo org — code-heavy by design. Schema, UI, and data layer transcribed to a runnable app you own; every Apex class and LWC measured and priced, nothing glossed. 33.1% rebuilds deterministically today — the rest inventoried per item, then rebuilt by our migration team.

6Postgres models
44React app files
100%migration docs
View the full X-Ray report →
Case file 02 · scale stress-test

NPSP — Nonprofit Success Pack

One of the largest managed packages in the ecosystem — parsed without crashing.

317,863lines of Apex
1,035Apex types
12,194files parsed
View the full X-Ray report →
Case file 03 · real production package · shipped 2020

PMM — Program Management Module

A Salesforce.org package running in live nonprofits today, not a demo — program management with 107 Apex classes, 43 LWC and Aura components, and an Experience Cloud volunteer community. Scored with no rounding, it rebuilds 54.1% deterministically right now; every remaining class is inventoried, each flag carrying a reason. This is the honest shape of a genuinely code-heavy org — and the flagged list isn't work left on your desk. It's the plan our migration engineers run for you: a person in the lead, AI as leverage, every item checked for parity against your Salesforce source and recorded verified, draft, or reworked before it ships.

54.1%ready today (deterministic)
107Apex classes measured
43LWC + Aura components
View the full X-Ray report →
Translator deep-run · apex-recipes

17,650 lines of living Apex, parsed clean.

apex-recipes is Salesforce's continuously-updated reference library — the modern canon for how Apex is actually written. We ran the whole thing through the engine. Not a readiness score — a compiler proof: every class parsed, every query classified, nothing hand-waved.

Zero unparsed files. 87.8% of SOQL rewritten to Postgres automatically; the remaining flagged statements — dynamic SOQL, SOSL, cross-object subqueries — each carry a reason and are never silently dropped.

View the full deep-run report →

We never fabricate. The flags are the plan — and the plan is ours to run.

Vendors over-promise a magic 100%. We don't. Everything that deterministically converts is generated; everything that needs a human is flagged with a reason. That flagged list isn't homework for your team — it's the scoped work our migration engineers deliver: a person in the lead, AI as leverage, every item checked for parity against your Salesforce source and recorded verified, draft, or reworked before it ships. Trust is the moat.

§ 04The moat · MetadataContextof 04
The moat · MetadataContext

We document what your org actually does

Fifteen years of admin decisions live in field names no one remembers the reason for. MetadataContext uses an LLM to infer and record the business purpose of every object and field — so the app you own is also the app you understand.

  • Purpose inference for every field & object, in plain language.
  • Verification harness confirms translated logic matches original behavior.
  • Living documentation ships with the code — not a stale wiki.
Read the MetadataContext position paper →

See what your org would save. Free Org X-Ray.