The Sales Elite Engine

One engine.
Every operator runs on it.

Purpose-built for commission programs. Not adapted from a CRM, not bolted onto a spreadsheet — designed from the beginning for how rep firms actually run spiff programs.

01 — Commission engine

Tiers, caps, split. Bit-exact, every time.

Progressive tiered bands apply the right rate to the right slice of revenue. Proportional caps prevent overpayment. The firm/manufacturer split is calculated server-side using decimal arithmetic — no floating-point rounding, no reconciliation spreadsheet.

  • Progressive tiers with proportional slicing
  • Per-brand commission caps
  • Firm / manufacturer split in one calculation
  • Server-side decimal — never a floating-point error
  • Live preview at claim step 6 before submit
Commission tiers · example
$0 – 5K
0%
no commission
$5 – 15K
1.5%
of the band
$15K +
3%
of the band
Against $3,000 cap72%
Firm share $593.42
Manufacturer share $328.12
Total $921.54
02 — The claim experience

A claim that fills itself in.

Field sellers don't think in manufacturer invoices — they think in what they sold. So the claim starts from their own customer PO, and the engine does the mapping.

Step 1
Brand
Pick the manufacturer.
The centerpiece
Step 2
Your PO
Drop the customer PO. AI reads it in the background.
Pre-checked
Step 3
Products
Tiles already checked from the PO. Adjust freely.
Only if required
Step 4
Serials
Pre-checked too. Skipped when the brand doesn't track them.
Pre-filled
Step 5
Details
Sale date, end user, PO number — already in.
Step 6
Submit
You confirm. It's on its way.

Manual entry stays available at every step — the PO is a shortcut, never a requirement.

Claim · Northline PO-8841 · customer-po.pdf
Reading your PO0%
M-2100
Digital mobile radio
12 availableserial-tracked3 selected
P-705
Portable radio
6 availableserial-tracked2 selected
SPK-30
Speaker microphone
23 availableby quantity4 selected
CHG-213
Rapid charger
9 availableby quantity
1 item sold, manufacturer invoice on its way — the claim still goes out; the line resolves itself when the invoice lands.

Fed by the manufacturer

Manufacturers drop their invoices on their side. Every scanned invoice becomes claimable product here — sellers never chase paperwork.

Only what you can claim

Never the full catalog — just the models with quantity available to you, transfers between branches included.

Three characters to find anything

Model or serial number — results adjust live, and what you've checked stays checked while you search.

One PO, one claim per brand

The same PO can carry three brands — three clean claims. A second claim on the same PO and brand is blocked at the database.

Running today

Volume tiers

Percentage of cost, tier by tier — bit-exact basis points, recomputed as volume crosses each threshold.

On the roadmap

Per-unit spiffs

A flat dollar amount per unit sold, funded by the manufacturer — same claims, same settlement, different math.

03 — AI advisory verdict
AI analysis · brand knowledge base
Review recommended
Primary reason

Serial number already claimed 94 days ago by another seller

78% confidence
Rules evaluated
  • Product eligible for the brand programOK
  • Serial number uniquenessFAIL
  • SN within the tracking window (12 mo)OK
  • Banner recognized + activeOK
  • Invoice compliant (date + vendor)WARN
Suggested actionFlag for manual review — confirm with the original seller
Live component from the product. Advisory only — the manufacturer approves, requests info, or rejects. The seller never sees this score.

A second pair of eyes, grounded in the catalog.

Before a claim reaches the manufacturer, the engine runs it against a private AI knowledge base — built from each brand's product catalog, pricing rules, and program terms. It surfaces a confidence score and specific flags.

The seller never sees the score. The manufacturer sees it alongside three buttons: approve, request info, reject. The AI advises. The human decides.

  • Per-brand private knowledge base (your catalog, your rules)
  • Checks pricing, serial number, model eligibility
  • Confidence score + specific flag reasons
  • Advisory only — never autonomous
  • Anthropic API · not used to train · 30-day retention
04 — Rules encoded

Your program rules, encoded in the product.

Each manufacturer gets a private AI knowledge base — a structured document of their program terms, eligible products, pricing bands, and claim requirements. The engine reads this to evaluate every claim. It's not configuration in a UI; it's knowledge, encoded.

brand/rules.md · private knowledge base
# Eligible products
NL-4500 series, NL-220, MR-9100 ...
# Commission tiers (CAD)
$0–5K → 0% | $5–15K → 1.5% | $15K+ → 3%
# Pricing guardrails
Max 5% above MSRP. Flag if below cost.
# Claim requirements
Serial # required. End-user invoice attached.
AI verdict
93%
grounded in your rules
05 — Brand isolation

"You see your brand. Only your brand."

Every manufacturer sees only their own brand — claims, reports, leaderboard, verdicts. Not a permission filter applied after the fact: isolation is structural, enforced at the database level.

The rep firm operates all brands in one engine. Each brand's data never touches another brand's workspace. The data is physically separate.

  • Enforced at the database — not a filter
  • Each brand has its own workspace, reports, leaderboard
  • Sellers only see brands they're certified for
  • Cross-brand data never surfaces to any party
One engine · every brand the firm reps
Your brand
← You see this
Northline
Crestwave
Helix RF
Stratus
Meridian
Vantage
Demo brands — each one physically isolated at the database level
06 — Settlement & clawback

Preview. Confirm. Execute.

Settlement runs in three steps: review the preview, confirm the batch, execute — irreversible. Split to the cent. Every claim batched, every seller paid, every manufacturer accounted for.

Reversal is one click, netted on the next batch. CSV and T4A accounting exports on every settlement.

  • Preview → confirm → execute flow
  • Irreversible once committed
  • One-click clawback, netted next batch
  • CSV + T4A exports for accounting
  • Batched across all eligible claims
Settlement preview · example
Seller Mfr Sales Commission
A. Lévesque Northline $12,687 $40.30
M.-È. Dubois Northline $46,875 $881.24
Total $59,562 $921.54
Payout cycle

From claim to payout

Automatic quarterly payout. No follow-up required.

  1. Claims accumulate
    April → June 2026
    Apr 1 → Jun 30
  2. Settlement prepared
    Admin + MFR validation
    July 5
  3. Final payout review
    Seller confirmation
    July 10
  4. Bank transfer
    Direct deposit (1–3 days)
    July 15
Amount to be paid$0
Account•••• 4729
Processing time1–3 business days
07 — Reports & data

13 reporting surfaces. Data you own.

The longer the engine runs, the more valuable the data becomes. Year-over-year performance, dealer rankings, product heatmaps — scoped to each brand.

Sales heatmap
Month × seller matrix — see who's selling what, when.
Year-over-year
Compare quarters, track growth, spot seasonal patterns.
By dealer / industry
Rank dealers, slice by vertical, find the high performers.
Leaderboard
Brand-isolated seller ranking + goal tracking.
Settlement history
Every settlement, every split, full audit trail.
By product
Which models drive volume — per brand, per period.
Report

Sales activity · 53 weeks

Sales$649,388
Claims249
Hover over a cellBest day: $6,886

Live components from the product — every report exports to PDF, CSV, or Excel.

08 — Trust & compliance

The foundation, not a feature.

Every layer of the stack was built for money. Full security page →

AES-256 banking
Banking data encrypted at rest.
Audit log
Every action, actor, timestamp — immutable.
Loi 25 / GDPR
Versioned consent, erasure, 30-day SLA.
SSO + 4 roles
Seller · Manufacturer · Admin · Director.
Bilingual
EN/FR per user, by design.

See it run on a real account.

We walk through a live instance — your manufacturers, your tiers, your program. No slides, no demos with fake data.