Takes over when the salesperson has finished the trial close and the customer is ready to move. It structures the pencil, term, down payment and payment, and shows which lenders are most likely to approve the deal and at what gross. Every number deterministic, every route approved by a manager.
A published tier sheet tells you what a lender says it will do. Your own funded history tells you what that lender has actually done, with deals like this one, at your store. Those are different numbers, and only one of them is worth desking on.
Payment, rate, term, LTV, advance and reserve are calculated, not estimated. The same inputs give the same numbers to every manager on every shift, which is the minimum bar for a desk tool.
Which lenders are most likely to approve this structure, ranked on what your store has funded before, and what each route is worth in gross.
Compare structures with the gross on each one, so the conversation with the customer is about a choice rather than a single take-it-or-leave-it pencil.
Each route says what it assumes, what restricts it, and what is missing from the deal. A manager approves a decision, not a black box.
Funded 31 of your last 37 deals in this tier and LTV band. Prefers the 72 month term.
Same rate to the customer, but pays a flat. Subvented program expires in 9 days.
Advance is short at this LTV. Needs proof of income, which is not on file.
Illustrative product interface. Synthetic data, not customer data. Lender names, rates and approval odds are invented for this mockup.
The odds come from your own funded history at this tier and LTV band. The route that cannot fund says why it cannot: the advance is short and proof of income is not on file. That is more useful than a rejection three days later.
The recommendation is a recommendation. The sales manager approves it, and the outcome is recorded against it.
Recording the outcome against a matched control is what makes the lender matching improve. Without it, a routing engine is just an opinion that never finds out whether it was right.
The desk is the middle of a chain, not the start of one. Everything the salesperson learned on the floor is already here, and everything decided here reaches F and I intact.
The customer arrives at the desk with the needs analysis, the trade, the vehicle and the buy-today decision already attached. Nothing is retyped.
The pencil is structured, the routes are ranked, the manager approves.
The approved structure goes to F and I with the reason it was chosen still attached, so the finance manager is not reconstructing the logic from a printout.
The desk recommends. The sales manager decides. Every number is deterministic and reproducible, every route is human-approved before it is submitted, and the structure that was approved is recorded alongside the recommendation that was made.
Velocity Credit feeds the tier this module structures against. It is a separate module because a credit pull carries obligations that deal structuring does not.
In build
Desk is being built alongside Engage, because lender routing and deal structure is where the platform is commercially different from anything else a store could buy. It is also the hardest module to fake, which is a reason to build it early rather than late.
Reviews the credit profile and the approved structure, then recommends the lender most likely to fund it on the best terms for the customer and the store.
Soft pull prequalification and authorized hard pulls, with consent recorded before any inquiry runs and a full adverse action trail behind every decline.
