The Agentic Core: How Vandoko Reads Your Market
Dealer promotions are published as artwork, not data. Reading them is the hard part. Inside the engine that parses every competitor's offer, down to the model code and the fine print.
The input is artwork, not data
Here is the part nobody tells you about competitor intelligence: the hardest data in your market is not hidden. It is published every day, by the dealers themselves, as marketing artwork.
A lease special is a banner. The payment is display type, the term is a subhead, the due-at-inception is a smaller line, the trim and model code live in the fine print next to a starburst about MPG. There is no feed, no standardized format, no API. Which is why nobody was collecting it.
Vandoko reads it anyway. That is the job.
What the Core reads out of an offer
From each competitor's published promotions, the Agentic Core parses the parts that were never structured in the first place:
- Lease structure. Payment, term, and what is due at inception, not just a monthly number out of context.
- Model code. The specific configuration behind the offer, so a comparison is between the same vehicle, not two things that share a name.
- Finance offers, discounts, and rebates. APR terms and the money behind them.
- The promotion itself. The actual artwork is passed through to your dashboard next to the parsed data, so you can check the comparison with your own eyes.
Model code, not model name
Offer comparison runs at model code, because that is the level a deal is actually structured at. Inventory and market mix run at model name, how many of a given model a store carries, with the code one click deeper. Two views, because the desk and the principal are asking two different questions.
The compliance read
Every offer's disclaimers are evaluated against OEM guidelines and the federal advertising rules that govern them: Regulation M for lease advertising and Regulation Z for credit and APR. When a competitor's advertised offer is missing required terms, your team sees that flag inside the product, alongside the offer itself.
The daily cycle
Every competitor's offers are checked daily and logged. That log is what makes the intelligence trustworthy: the morning briefing's narrative is written strictly over computed facts, three to five sentences grounded in the numbers underneath, not a language model's guess about your market.
Beaten on structure, not price
Here is why the parsing depth matters. Same year, make, model, and model code. A competitor publishes a compliant offer structured differently, and it lands at a lower monthly payment. The shopper does not read structure. The shopper reads the better deal, and goes.
If your comparison runs at the model-name level, you never see it happen. At the model-code level, with the structure parsed, you see exactly what you are up against and what a response would take.
None of this is hidden
Your competitors publish this every day. Nobody was collecting it, because the input is artwork and reading it required building for that. The data was always public. Reading it was the hard part.
See what the Core reads in your market. Start my trial.
