What the AI does, in plain terms
Elastly's models learn three things from your history: how much each customer segment will pay for each part family before win rate drops, how each part's demand moves through the year, and how the market around you behaves when you track competitors. From that, the engine sets a price for every line that keeps the margin at a stable win rate. The AI is the learning; the price is still bounded by your policy.
There is no language model guessing at prices. The models are statistical, trained on your own quotes, and every output shows the factors that produced it.
Trained on your data, not someone else's
Benchmarks from other companies tell you what a bearing sells for somewhere. Your history tells you what your workshop in Sharjah accepted last Tuesday. The second is what prices your next quote.
- Trained only on your workspace's quotes, orders, cost and stock
- Never used for another customer
- Retrained as outcomes sync, with drift shown against your policy
- Confidence per price, and a threshold below which the engine does not move it
Why every AI price shows its reasons
A rep will not send a price they cannot defend, and a pricing manager will not approve a model they cannot see into. So each price carries the factors (customer tier, volume, competitor index, stock cover) with their size, the policy that capped it and the confidence. The decision inspector keeps all of it, so the question "why did the AI do that" has an answer months later.
Where AI helps the team, not just the price
The strategy drafter writes a rule from a sentence ("protect margin on the long tail for workshops") and shows the scope it chose. The quote assistant proposes line changes with their margin effect. Insights surface where discount runs ahead of volume. All of it is reviewed by a person before it changes a price.