What is commerce decision intelligence?
Commerce decision intelligence is software that turns market observations into specific, evidence-backed decisions for a merchant — what to reprice, what to leave, what to restock — and keeps a ledger of every recommendation, who approved it and what happened next. It sits between price tracking, which only reports, and repricing automation, which acts without a record.
How is decision intelligence different from price tracking?
Price tracking ends at the dashboard: a competitor moved, here is the chart. Decision intelligence continues: given that move, your margin, your stock and your rules, here is the one thing to do, here is the evidence, and here is what it will cost you to be wrong. The difference is a recommendation you can act on and audit, not a number you have to interpret.
How is it different from repricing automation?
Repricing automation acts; it rarely explains, and it almost never grades itself afterwards. Decision intelligence keeps the human decision in the loop and records it: recommendation, evidence, approval, outcome. Automation can be added later, gated on a track record, but the ledger comes first. Melqart is built in that order — recommendations and approvals today, price writes when write access launches — because a merchant should be able to check the advice before trusting the hand.
What does a decision ledger contain?
Every recommendation with the observation that caused it (source URL, timestamp, content hash, snapshot), the guardrails it passed or failed, who approved or rejected it and when, and the outcome measured afterwards. A recommendation that turned out wrong stays in the ledger; that is the point.
- See: dated observations with receipts
- Decide: one recommendation, with its evidence and its risk
- Gate: guardrails, then a person
- Act: the write, approval-gated (launching)
- Grade: the outcome, measured and kept
Who is commerce decision intelligence for?
Merchants with a catalogue too large to price by hand and too valuable to price by a black box: a few hundred to a few thousand SKUs, real margin at stake, and at least three competitors a customer would compare against. Enterprise pricing teams have had this for years; the point of the category is to make it available to a store on Shopify, Salla or Zid without an analyst.
Questions people ask next
Is Melqart a repricer?
It is a decision system with a repricing action behind an approval gate. Today it watches, recommends, records approvals and keeps evidence; the write to your store ships when write access launches and is dated on the changelog.
Does it use AI?
For matching and for drafting explanations, yes; for the decision to change a price, no — that is guardrails and a person. Model providers used are contractually barred from training on your data.