The MDM Hub Is Dead. Long Live the Bond.
Traditional MDM builds one hub, matches and merges everything into it, then watches the golden record decay. Next-gen MDM inverts the model: bond data where it lives — virtual, confidence-scored, continuous.
Nir Rachmani, Head of Growth
· 3 min read

For thirty years, Master Data Management meant one thing: build a hub. Pull every customer, supplier, and product record out of the systems that hold them, run match-and-merge, and crown a single winning row as the golden record. The hub owns the truth. Everything else defers to it. That was the whole idea — and in an AI era, it's finished.
The hub was obsolete the moment it shipped
A centralized golden record starts decaying the instant updates begin happening in more than one place at a time — which, in a real company, is always. Sales edits an account. Billing corrects a legal entity. A regional team applies its own naming convention. The hub can't see those edits until the next sync, so the “master” copy is quietly wrong between refreshes. You didn't build a single source of truth. You built a fourth version of it and gave it a certificate.
Match-and-merge makes it worse. Every merge is a destructive guess. Collapse two records that shouldn't have been collapsed and you've corrupted the golden record for everyone downstream — with no confidence score, no lineage, no way to ask how sure the system actually was.
The old primitive was the record. The new one is the Bond.
Next-gen MDM stops asking “which copy wins?” and starts asking “how do these records relate?” Instead of one hub owning the record, you connect data where it already lives — virtually, continuously, and without moving or storing a single raw row. The unit of mastering is no longer the merged record in the hub. It's the Bond.
A Bond defines how data in one system relates to data in another: which fields link, and how records match. Bondata's AI reads metadata and hashed values across your connected systems and proposes these relationships as Suggested Bonds — each one carrying a confidence score. You approve or reject from the Suggested tab; confirmed Bonds become the load-bearing connections everything else is built on. Nothing is silently merged. Every link is either something the AI is scoring for you or something you've explicitly confirmed.
Confidence beats consolidation
This is the part the hub could never do. Bonds aren't binary. You can match on exact values, or set a partial match with a confidence threshold — say 90 to 100 percent — to absorb typos and formatting drift without fabricating a false merge. You can compound fields, matching on first name plus last name plus domain instead of betting everything on one column. The relationship is expressed as a probability, not a destructive assertion.
So instead of a golden record that hides its own uncertainty, you get a living graph that surfaces it. That's the difference between a system that says “trust me” and one that shows its work.
Zero-copy, or it doesn't scale
The hub model requires copying your most sensitive data into one more place — the exact opposite of what you want when AI agents are about to act on it. Bondata inverts that too. It stores only metadata, connection details, and hashed values. Raw data is accessed only while a workflow runs, never warehoused in a central master. SOC 2 Type II, ISO 27001, GDPR — mastering without the honeypot.
The map you can actually see
Because Bonds live between systems instead of inside a monolith, the whole model is legible. The Data Map is a visual graph of every Bond across your stack — click an entity and see its active and suggested relationships, approve inline, filter by system or bond type. It's not a snapshot from last quarter's mapping project. It's the current shape of your data, maintained as the systems underneath it change.
What next-gen MDM actually means
Not one hub owning the record, but a network of confidence-scored Bonds connecting the records where they already are. Not a months-long merge project that's stale on delivery, but continuous, self-maintaining relationships. Not a copy of your data, but a context layer over it — the substrate AI agents need to act without hallucinating.
The hub gave you a single row and asked you to believe it. The Bond gives you the whole relationship and shows you how sure it is. The record was always going to decay. The Bond keeps up.
See your Golden Record map itself
Every team needs MDM. Walk your systems with us and watch fragmented data bond into one trusted source of truth — in hours, not months.
Book a Demo

