Your ERP calls a supplier 0000104471, the acquired division calls it 104471, and the portal calls it DF-1042.
The values carry the evidence; your team carries the decision.
Mapping told Juno what each column means. Linking answers a second question: when two systems both hold suppliers, are they the same suppliers. The evidence is in the values, and the knowledge index already holds a profile of every column.
Juno picks the pairs worth comparing: two columns mapped to the same ontology attribute, or two columns whose values follow the same shape. A ten-digit number column and a date column stay out of the comparison.
For each pair it compares the values themselves and counts how many appear on both sides. That count becomes the share matched: 94% of the 9,340 supplier numbers in the showcase estate appear in both systems.
The pair arrives on the review screen with the share matched, a sample of the values that agree and a sample of those that stay apart. Your team confirms the link, adjusts it or adds a cross-reference table for the codes that follow their own shape.
Each pair carries its evidence onto the review screen, so a reviewer can judge it in seconds.
Nine digits in SAP and the same nine digits in Oracle, once the leading zeros are set aside. This is the strongest evidence, and it needs a few seconds to measure.
“Delta Fasteners Ltd” beside “Delta Fasteners Limited”. Juno measures how close two names are, and shows the pairs so a person can agree or disagree.
A portal code follows its own shape, so the link runs through a table your team maintains, or through a tax number both systems carry.
The comparison reads the sample of values kept in the knowledge index. The rows stay in the source systems. Product.
One number per pair tells your data owners where the estate agrees and where it needs work.
| Columns compared | What Juno found | Values that agree | What your team did |
|---|---|---|---|
SAP LFA1.LIFNR · Oracle suppliers.SupplierNumber | The same nine digits on both sides, with leading zeros in SAP | 94% of 9,340 | Confirmed as one supplier key |
SAP LFA1.NAME1 · Portal VENDOR_NAME | The same names with small spelling differences | 88% of 1,204 | Confirmed; 142 names reviewed by hand |
SAP MARA.MATNR · Oracle item number | The same material numbers on both sides | 99% of 412,880 | Confirmed as one material key |
| SAP plant code · Warehouse system site | A short code table joins the two | 100% of 14 | Confirmed against the site list |
Portal VENDOR_ID · SAP LFA1.LIFNR | A portal code with a shape of its own | 71% of 860 | A cross-reference table added by hand |
Five pairs from the showcase estate. The share matched is the count of values that appear on both sides. Showcase estate.
The share matched is the number your data owners ask for first. A pair at 99% is a key both systems keep well. A pair at 71% says that most records link and that the rest need a route of their own, usually a cross-reference table or a correction in the source.
Records that stay apart are listed, with the system each one came from. Your team works through them, and each decision joins the evidence for the next run.
Linking makes a question whole: the late rate of Delta Fasteners counts the deliveries recorded in both ERPs and the confirmations on the portal, as one supplier.
Where a client wants one governed record for the supplier itself, with a stewarded golden record and a survivorship rule per field, Lima Axiom holds it and Juno reads it as another registered source.
Showcase estate.
The cancellation case needed the late rate of each supplier, and Delta Fasteners sat in three systems. SAP held 0000104471, the acquired division held 104471 in Oracle, and the portal held DF-1042 with the confirmed ship dates.
Juno compared the supplier columns at 14:05 and reported 94% of 9,340 numbers on both sides, with the portal code at 71%. P. Sharma confirmed the SAP and Oracle link, and uploaded a cross-reference table for the portal codes that afternoon. Delta Fasteners then carried one late rate, 7.4%, and the buyer rule that keeps allocation-bound lines could name the supplier once.
With the estate registered, mapped and linked, a person can ask a question in plain language and see the working behind the answer.
The link is held in the ontology, beside the mapping. Each record stays in the system that holds it, with the key that system uses, and questions reach all of them through the link.
It stays as it is and is listed among the records that stay apart, with the system it came from. Your team decides whether it is a separate supplier or a record to correct.
Linking recognises that two records describe one thing, so a question can join them. Lima Axiom holds one governed record for the thing itself, with stewards and survivorship rules. Estates often run both.
It runs as a background task on the sample values in the knowledge index. Supplier and material columns in the showcase estate were compared in minutes, and the review took the rest of the afternoon.
Each one builds on the one before it.
Reads your rows in place.
Connect and move ↗One record per real thing.
Master and govern ↗Apps for every decision.
Operate and decide ↗Asks and acts, at your level.
Ask and act ↗Bring two systems and one decision your planners make every week. In the first week we register the data, confirm the mapping with your team, answer your questions in ChatLima and run the first agent at the level you choose.