Algonomy Data Agent
Intelligent Data Discovery and Data Quality for Retail Data
Sources Analysed
3
Across UAE region
New Sources Detected
4
Awaiting discovery
Discovery Time Saved
82%
vs. manual baseline
Onboarding Speed-up
8x
KSA reused UAE approved mappings + DQ rules
DQ Rules Proposed
24
8 critical / high
Critical Issues Open
2
Awaiting action
Avg Data Quality Score
91%
Trusted Sales layer
End-to-End Data Product Flow
From raw source metadata to a governed, trusted data product.
Source Metadata
Discovery Agent
BA Approval
Approved Mapping
Data Quality Agent
Governance Approval
Trusted Data Product
Discovery Agent
Scan source metadata, infer semantics, and propose source-to-target mappings for BA review.
Data Quality Agent
Profile trusted data, propose DQ rules, detect incidents, and recommend remediation actions.
Discovery Feed
Live activity from Source Radar, schema drift monitors, and residency guards.
Schema drift: 2 new columns detected in pos_trn_dtl (Retail POS · UAE) since last scan
Review proposed mappings · 2h ago
Retail POS · KSA onboarded via mapping reuse — 6 deltas approved
47 mappings inherited from UAE · 1d ago
New source detected: ecom_orders_ksa
Awaiting discovery kickoff · 2d ago
natl_id flagged residency-restricted under Saudi PDPL
Data must remain in KSA region · 3d ago
Model Learning · Steward feedback loop
Every approved edit, rejected proposal, and steward correction feeds back into the agents. Proposal quality improves measurably week over week — not a static rules engine.
Proposal acceptance rate
89%
+18pp vs 12 weeks ago
Avg BA edits per mapping
0.7
-62% since launch
Corrections replayed into model
1,240
last 90 days
Rules auto-tightened
34
from steward feedback
| Week | Proposals | Accepted | Edited | Rejected | Acceptance |
|---|---|---|---|---|---|
| W-12 | 420 | 297 | 88 | 35 | 71% |
| W-8 | 512 | 400 | 82 | 30 | 78% |
| W-4 | 604 | 502 | 78 | 24 | 83% |
| W-0 (this week) | 688 | 612 | 62 | 14 | 89% |
Auto-tightened after 8 similar corrections
Freshness SLA on pos_trn_hdr: 45m → 30m. Stewards resolved 8 late-batch tickets by tightening this threshold; agent proposed the same tightening on pos_trn_hdr_ksa unprompted.
Rejected proposal, learned
Agent no longer proposes promo_cd → Campaign.Code matches with confidence below 92% — BA rejected 6 low-confidence variants in a row and the model down-weighted the pattern.
Tip: open the Review Queue to see proposed mappings, rules and incidents awaiting human approval, or try Ask the Catalog to query metadata in natural language.