Hours of mapping work compressed into minutes by MAPR AI
Source and target formats — X12, EDIFACT, IDoc, XML, CSV, JSON
Every rule verified against your real samples before going live
MAPR runs once at build. BPI executes the rules in production.
Trusted by mapping teams worldwide
The mapping agent
80 hours of mapping work, done in 8 minutes.
MAPR AI doesn't suggest mappings — it builds them. Hand it two sample payloads and a sentence of intent. It parses both schemas, drafts a complete rule set, dry-runs against your samples, repairs failures, and applies the live mapping in your IDE.
- Reads X12, EDIFACT, IDoc, XML, CSV, JSON
- Writes complex JSONata — qualifiers, fallbacks, conditionals
- Dry-runs verified before any rule goes live
- Sessions persist — pick up tomorrow where you left off today
MAPR runs at design time. The rules it generates execute inside Orbit BPI's runtime — with no AI in the production path.
How it works
Five steps. Eight minutes.
From paste to publish, mapped by AI, verified by dry-run, and ready to ship into your BPI process flow.
Paste source
X12, EDIFACT, JSON, XML, IDoc, CSV, or flat-file sample payload
Describe target
Pick a schema, paste a sample, or describe the target in plain English
AI drafts rules
MAPR proposes direct paths, JSONata expressions, constants, loops, fallbacks
Dry-run + repair
Run against your samples. Repair anything that doesn't match. Iterate.
Apply to live
Promote the rule set into your BPI mapping IDE. Production runs them.
Before vs after
The same mapping. A different week.
- Build 178 rules by hand, one at a time
- Re-test after every change against full samples
- Patch each edge case (qualifier filters, missing fields) manually
- Wait days for a peer review of a mapping that should take minutes
- Paste two samples. Wait 8 minutes. Get 178 rules.
- Dry-run runs automatically against your real samples
- AI proposes fallbacks, qualifiers, conditional rules — verified
- Reviewer ships changes the same hour, not the same week
Real capabilities
Not a wrapper. A mapping engine.
MAPR understands schemas, qualifiers, loops, and Orbit-specific JSONata patterns. It writes the same kind of rules a senior integration engineer would write — and dry-runs them before you sign off.
Schema-aware rule drafting
MAPR reads field trees, repeated groups, qualifiers, and target structure before drafting rules.
Orbit-safe JSONata
Generated expressions follow Orbit's expression runtime conventions — fallbacks, prefixed functions, conditional formatting.
Loop and qualifier handling
Repeated rows, qualifier-filtered addresses, sibling fields, and target arrays stay connected to the right driver.
Confidence & repair
Every proposed rule carries a confidence score. Low-confidence rules trigger automatic repair attempts.
Validation & governance
Reviewer-grade, not vibe-coded.
MAPR is an agent for production mappings. Every drafted rule is dry-run verified, versioned, reviewer-approvable, and rolling back is one click.
Dry-run before publish
Every proposed mapping runs against your representative samples first. Resolved values, skipped fields, and errors surface before any rule reaches production.
Reviewer-ready output
MAPR emits structured diffs so a reviewer can approve / reject specific rules, not the whole mapping. Audit trail captures every decision.
Versioned and rollback-safe
Every AI-drafted version is captured in the mapping history. Roll back to the previous active version in one click if something breaks downstream.
90-second tour
See MAPR AI draft a mapping in real time.
Bring a real mapping.
Watch MAPR draft it.
Bring two sample payloads — your hardest current mapping. We'll run MAPR AI on them live, dry-run the rules, and review the output together. 30 minutes. Engineer-led.