Features

Everything the reconciliation needs

Built to handle non-standardized formats and variable schemas from real-world sources — with the AI doing the tedious part and you keeping control.

01✨ AI-powered

Validation in plain language

Stop writing regex and nested formulas. Describe what a good record looks like in plain English and the local AI turns it into rules.

Pattern & format checks

Regex, date formats, and identifier standards.

Conditional business logic

If-then rules with mathematical operations.

Duplicate detection

Exact and semantic duplicates, across files.

Validation Rules

5 rules configured
Completeness2 rules
Presence / Mandatory Checke.g. email cannot be null
Accuracy & Reasonableness3 rules
Range / Business Rulee.g. premium 1K–500K
Temporal / Domain Plausibilitye.g. date not in future
AI Recommendations
→ Likely Numeric Columns (47 candidates)
AI Plausibility Checks20 / 31 columns
Premium Amount: Must be positiveApplied
Effective Date: Cannot be in futureApplied
Risk Score: Range 0–100Suggested
Deductible: Must not exceed premiumSuggested
02✨ AI-powered

Smart column mapping

A three-layer match — exact, string similarity, and LLM context — proposes how every source column lines up with your target schema.

Exact matching

Handles casing, separators, spaces, camelCase.

String similarity

Edit-distance algorithms for close matches.

LLM context ✨

Semantic matching with sample values and types.

Column Mapping

3 source files → Insurance_v1 dictionary

92%
486 of 528 mapped
Source ColumnDictionary Column
Policy_ID
Policy Number
ClientName
Insured Name
Coverage
Coverage Type
Annual_Premium
Premium Amount
Start_Date
Effective Date
RiskLevel
Risk Score
Deductible_Amt
Deductible
Max_Coverage
Coverage Limit
03✨ AI-powered

AI translation, in action

Two-tier AI — fast embeddings for the simple cases, LLM reasoning for the hard ones — standardizes source-specific codes and values.

LLM reasoning

Deep semantic understanding for complex values.

Semantic embeddings

Fast similarity matching for the obvious cases.

Pattern detection ✨

Learns from the mappings you approve.

Value Translation

Status column · 3 source files · 127 unique values

Exact: 4AI: 4
94%
119 of 127 mapped
Source ValueTarget Value
Active
ACTIVE
Pending
PENDING
In Progress
PENDING
Completed
CLOSED
On Hold
SUSPENDED
Cancelled
CANCELLED
Under Review
PENDING
Closed
CLOSED
Pattern detected: “In Progress”, “Under Review” → “PENDING” (awaiting action). 95% confidence · Apply to 3 similar values
04

The full pipeline, measured

Validation, mapping, translation, and consolidation in one flow — with quality scores at every stage so you trust the output.

Pipeline metrics

Progressive analysis from source files to output.

Quality scores

Integrity, quality, and completeness in real time.

Consolidated output

Clean, standardized, exportable to CSV or Excel.

Consolidation Pipeline

Complete

3 source files → 1 consolidated output

99%
overall quality
Source Integrity
100%

3 of 3 validated

Data Quality
100%

1,847 unique rows

Mapping Coverage
100%

486 of 486 applied

Translation
100%

92 columns translated

Target Completeness
97%

156 of 161 columns

Output Preview 1,847 rows × 161 columns
Policy NumberInsured NameCoveragePremiumEffective DateRisk
POL2024-1847Anderson CorpCommercial Property$24,5002024-01-1578
POL2024-1848TechStart IncCyber Liability$18,7502024-01-1682
POL2024-1849Global LogisticsMarine Cargo$32,1002024-01-1771
05New cycle

Recurring workflows, reused

Swap in next month's source files and keep every mapping, translation, and rule intact. Built for recurring submissions and monthly consolidations.

Step 1

Upload new files

Drop updated sources into an existing session.

Step 2

Auto-match files

Matched by column structure and filename.

Step 3

Re-process instantly

Every configuration is preserved automatically.

What carries forward

Column mappingsTranslationsValidation rulesDictionary & schema

06 · For teams

Built for enterprise workflows

The capabilities that turn a one-off cleanup into a repeatable, auditable operation.

Exception Explorer

Filter and group validation exceptions with AI duplicate detection, so you focus on what matters.

Session Management

Version-controlled, auto-saved workflows. Reuse templates across similar consolidation jobs.

Reference Dictionary

Maintain translation dictionaries and apply proven mappings to new files for consistency.

Progress Tracking

Real-time metrics for data quality, mapping coverage, and translation accuracy at every stage.

Help shape what's next.

Datally is built with the people who use it. Try the demo, then tell us what your data throws at you.

Join the community