Features

Automatic by default. Detailed when you need it.

One place to add your files, follow the work, and review the result. The full preparation workbench is there in Advanced mode.

Automatic preparation

Datally takes responsibility for the next step.

Start with source files and an optional target. Datally coordinates the work through to the first output, bringing unresolved choices back to you.

With or without a target

Use your reference structure, or review one proposed from your source files.

Coordinated preparation

Analysis, mapping, translation, validation, and the first output run are connected for you.

Visible decisions and progress

Follow the activity feed and resolve uncertain choices before continuing.

Access to the details

Open Advanced to inspect the dictionaries, mappings, translations, and rules behind your result.

Start your first project
Interactive example · sample data01 / 04

Project

Regional customer files

Automatic
  1. Files
  2. Prepare
  3. Review
  4. Output

Start with the files you have.

Different headers. The same kind of records. No target template supplied.

north.csv

Customer ID · Region · Status

2 rows

south.xlsx

Client ID · Region · Status

2 rows
Advanced controls are there when you need them.
Explore the Advanced workbench
01AI assistance available

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.

Reviewable findings

Inspect the records that fail each configured check.

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

Illustrative workbench view with sample data.

02AI assistance available

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.

AI context

Model-assisted 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

Illustrative workbench view with sample data.

03AI assistance available

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.

Saved decisions

Keep approved translations for subsequent runs.

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

Illustrative workbench view with sample data.

04

The full pipeline, measured

Validation, mapping, translation, and consolidation in one flow, with coverage and exception counts to help you assess the output.

Pipeline metrics

Progressive analysis from source files to output.

Review evidence

Mapping coverage, row counts, and validation findings.

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

Illustrative workbench view with sample data.

05New cycle

Recurring workflows, reused

Reuse accepted mappings, translations, and rules for the next set of source files. Review changes in structure or values before producing a new result.

Step 1

Add new files

Add updated sources to a project or a new Pipeline run.

Step 2

Auto-match files

Matched by column structure and filename.

Step 3

Review and run

Reuse accepted settings and review the impact of changed files.

What carries forward

Column mappingsTranslationsValidation rulesDictionary & schema

A lasting workspace

Keep the work behind the result

Saved projects and reusable Pipelines help you carry accepted decisions into the next preparation.

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

Follow preparation stages, mapping coverage, and validation findings as the work runs.

Try Automatic with your own files.

Start with a small project and inspect the results before moving on to a larger dataset.

Explore trial access