DataClean Agent

Profiles and cleans tables into a trusted silver layer.

ServiceInsightLangGraph

Book a demo of this agent Run it in the console

Found31

Fixed + verified31

Held for approval0

Recall1.00

One offline run on demo data with planted problems (recorded platform test, November 2026).

How DataClean heals data, step by step

INData inNYC TLC taxi trips.

01Detect31 problems found

02Diagnose5 problem types named

03Plan31 safe to fix · 0 need a person

04Fix + verify31 fixed and re-checked

05ReportPlain-English report and Control Tower record

Client problem: Analysts spend most of their time cleaning data.
What it does: Finds nulls, outliers, duplicates and impossible timestamps.
How it heals: Imputes, quarantines and de-duplicates with a quality log.
Human gate: None.

Typical buyer: Mid-market data teams
Built with: LangGraph self-healing loop
Demo data: NYC TLC taxi trips.
Tables: taxi_trips ← public URL
Target outcome: Target: >95% of planted faults caught.
Path to production: Service.

Issues detected on the demo data

Issue	Risk	Proposed fix	Count
missing_passenger_count 'passenger_count' is empty	low	Impute with the mode (1 passenger)	10
out_of_range_fare_amount fare_amount=-12.61 outside [0, None]	low	Quarantine – refund/void record	6
out_of_range_trip_distance trip_distance=900.0 outside [0, 200]	medium	Quarantine impossible distance	5
duplicate_record Duplicate on trip_id	low	Drop duplicate trip	6
dropoff_before_pickup Drop-off time earlier than pick-up	low	Quarantine – meter clock error	4

How we demo it

Run DataClean on taxi trips.
Show the quarantine table.

Built on AWS

Runs in the platform container on EC2 Graviton (or on demand in Lambda), writes healed tables to the S3 lakehouse, uses Amazon Bedrock only to write explanations, and reports six metrics to CloudWatch.

More Data Automation agents

Related agents

31 · Data Automation