Problem Definition: Tasks / Suspects
By investing in an MDM / EMPI / MPI platform you recognize the need to match your data accurately and completely but all of these platforms fail to match the entire population of data and leave the unresolved matches as potential duplicates (also known as tasks, suspects, worklists, reviews etc...) for humans to work manually.
MDM/EMPI Implementation's are configured to raise tasks / suspects
Tasks / Suspects are resolved by data stewards
Human Data Remediation
Can be inconsistent between stewards
Confined to workdays/hours (M-F, 8-hour days)
Large Backlogs are difficult to overcome (100k – millions)
Adding sources will spike backlogs
Meanwhile new Tasks/Suspects generation continues
Tasks/Suspects typically account for 10-15% of the population – affects match rate
AI Steward Implementation Process:
Production in Weeks
Implementation done in weeks (unlike an MDM implementation)
EntityWise does most of the heavy lifting
Train the Model
(no cost POC)
3 people: approximately 8 hour
End of POC-Functioning AI Steward
Deploy & Test in
Fast/slow as desired
Test n number of Tasks
Deploy in Production
Fast/slow as desired
Solve 1 task to n Tasks
There are 3 steps to resolving any potential duplicate on any platform:
Get the Potential Duplicate out of the Platform
Examine and Make Decision on the records
Put the Answer back into the Platform
This is true whether you are doing this through a team of people or through an automated process.
AI Steward works in conjunction with your existing platform to help resolve these tasks.
How do we engage?
We can engage:
As a one-time project to eliminate a backlog and allow your data team to catchup (a services engagement)
As a subscription where we continuously resolve an ongoing workload 7x24x365
As a file based engagement where we receive a file of potential duplicates and return them resolved.
How AI Steward Works
Why are Data Stewards able to resolve tasks when the MDM/EMPI/MPI cannot.
Data Stewards examine a pair of records differently than a match platform that employs a Deterministic or Probabilistic match approach.
A human looks at each of the attributes individually and makes assessments that weigh the balance of which attributes are more important in addition to how well the values match and strike a balance that is not possible from traditional approaches.
Machine Learning, what AI Steward uses, approaches the decision making process exactly the same way a human does.
It learns from the data steward and can see what the data steward see's when making the decision.
AI Steward can resolve tasks on platforms or file based projects (e.g. MPI Cleaups) at 20x the speed of a data team and at 1/3 the cost. And as a machine learning technology it will learn the match behavior from your team and mimic their accuracy while saving money.