AI in Duplicare
Where AI is used in Duplicare
With the release of Duplicare 8, we have integrated AI agents to help in the process of configuring the product to match your specification.
Currently there are 4 main AI agents included in Duplicare:
Dedupe+ Rule Agent
This agent is used to help create new Dedupe+ rules, update existing Dedupe+ rules as well as explain existing rules are doing by using natural language.
This agent will be assisted by the metadata of your system to help the agent make better decisions.
No record data will be passed to the agent at any point.
Master Sort Agent
The Master Sort Agent is used to provide assistance when creating a Master Selection Rule. It can create, edit or explain master selection rules, therefore removing the prerequisite of having any knowledge of SQL.
This agent will be assisted by the metadata of your system to help the agent make better decisions.
No record data will be passed to the agent at any point.
Merge+ Rule Agent
When configuring Merge+ Rules on an table column , this agent can be used to create, edit and describe merge+ rules.
This agent will be assisted by the metadata of your system to help the agent make better decisions.
No record data will be passed to the agent at any point.
Dedupe Results Analysis Agent
You can quickly compare 2 records at a time and gain a visual representation of if those records were, or were not, identified as duplicates to help in building a Dedupe+ rule that meets your specification.
This agent will provide a summary paragraph explaining how exactly a Dedupe+ rule works when identifying duplicates.
This agent will be assisted by the metadata of your system to help the agent make better decisions. This agent will also be passed some boolean values on if a rule has matched – those values are calculated within your Dataverse environment meaning no record data is passed to the agent at any point.
Data Collection and Use
Data8 operates within strict data protection and security guidelines. Only metadata from your environment is used by the agent to make decisions. This metadata helps the agent understand which fields are available but does not include any actual data, records, or personally identifiable information (PII).
The agent also has access to your existing rule configurations as a reference point, enabling it to provide more accurate and contextually relevant responses.
To continuously improve our service, Data8 offers a feedback feature that allows our team to review anonymised chat history with the agents. This helps us enhance the quality and reliability of responses and overall agent behaviour.
Limitations and Risk
It is important to note that content that is generated from these agents may be inaccurate and it is important to verify the information provided from them.