Reltio Connect

Hi All,

I am trying to understand or get more information on the below items, would really appreciate if one can share some light into these areas.

1. How the recently introduced MCP server integrates Reltio and other AI tools/applications.
2. Does the MCP server uses  Reltio APIs to facilitate the integration between the two sides. 
3. Does the MCP server translates any prompts from AI tools to construct the response.
4. What is the level of details can be made available to AI tools/apps using the MCP server.
5. Apart from the MCP server, are there any options to integrate MDM with Gen AI/Open AI tools/apps.

Thanks!

Hi Vinoth, did you manage to look through this post shared Sergio Abraham on the Reltio MCP Server?

I have been playing  / hacking around with the MCP Server so can give you my take on it so long but am interested in what others have to say too.

I'm using the Claude desktop app on a Mac just so you know.  After install, the MCP Server is exposed in Claude under Settings -> Developer. If you click on the "reltio-mcp-server", it should show a state of "running". If not you may need to troubleshoot a bit based on looking at the logs.

The Reltio MCP Server exposes a series of "tools" (written in python) that act as wrappers to various Reltio API's (search entity, get entity, get matches etc).  By including markup in these python files along with the actual API calls, the LLM is able to intelligently figure out which tools to use based on your prompt.  It's actually very easy to extend the MCP Server with your own tools if you want to as part of learning how this works.  I have very little experience with python myself but with the help of the Claude app, I was able to generate some python very easily to add additional tools to the MCP Server.  However you certainly don't have to add additional tools in order to use the MCP Server as it is. 

Once it's running just start asking it some questions such as "search for a person with a phone number of bla bla" or "fetch the details about entity id bla bla".  It's amazing to watch how Claude will take your prompt and then use the tools (assuming your MCP Server is running of course). It may even try various tools if it doesn't get a result with one tool.  I've been stunned at times just to watch what it does and how it returns information.  Improved prompting will render better results so try different prompts and see how your output varies.  If you're interested, I can share a short video too about some new tools I added to my local version of the MCP Server in case you would like to try that too.

This is such a fascinating area and as we all know it's evolving almost daily. Let's all help each other get up to speed quickly and keep improving our knowledge in this area because this MCP Server concept coupled with the power of LLM's is an amazing capability!

Hi All,

I am trying to understand or get more information on the below items, would really appreciate if one can share some light into these areas.

1. How the recently introduced MCP server integrates Reltio and other AI tools/applications.
2. Does the MCP server uses  Reltio APIs to facilitate the integration between the two sides. 
3. Does the MCP server translates any prompts from AI tools to construct the response.
4. What is the level of details can be made available to AI tools/apps using the MCP server.
5. Apart from the MCP server, are there any options to integrate MDM with Gen AI/Open AI tools/apps.

Thanks!

Hi @Guy Vorster,

Thank you for the response, it helped for sure. 
I was able to setup the MCP server using Claude, but the response is not  great.
Simple search/get prompts are not giving results except the one get details using entity ID.

Would be great if you could share any content to better understand or customize the solution. 
Appreciate your inputs.

Hi Vinoth, did you manage to look through this post shared Sergio Abraham on the Reltio MCP Server?

I have been playing  / hacking around with the MCP Server so can give you my take on it so long but am interested in what others have to say too.

I'm using the Claude desktop app on a Mac just so you know.  After install, the MCP Server is exposed in Claude under Settings -> Developer. If you click on the "reltio-mcp-server", it should show a state of "running". If not you may need to troubleshoot a bit based on looking at the logs.

The Reltio MCP Server exposes a series of "tools" (written in python) that act as wrappers to various Reltio API's (search entity, get entity, get matches etc).  By including markup in these python files along with the actual API calls, the LLM is able to intelligently figure out which tools to use based on your prompt.  It's actually very easy to extend the MCP Server with your own tools if you want to as part of learning how this works.  I have very little experience with python myself but with the help of the Claude app, I was able to generate some python very easily to add additional tools to the MCP Server.  However you certainly don't have to add additional tools in order to use the MCP Server as it is. 

Once it's running just start asking it some questions such as "search for a person with a phone number of bla bla" or "fetch the details about entity id bla bla".  It's amazing to watch how Claude will take your prompt and then use the tools (assuming your MCP Server is running of course). It may even try various tools if it doesn't get a result with one tool.  I've been stunned at times just to watch what it does and how it returns information.  Improved prompting will render better results so try different prompts and see how your output varies.  If you're interested, I can share a short video too about some new tools I added to my local version of the MCP Server in case you would like to try that too.

This is such a fascinating area and as we all know it's evolving almost daily. Let's all help each other get up to speed quickly and keep improving our knowledge in this area because this MCP Server concept coupled with the power of LLM's is an amazing capability!

Hi All,

I am trying to understand or get more information on the below items, would really appreciate if one can share some light into these areas.

1. How the recently introduced MCP server integrates Reltio and other AI tools/applications.
2. Does the MCP server uses  Reltio APIs to facilitate the integration between the two sides. 
3. Does the MCP server translates any prompts from AI tools to construct the response.
4. What is the level of details can be made available to AI tools/apps using the MCP server.
5. Apart from the MCP server, are there any options to integrate MDM with Gen AI/Open AI tools/apps.

Thanks!

Hey Vinoth

Give me a bit more context about the tenant you're using (what type of data, what industry) and send a few examples of the prompts you were using (feel free to sanitize them a bit if needed).  I can try them on my side (if they involve people and or organizations) and see how they work for me.  Granted we will have different data but if we are dealing with people or organizations at least the questions will be similar.

Happy to help.

Hi @Guy Vorster,

Thank you for the response, it helped for sure. 
I was able to setup the MCP server using Claude, but the response is not  great.
Simple search/get prompts are not giving results except the one get details using entity ID.

Would be great if you could share any content to better understand or customize the solution. 
Appreciate your inputs.

Hi Vinoth, did you manage to look through this post shared Sergio Abraham on the Reltio MCP Server?

I have been playing  / hacking around with the MCP Server so can give you my take on it so long but am interested in what others have to say too.

I'm using the Claude desktop app on a Mac just so you know.  After install, the MCP Server is exposed in Claude under Settings -> Developer. If you click on the "reltio-mcp-server", it should show a state of "running". If not you may need to troubleshoot a bit based on looking at the logs.

The Reltio MCP Server exposes a series of "tools" (written in python) that act as wrappers to various Reltio API's (search entity, get entity, get matches etc).  By including markup in these python files along with the actual API calls, the LLM is able to intelligently figure out which tools to use based on your prompt.  It's actually very easy to extend the MCP Server with your own tools if you want to as part of learning how this works.  I have very little experience with python myself but with the help of the Claude app, I was able to generate some python very easily to add additional tools to the MCP Server.  However you certainly don't have to add additional tools in order to use the MCP Server as it is. 

Once it's running just start asking it some questions such as "search for a person with a phone number of bla bla" or "fetch the details about entity id bla bla".  It's amazing to watch how Claude will take your prompt and then use the tools (assuming your MCP Server is running of course). It may even try various tools if it doesn't get a result with one tool.  I've been stunned at times just to watch what it does and how it returns information.  Improved prompting will render better results so try different prompts and see how your output varies.  If you're interested, I can share a short video too about some new tools I added to my local version of the MCP Server in case you would like to try that too.

This is such a fascinating area and as we all know it's evolving almost daily. Let's all help each other get up to speed quickly and keep improving our knowledge in this area because this MCP Server concept coupled with the power of LLM's is an amazing capability!

Hi All,

I am trying to understand or get more information on the below items, would really appreciate if one can share some light into these areas.

1. How the recently introduced MCP server integrates Reltio and other AI tools/applications.
2. Does the MCP server uses  Reltio APIs to facilitate the integration between the two sides. 
3. Does the MCP server translates any prompts from AI tools to construct the response.
4. What is the level of details can be made available to AI tools/apps using the MCP server.
5. Apart from the MCP server, are there any options to integrate MDM with Gen AI/Open AI tools/apps.

Thanks!

Hi @Guy Vorster,

Am using a partner tenant, that has Individual and Org customer data.
I was using the sample prompts below ,

1. Search for Individual profiles from city Chennai
2. Search for individual profiles with First Name Vinoth
3. Get the list of profiles without Last Name

Thanks again for helping out.

Hey Vinoth

Give me a bit more context about the tenant you're using (what type of data, what industry) and send a few examples of the prompts you were using (feel free to sanitize them a bit if needed).  I can try them on my side (if they involve people and or organizations) and see how they work for me.  Granted we will have different data but if we are dealing with people or organizations at least the questions will be similar.

Happy to help.

Hi @Guy Vorster,

Thank you for the response, it helped for sure. 
I was able to setup the MCP server using Claude, but the response is not  great.
Simple search/get prompts are not giving results except the one get details using entity ID.

Would be great if you could share any content to better understand or customize the solution. 
Appreciate your inputs.

Hi Vinoth, did you manage to look through this post shared Sergio Abraham on the Reltio MCP Server?

I have been playing  / hacking around with the MCP Server so can give you my take on it so long but am interested in what others have to say too.

I'm using the Claude desktop app on a Mac just so you know.  After install, the MCP Server is exposed in Claude under Settings -> Developer. If you click on the "reltio-mcp-server", it should show a state of "running". If not you may need to troubleshoot a bit based on looking at the logs.

The Reltio MCP Server exposes a series of "tools" (written in python) that act as wrappers to various Reltio API's (search entity, get entity, get matches etc).  By including markup in these python files along with the actual API calls, the LLM is able to intelligently figure out which tools to use based on your prompt.  It's actually very easy to extend the MCP Server with your own tools if you want to as part of learning how this works.  I have very little experience with python myself but with the help of the Claude app, I was able to generate some python very easily to add additional tools to the MCP Server.  However you certainly don't have to add additional tools in order to use the MCP Server as it is. 

Once it's running just start asking it some questions such as "search for a person with a phone number of bla bla" or "fetch the details about entity id bla bla".  It's amazing to watch how Claude will take your prompt and then use the tools (assuming your MCP Server is running of course). It may even try various tools if it doesn't get a result with one tool.  I've been stunned at times just to watch what it does and how it returns information.  Improved prompting will render better results so try different prompts and see how your output varies.  If you're interested, I can share a short video too about some new tools I added to my local version of the MCP Server in case you would like to try that too.

This is such a fascinating area and as we all know it's evolving almost daily. Let's all help each other get up to speed quickly and keep improving our knowledge in this area because this MCP Server concept coupled with the power of LLM's is an amazing capability!

Hi All,

I am trying to understand or get more information on the below items, would really appreciate if one can share some light into these areas.

1. How the recently introduced MCP server integrates Reltio and other AI tools/applications.
2. Does the MCP server uses  Reltio APIs to facilitate the integration between the two sides. 
3. Does the MCP server translates any prompts from AI tools to construct the response.
4. What is the level of details can be made available to AI tools/apps using the MCP server.
5. Apart from the MCP server, are there any options to integrate MDM with Gen AI/Open AI tools/apps.

Thanks!

Thanks for sharing these examples Vinoth. Let me try similar ones in my tenant and I will report back.  One question though; are you getting errors in the mcp server or just zero results for all of these?  thanks

Hi @Guy Vorster,

Am using a partner tenant, that has Individual and Org customer data.
I was using the sample prompts below ,

1. Search for Individual profiles from city Chennai
2. Search for individual profiles with First Name Vinoth
3. Get the list of profiles without Last Name

Thanks again for helping out.

Hey Vinoth

Give me a bit more context about the tenant you're using (what type of data, what industry) and send a few examples of the prompts you were using (feel free to sanitize them a bit if needed).  I can try them on my side (if they involve people and or organizations) and see how they work for me.  Granted we will have different data but if we are dealing with people or organizations at least the questions will be similar.

Happy to help.

Hi @Guy Vorster,

Thank you for the response, it helped for sure. 
I was able to setup the MCP server using Claude, but the response is not  great.
Simple search/get prompts are not giving results except the one get details using entity ID.

Would be great if you could share any content to better understand or customize the solution. 
Appreciate your inputs.

Hi Vinoth, did you manage to look through this post shared Sergio Abraham on the Reltio MCP Server?

I have been playing  / hacking around with the MCP Server so can give you my take on it so long but am interested in what others have to say too.

I'm using the Claude desktop app on a Mac just so you know.  After install, the MCP Server is exposed in Claude under Settings -> Developer. If you click on the "reltio-mcp-server", it should show a state of "running". If not you may need to troubleshoot a bit based on looking at the logs.

The Reltio MCP Server exposes a series of "tools" (written in python) that act as wrappers to various Reltio API's (search entity, get entity, get matches etc).  By including markup in these python files along with the actual API calls, the LLM is able to intelligently figure out which tools to use based on your prompt.  It's actually very easy to extend the MCP Server with your own tools if you want to as part of learning how this works.  I have very little experience with python myself but with the help of the Claude app, I was able to generate some python very easily to add additional tools to the MCP Server.  However you certainly don't have to add additional tools in order to use the MCP Server as it is. 

Once it's running just start asking it some questions such as "search for a person with a phone number of bla bla" or "fetch the details about entity id bla bla".  It's amazing to watch how Claude will take your prompt and then use the tools (assuming your MCP Server is running of course). It may even try various tools if it doesn't get a result with one tool.  I've been stunned at times just to watch what it does and how it returns information.  Improved prompting will render better results so try different prompts and see how your output varies.  If you're interested, I can share a short video too about some new tools I added to my local version of the MCP Server in case you would like to try that too.

This is such a fascinating area and as we all know it's evolving almost daily. Let's all help each other get up to speed quickly and keep improving our knowledge in this area because this MCP Server concept coupled with the power of LLM's is an amazing capability!

Hi All,

I am trying to understand or get more information on the below items, would really appreciate if one can share some light into these areas.

1. How the recently introduced MCP server integrates Reltio and other AI tools/applications.
2. Does the MCP server uses  Reltio APIs to facilitate the integration between the two sides. 
3. Does the MCP server translates any prompts from AI tools to construct the response.
4. What is the level of details can be made available to AI tools/apps using the MCP server.
5. Apart from the MCP server, are there any options to integrate MDM with Gen AI/Open AI tools/apps.

Thanks!

Hi Vinoth. I just tried the query against my demo tenant "Search for Individual profiles from city Boston" which is exactly the same as your query except for using Boston vs Chennai. I received results back from my search without any problem so am interested to learn more about why you're not seeing any results.  Perhaps share a screenshot of the faceted search from your reltio tenant with Chennai as the selected city and then share a screenshot of what you get in Claude via the MCP Server.

Thanks for sharing these examples Vinoth. Let me try similar ones in my tenant and I will report back.  One question though; are you getting errors in the mcp server or just zero results for all of these?  thanks

Hi @Guy Vorster,

Am using a partner tenant, that has Individual and Org customer data.
I was using the sample prompts below ,

1. Search for Individual profiles from city Chennai
2. Search for individual profiles with First Name Vinoth
3. Get the list of profiles without Last Name

Thanks again for helping out.

Hey Vinoth

Give me a bit more context about the tenant you're using (what type of data, what industry) and send a few examples of the prompts you were using (feel free to sanitize them a bit if needed).  I can try them on my side (if they involve people and or organizations) and see how they work for me.  Granted we will have different data but if we are dealing with people or organizations at least the questions will be similar.

Happy to help.

Hi @Guy Vorster,

Thank you for the response, it helped for sure. 
I was able to setup the MCP server using Claude, but the response is not  great.
Simple search/get prompts are not giving results except the one get details using entity ID.

Would be great if you could share any content to better understand or customize the solution. 
Appreciate your inputs.

Hi Vinoth, did you manage to look through this post shared Sergio Abraham on the Reltio MCP Server?

I have been playing  / hacking around with the MCP Server so can give you my take on it so long but am interested in what others have to say too.

I'm using the Claude desktop app on a Mac just so you know.  After install, the MCP Server is exposed in Claude under Settings -> Developer. If you click on the "reltio-mcp-server", it should show a state of "running". If not you may need to troubleshoot a bit based on looking at the logs.

The Reltio MCP Server exposes a series of "tools" (written in python) that act as wrappers to various Reltio API's (search entity, get entity, get matches etc).  By including markup in these python files along with the actual API calls, the LLM is able to intelligently figure out which tools to use based on your prompt.  It's actually very easy to extend the MCP Server with your own tools if you want to as part of learning how this works.  I have very little experience with python myself but with the help of the Claude app, I was able to generate some python very easily to add additional tools to the MCP Server.  However you certainly don't have to add additional tools in order to use the MCP Server as it is. 

Once it's running just start asking it some questions such as "search for a person with a phone number of bla bla" or "fetch the details about entity id bla bla".  It's amazing to watch how Claude will take your prompt and then use the tools (assuming your MCP Server is running of course). It may even try various tools if it doesn't get a result with one tool.  I've been stunned at times just to watch what it does and how it returns information.  Improved prompting will render better results so try different prompts and see how your output varies.  If you're interested, I can share a short video too about some new tools I added to my local version of the MCP Server in case you would like to try that too.

This is such a fascinating area and as we all know it's evolving almost daily. Let's all help each other get up to speed quickly and keep improving our knowledge in this area because this MCP Server concept coupled with the power of LLM's is an amazing capability!

Hi All,

I am trying to understand or get more information on the below items, would really appreciate if one can share some light into these areas.

1. How the recently introduced MCP server integrates Reltio and other AI tools/applications.
2. Does the MCP server uses  Reltio APIs to facilitate the integration between the two sides. 
3. Does the MCP server translates any prompts from AI tools to construct the response.
4. What is the level of details can be made available to AI tools/apps using the MCP server.
5. Apart from the MCP server, are there any options to integrate MDM with Gen AI/Open AI tools/apps.

Thanks!

Hi @Guy Vorster,

Adding the sample responses from Hub and Claude.

Thanks for helping out.

Hi Vinoth. I just tried the query against my demo tenant "Search for Individual profiles from city Boston" which is exactly the same as your query except for using Boston vs Chennai. I received results back from my search without any problem so am interested to learn more about why you're not seeing any results.  Perhaps share a screenshot of the faceted search from your reltio tenant with Chennai as the selected city and then share a screenshot of what you get in Claude via the MCP Server.

Thanks for sharing these examples Vinoth. Let me try similar ones in my tenant and I will report back.  One question though; are you getting errors in the mcp server or just zero results for all of these?  thanks

Hi @Guy Vorster,

Am using a partner tenant, that has Individual and Org customer data.
I was using the sample prompts below ,

1. Search for Individual profiles from city Chennai
2. Search for individual profiles with First Name Vinoth
3. Get the list of profiles without Last Name

Thanks again for helping out.

Hey Vinoth

Give me a bit more context about the tenant you're using (what type of data, what industry) and send a few examples of the prompts you were using (feel free to sanitize them a bit if needed).  I can try them on my side (if they involve people and or organizations) and see how they work for me.  Granted we will have different data but if we are dealing with people or organizations at least the questions will be similar.

Happy to help.

Hi @Guy Vorster,

Thank you for the response, it helped for sure. 
I was able to setup the MCP server using Claude, but the response is not  great.
Simple search/get prompts are not giving results except the one get details using entity ID.

Would be great if you could share any content to better understand or customize the solution. 
Appreciate your inputs.

Hi Vinoth, did you manage to look through this post shared Sergio Abraham on the Reltio MCP Server?

I have been playing  / hacking around with the MCP Server so can give you my take on it so long but am interested in what others have to say too.

I'm using the Claude desktop app on a Mac just so you know.  After install, the MCP Server is exposed in Claude under Settings -> Developer. If you click on the "reltio-mcp-server", it should show a state of "running". If not you may need to troubleshoot a bit based on looking at the logs.

The Reltio MCP Server exposes a series of "tools" (written in python) that act as wrappers to various Reltio API's (search entity, get entity, get matches etc).  By including markup in these python files along with the actual API calls, the LLM is able to intelligently figure out which tools to use based on your prompt.  It's actually very easy to extend the MCP Server with your own tools if you want to as part of learning how this works.  I have very little experience with python myself but with the help of the Claude app, I was able to generate some python very easily to add additional tools to the MCP Server.  However you certainly don't have to add additional tools in order to use the MCP Server as it is. 

Once it's running just start asking it some questions such as "search for a person with a phone number of bla bla" or "fetch the details about entity id bla bla".  It's amazing to watch how Claude will take your prompt and then use the tools (assuming your MCP Server is running of course). It may even try various tools if it doesn't get a result with one tool.  I've been stunned at times just to watch what it does and how it returns information.  Improved prompting will render better results so try different prompts and see how your output varies.  If you're interested, I can share a short video too about some new tools I added to my local version of the MCP Server in case you would like to try that too.

This is such a fascinating area and as we all know it's evolving almost daily. Let's all help each other get up to speed quickly and keep improving our knowledge in this area because this MCP Server concept coupled with the power of LLM's is an amazing capability!

Hi All,

I am trying to understand or get more information on the below items, would really appreciate if one can share some light into these areas.

1. How the recently introduced MCP server integrates Reltio and other AI tools/applications.
2. Does the MCP server uses  Reltio APIs to facilitate the integration between the two sides. 
3. Does the MCP server translates any prompts from AI tools to construct the response.
4. What is the level of details can be made available to AI tools/apps using the MCP server.
5. Apart from the MCP server, are there any options to integrate MDM with Gen AI/Open AI tools/apps.

Thanks!

OK Vinoth I suspect it may be because your tenant does not have "search by ov" enabled.  In order to confirm this, please can you check in your claude logs. On a Mc they are in your Library folder, then Logs, then Claude.  You should see a file (likely more than one) with the name of your reltio mcp server.  Scroll down and see if you can see search errors.  Feel free to attach your log file so I can have a look

Hi @Guy Vorster,

Adding the sample responses from Hub and Claude.

Thanks for helping out.

Hi Vinoth. I just tried the query against my demo tenant "Search for Individual profiles from city Boston" which is exactly the same as your query except for using Boston vs Chennai. I received results back from my search without any problem so am interested to learn more about why you're not seeing any results.  Perhaps share a screenshot of the faceted search from your reltio tenant with Chennai as the selected city and then share a screenshot of what you get in Claude via the MCP Server.

Thanks for sharing these examples Vinoth. Let me try similar ones in my tenant and I will report back.  One question though; are you getting errors in the mcp server or just zero results for all of these?  thanks

Hi @Guy Vorster,

Am using a partner tenant, that has Individual and Org customer data.
I was using the sample prompts below ,

1. Search for Individual profiles from city Chennai
2. Search for individual profiles with First Name Vinoth
3. Get the list of profiles without Last Name

Thanks again for helping out.

Hey Vinoth

Give me a bit more context about the tenant you're using (what type of data, what industry) and send a few examples of the prompts you were using (feel free to sanitize them a bit if needed).  I can try them on my side (if they involve people and or organizations) and see how they work for me.  Granted we will have different data but if we are dealing with people or organizations at least the questions will be similar.

Happy to help.

Hi @Guy Vorster,

Thank you for the response, it helped for sure. 
I was able to setup the MCP server using Claude, but the response is not  great.
Simple search/get prompts are not giving results except the one get details using entity ID.

Would be great if you could share any content to better understand or customize the solution. 
Appreciate your inputs.

Hi Vinoth, did you manage to look through this post shared Sergio Abraham on the Reltio MCP Server?

I have been playing  / hacking around with the MCP Server so can give you my take on it so long but am interested in what others have to say too.

I'm using the Claude desktop app on a Mac just so you know.  After install, the MCP Server is exposed in Claude under Settings -> Developer. If you click on the "reltio-mcp-server", it should show a state of "running". If not you may need to troubleshoot a bit based on looking at the logs.

The Reltio MCP Server exposes a series of "tools" (written in python) that act as wrappers to various Reltio API's (search entity, get entity, get matches etc).  By including markup in these python files along with the actual API calls, the LLM is able to intelligently figure out which tools to use based on your prompt.  It's actually very easy to extend the MCP Server with your own tools if you want to as part of learning how this works.  I have very little experience with python myself but with the help of the Claude app, I was able to generate some python very easily to add additional tools to the MCP Server.  However you certainly don't have to add additional tools in order to use the MCP Server as it is. 

Once it's running just start asking it some questions such as "search for a person with a phone number of bla bla" or "fetch the details about entity id bla bla".  It's amazing to watch how Claude will take your prompt and then use the tools (assuming your MCP Server is running of course). It may even try various tools if it doesn't get a result with one tool.  I've been stunned at times just to watch what it does and how it returns information.  Improved prompting will render better results so try different prompts and see how your output varies.  If you're interested, I can share a short video too about some new tools I added to my local version of the MCP Server in case you would like to try that too.

This is such a fascinating area and as we all know it's evolving almost daily. Let's all help each other get up to speed quickly and keep improving our knowledge in this area because this MCP Server concept coupled with the power of LLM's is an amazing capability!

Hi All,

I am trying to understand or get more information on the below items, would really appreciate if one can share some light into these areas.

1. How the recently introduced MCP server integrates Reltio and other AI tools/applications.
2. Does the MCP server uses  Reltio APIs to facilitate the integration between the two sides. 
3. Does the MCP server translates any prompts from AI tools to construct the response.
4. What is the level of details can be made available to AI tools/apps using the MCP server.
5. Apart from the MCP server, are there any options to integrate MDM with Gen AI/Open AI tools/apps.

Thanks!

Hi @Guy Vorster,

Yes, I can see the log with error "Search by ov is not  available for tenant:" I was looking at another log file, that didn't have enough to debug. I'll reach out to enable this for the tenant.

Thank you for all the help and inputs!

OK Vinoth I suspect it may be because your tenant does not have "search by ov" enabled.  In order to confirm this, please can you check in your claude logs. On a Mc they are in your Library folder, then Logs, then Claude.  You should see a file (likely more than one) with the name of your reltio mcp server.  Scroll down and see if you can see search errors.  Feel free to attach your log file so I can have a look

Hi @Guy Vorster,

Adding the sample responses from Hub and Claude.

Thanks for helping out.

Hi Vinoth. I just tried the query against my demo tenant "Search for Individual profiles from city Boston" which is exactly the same as your query except for using Boston vs Chennai. I received results back from my search without any problem so am interested to learn more about why you're not seeing any results.  Perhaps share a screenshot of the faceted search from your reltio tenant with Chennai as the selected city and then share a screenshot of what you get in Claude via the MCP Server.

Thanks for sharing these examples Vinoth. Let me try similar ones in my tenant and I will report back.  One question though; are you getting errors in the mcp server or just zero results for all of these?  thanks

Hi @Guy Vorster,

Am using a partner tenant, that has Individual and Org customer data.
I was using the sample prompts below ,

1. Search for Individual profiles from city Chennai
2. Search for individual profiles with First Name Vinoth
3. Get the list of profiles without Last Name

Thanks again for helping out.

Hey Vinoth

Give me a bit more context about the tenant you're using (what type of data, what industry) and send a few examples of the prompts you were using (feel free to sanitize them a bit if needed).  I can try them on my side (if they involve people and or organizations) and see how they work for me.  Granted we will have different data but if we are dealing with people or organizations at least the questions will be similar.

Happy to help.

Hi @Guy Vorster,

Thank you for the response, it helped for sure. 
I was able to setup the MCP server using Claude, but the response is not  great.
Simple search/get prompts are not giving results except the one get details using entity ID.

Would be great if you could share any content to better understand or customize the solution. 
Appreciate your inputs.

Hi Vinoth, did you manage to look through this post shared Sergio Abraham on the Reltio MCP Server?

I have been playing  / hacking around with the MCP Server so can give you my take on it so long but am interested in what others have to say too.

I'm using the Claude desktop app on a Mac just so you know.  After install, the MCP Server is exposed in Claude under Settings -> Developer. If you click on the "reltio-mcp-server", it should show a state of "running". If not you may need to troubleshoot a bit based on looking at the logs.

The Reltio MCP Server exposes a series of "tools" (written in python) that act as wrappers to various Reltio API's (search entity, get entity, get matches etc).  By including markup in these python files along with the actual API calls, the LLM is able to intelligently figure out which tools to use based on your prompt.  It's actually very easy to extend the MCP Server with your own tools if you want to as part of learning how this works.  I have very little experience with python myself but with the help of the Claude app, I was able to generate some python very easily to add additional tools to the MCP Server.  However you certainly don't have to add additional tools in order to use the MCP Server as it is. 

Once it's running just start asking it some questions such as "search for a person with a phone number of bla bla" or "fetch the details about entity id bla bla".  It's amazing to watch how Claude will take your prompt and then use the tools (assuming your MCP Server is running of course). It may even try various tools if it doesn't get a result with one tool.  I've been stunned at times just to watch what it does and how it returns information.  Improved prompting will render better results so try different prompts and see how your output varies.  If you're interested, I can share a short video too about some new tools I added to my local version of the MCP Server in case you would like to try that too.

This is such a fascinating area and as we all know it's evolving almost daily. Let's all help each other get up to speed quickly and keep improving our knowledge in this area because this MCP Server concept coupled with the power of LLM's is an amazing capability!

Hi All,

I am trying to understand or get more information on the below items, would really appreciate if one can share some light into these areas.

1. How the recently introduced MCP server integrates Reltio and other AI tools/applications.
2. Does the MCP server uses  Reltio APIs to facilitate the integration between the two sides. 
3. Does the MCP server translates any prompts from AI tools to construct the response.
4. What is the level of details can be made available to AI tools/apps using the MCP server.
5. Apart from the MCP server, are there any options to integrate MDM with Gen AI/Open AI tools/apps.

Thanks!

Yup that's what I suspected was the error.  This is good to know.  Unfortunately it means the search won't work until search by ov is enabled in that tenant. can you send me your tenant id please?

Hi @Guy Vorster,

Yes, I can see the log with error "Search by ov is not  available for tenant:" I was looking at another log file, that didn't have enough to debug. I'll reach out to enable this for the tenant.

Thank you for all the help and inputs!

OK Vinoth I suspect it may be because your tenant does not have "search by ov" enabled.  In order to confirm this, please can you check in your claude logs. On a Mc they are in your Library folder, then Logs, then Claude.  You should see a file (likely more than one) with the name of your reltio mcp server.  Scroll down and see if you can see search errors.  Feel free to attach your log file so I can have a look

Hi @Guy Vorster,

Adding the sample responses from Hub and Claude.

Thanks for helping out.

Hi Vinoth. I just tried the query against my demo tenant "Search for Individual profiles from city Boston" which is exactly the same as your query except for using Boston vs Chennai. I received results back from my search without any problem so am interested to learn more about why you're not seeing any results.  Perhaps share a screenshot of the faceted search from your reltio tenant with Chennai as the selected city and then share a screenshot of what you get in Claude via the MCP Server.

Thanks for sharing these examples Vinoth. Let me try similar ones in my tenant and I will report back.  One question though; are you getting errors in the mcp server or just zero results for all of these?  thanks

Hi @Guy Vorster,

Am using a partner tenant, that has Individual and Org customer data.
I was using the sample prompts below ,

1. Search for Individual profiles from city Chennai
2. Search for individual profiles with First Name Vinoth
3. Get the list of profiles without Last Name

Thanks again for helping out.

Hey Vinoth

Give me a bit more context about the tenant you're using (what type of data, what industry) and send a few examples of the prompts you were using (feel free to sanitize them a bit if needed).  I can try them on my side (if they involve people and or organizations) and see how they work for me.  Granted we will have different data but if we are dealing with people or organizations at least the questions will be similar.

Happy to help.

Hi @Guy Vorster,

Thank you for the response, it helped for sure. 
I was able to setup the MCP server using Claude, but the response is not  great.
Simple search/get prompts are not giving results except the one get details using entity ID.

Would be great if you could share any content to better understand or customize the solution. 
Appreciate your inputs.

Hi Vinoth, did you manage to look through this post shared Sergio Abraham on the Reltio MCP Server?

I have been playing  / hacking around with the MCP Server so can give you my take on it so long but am interested in what others have to say too.

I'm using the Claude desktop app on a Mac just so you know.  After install, the MCP Server is exposed in Claude under Settings -> Developer. If you click on the "reltio-mcp-server", it should show a state of "running". If not you may need to troubleshoot a bit based on looking at the logs.

The Reltio MCP Server exposes a series of "tools" (written in python) that act as wrappers to various Reltio API's (search entity, get entity, get matches etc).  By including markup in these python files along with the actual API calls, the LLM is able to intelligently figure out which tools to use based on your prompt.  It's actually very easy to extend the MCP Server with your own tools if you want to as part of learning how this works.  I have very little experience with python myself but with the help of the Claude app, I was able to generate some python very easily to add additional tools to the MCP Server.  However you certainly don't have to add additional tools in order to use the MCP Server as it is. 

Once it's running just start asking it some questions such as "search for a person with a phone number of bla bla" or "fetch the details about entity id bla bla".  It's amazing to watch how Claude will take your prompt and then use the tools (assuming your MCP Server is running of course). It may even try various tools if it doesn't get a result with one tool.  I've been stunned at times just to watch what it does and how it returns information.  Improved prompting will render better results so try different prompts and see how your output varies.  If you're interested, I can share a short video too about some new tools I added to my local version of the MCP Server in case you would like to try that too.

This is such a fascinating area and as we all know it's evolving almost daily. Let's all help each other get up to speed quickly and keep improving our knowledge in this area because this MCP Server concept coupled with the power of LLM's is an amazing capability!

Hi All,

I am trying to understand or get more information on the below items, would really appreciate if one can share some light into these areas.

1. How the recently introduced MCP server integrates Reltio and other AI tools/applications.
2. Does the MCP server uses  Reltio APIs to facilitate the integration between the two sides. 
3. Does the MCP server translates any prompts from AI tools to construct the response.
4. What is the level of details can be made available to AI tools/apps using the MCP server.
5. Apart from the MCP server, are there any options to integrate MDM with Gen AI/Open AI tools/apps.

Thanks!

Hi @Guy Vorster, 

I have enabled the configuration to "Search By Ov" and am getting some results for the same prompts now, I'll explore and reach out incase of any other questions.  You mentioned you have added few custom actions to the MCP server, would you be able to share some details on adding custom tools to the model.

Really appreciate your support.

Yup that's what I suspected was the error.  This is good to know.  Unfortunately it means the search won't work until search by ov is enabled in that tenant. can you send me your tenant id please?

Hi @Guy Vorster,

Yes, I can see the log with error "Search by ov is not  available for tenant:" I was looking at another log file, that didn't have enough to debug. I'll reach out to enable this for the tenant.

Thank you for all the help and inputs!

OK Vinoth I suspect it may be because your tenant does not have "search by ov" enabled.  In order to confirm this, please can you check in your claude logs. On a Mc they are in your Library folder, then Logs, then Claude.  You should see a file (likely more than one) with the name of your reltio mcp server.  Scroll down and see if you can see search errors.  Feel free to attach your log file so I can have a look

Hi @Guy Vorster,

Adding the sample responses from Hub and Claude.

Thanks for helping out.

Hi Vinoth. I just tried the query against my demo tenant "Search for Individual profiles from city Boston" which is exactly the same as your query except for using Boston vs Chennai. I received results back from my search without any problem so am interested to learn more about why you're not seeing any results.  Perhaps share a screenshot of the faceted search from your reltio tenant with Chennai as the selected city and then share a screenshot of what you get in Claude via the MCP Server.

Thanks for sharing these examples Vinoth. Let me try similar ones in my tenant and I will report back.  One question though; are you getting errors in the mcp server or just zero results for all of these?  thanks

Hi @Guy Vorster,

Am using a partner tenant, that has Individual and Org customer data.
I was using the sample prompts below ,

1. Search for Individual profiles from city Chennai
2. Search for individual profiles with First Name Vinoth
3. Get the list of profiles without Last Name

Thanks again for helping out.

Hey Vinoth

Give me a bit more context about the tenant you're using (what type of data, what industry) and send a few examples of the prompts you were using (feel free to sanitize them a bit if needed).  I can try them on my side (if they involve people and or organizations) and see how they work for me.  Granted we will have different data but if we are dealing with people or organizations at least the questions will be similar.

Happy to help.

Hi @Guy Vorster,

Thank you for the response, it helped for sure. 
I was able to setup the MCP server using Claude, but the response is not  great.
Simple search/get prompts are not giving results except the one get details using entity ID.

Would be great if you could share any content to better understand or customize the solution. 
Appreciate your inputs.

Hi Vinoth, did you manage to look through this post shared Sergio Abraham on the Reltio MCP Server?

I have been playing  / hacking around with the MCP Server so can give you my take on it so long but am interested in what others have to say too.

I'm using the Claude desktop app on a Mac just so you know.  After install, the MCP Server is exposed in Claude under Settings -> Developer. If you click on the "reltio-mcp-server", it should show a state of "running". If not you may need to troubleshoot a bit based on looking at the logs.

The Reltio MCP Server exposes a series of "tools" (written in python) that act as wrappers to various Reltio API's (search entity, get entity, get matches etc).  By including markup in these python files along with the actual API calls, the LLM is able to intelligently figure out which tools to use based on your prompt.  It's actually very easy to extend the MCP Server with your own tools if you want to as part of learning how this works.  I have very little experience with python myself but with the help of the Claude app, I was able to generate some python very easily to add additional tools to the MCP Server.  However you certainly don't have to add additional tools in order to use the MCP Server as it is. 

Once it's running just start asking it some questions such as "search for a person with a phone number of bla bla" or "fetch the details about entity id bla bla".  It's amazing to watch how Claude will take your prompt and then use the tools (assuming your MCP Server is running of course). It may even try various tools if it doesn't get a result with one tool.  I've been stunned at times just to watch what it does and how it returns information.  Improved prompting will render better results so try different prompts and see how your output varies.  If you're interested, I can share a short video too about some new tools I added to my local version of the MCP Server in case you would like to try that too.

This is such a fascinating area and as we all know it's evolving almost daily. Let's all help each other get up to speed quickly and keep improving our knowledge in this area because this MCP Server concept coupled with the power of LLM's is an amazing capability!

Hi All,

I am trying to understand or get more information on the below items, would really appreciate if one can share some light into these areas.

1. How the recently introduced MCP server integrates Reltio and other AI tools/applications.
2. Does the MCP server uses  Reltio APIs to facilitate the integration between the two sides. 
3. Does the MCP server translates any prompts from AI tools to construct the response.
4. What is the level of details can be made available to AI tools/apps using the MCP server.
5. Apart from the MCP server, are there any options to integrate MDM with Gen AI/Open AI tools/apps.

Thanks!

Yes sure.  for example, assume you want to add a new tool to the mcp server to be able to return the graph of relationships surrounding any entity (this would be using the _hops api which is what is used in the grah view of reltio ui).  firstly you would modify the server.py file and add this python code in (this is defining the new mcp tool):

@mcp.tool()
async def get_entity_graph_tool(entity_id: str, tenant_id: str = RELTIO_TENANT) -> dict:
    """Get detailed information about the graph of related entities surrounding a Reltio entity by ID

    Args:
        entity_id (str): The ID of the entity to retrieve
        tenant_id (str): Tenant ID for the Reltio environment. Defaults to RELTIO_TENANT env value.

    Returns:
        A dictionary containing the related entities details as a graph form

    Raises:
        Exception: If there's an error getting the related entity details

    Examples:
        # Get entity details by ID
        get_entity_graph("entity_id", "tenant_id")
    """
    return await get_entity_graph(entity_id, tenant_id)

The easiest option then is to edit the file called entity.py and add the function call from below that makes the actual API call to Reltio (example below):

async def get_entity_graph(entity_id: str, tenant_id: str = RELTIO_TENANT) -> dict:
    """Get detailed information about the graph of related entities surrounding a Reltio entity by ID

    Args:
        entity_id (str): The ID of the entity to retrieve
        tenant_id (str): Tenant ID for the Reltio environment. Defaults to RELTIO_TENANT env value.

    Returns:
        A dictionary containing the related entities details as a graph form

    Raises:
        Exception: If there's an error getting the related entities details
    """
    try:
        # Validate inputs using Pydantic model
        try:
            request = EntityIdRequest(
                entity_id=entity_id,
                tenant_id=tenant_id
            )
        except ValueError as e:
            logger.warning(f"Validation error in get_entity_graph: {str(e)}")
            return create_error_response(
                "VALIDATION_ERROR",
                f"Invalid entity ID format: {str(e)}"
            )

        # Construct URL with validated entity ID
        url = get_reltio_url(f"entities/{request.entity_id}/_hops?deep=1&select=uri,type,label", request.tenant_id)

        try:
            headers = get_reltio_headers()

            # Validate connection security
            validate_connection_security(url, headers)
        except Exception as e:
            logger.error(f"Authentication or security error: {str(e)}")
            return create_error_response(
                "AUTHENTICATION_ERROR",
                "Failed to authenticate with Reltio API"
            )

        # Make the request with timeout
        try:
            entity_graph = http_request(url, headers=headers)
        except Exception as e:
            logger.error(f"API request error: {str(e)}")

            # Check if it's a 404 error (relation not found)
            if "404" in str(e):
                return create_error_response(
                    "RESOURCE_NOT_FOUND",
                    f"Entity with ID {request.entity_id} not found"
                )

            return create_error_response(
                "SERVER_ERROR",
                "Failed to retrieve entity graph details from Reltio API"
            )

        return entity_graph
    except Exception as e:
        # Log the error
        logger.error(f"Unexpected error in get_entity_graph: {str(e)}")

        # Return a sanitized error response
        return create_error_response(
            "SERVER_ERROR",
            "An unexpected error occurred while retrieving the entity graph details"
        )

An additional step which you now need to do (back in server.py file) is to include that new function in your import statement.  For example if you added that function call to entity.py then look for the import statement that looks like this "from src.tools.entity import" and add get_entity_graph at the end (after adding a comma).

Then quit claude completely and rerun setup.sh.  Upon opening claude you should see that new tool in the mcp server!  make sure you have a record that has a bunch of relationships to / from it and then try it out. just ask claude to fetch the relationships for entity bla bla.

let me know if you run into problems.

Hi @Guy Vorster, 

I have enabled the configuration to "Search By Ov" and am getting some results for the same prompts now, I'll explore and reach out incase of any other questions.  You mentioned you have added few custom actions to the MCP server, would you be able to share some details on adding custom tools to the model.

Really appreciate your support.

Yup that's what I suspected was the error.  This is good to know.  Unfortunately it means the search won't work until search by ov is enabled in that tenant. can you send me your tenant id please?

Hi @Guy Vorster,

Yes, I can see the log with error "Search by ov is not  available for tenant:" I was looking at another log file, that didn't have enough to debug. I'll reach out to enable this for the tenant.

Thank you for all the help and inputs!

OK Vinoth I suspect it may be because your tenant does not have "search by ov" enabled.  In order to confirm this, please can you check in your claude logs. On a Mc they are in your Library folder, then Logs, then Claude.  You should see a file (likely more than one) with the name of your reltio mcp server.  Scroll down and see if you can see search errors.  Feel free to attach your log file so I can have a look

Hi @Guy Vorster,

Adding the sample responses from Hub and Claude.

Thanks for helping out.

Hi Vinoth. I just tried the query against my demo tenant "Search for Individual profiles from city Boston" which is exactly the same as your query except for using Boston vs Chennai. I received results back from my search without any problem so am interested to learn more about why you're not seeing any results.  Perhaps share a screenshot of the faceted search from your reltio tenant with Chennai as the selected city and then share a screenshot of what you get in Claude via the MCP Server.

Thanks for sharing these examples Vinoth. Let me try similar ones in my tenant and I will report back.  One question though; are you getting errors in the mcp server or just zero results for all of these?  thanks

Hi @Guy Vorster,

Am using a partner tenant, that has Individual and Org customer data.
I was using the sample prompts below ,

1. Search for Individual profiles from city Chennai
2. Search for individual profiles with First Name Vinoth
3. Get the list of profiles without Last Name

Thanks again for helping out.

Hey Vinoth

Give me a bit more context about the tenant you're using (what type of data, what industry) and send a few examples of the prompts you were using (feel free to sanitize them a bit if needed).  I can try them on my side (if they involve people and or organizations) and see how they work for me.  Granted we will have different data but if we are dealing with people or organizations at least the questions will be similar.

Happy to help.

Hi @Guy Vorster,

Thank you for the response, it helped for sure. 
I was able to setup the MCP server using Claude, but the response is not  great.
Simple search/get prompts are not giving results except the one get details using entity ID.

Would be great if you could share any content to better understand or customize the solution. 
Appreciate your inputs.

Hi Vinoth, did you manage to look through this post shared Sergio Abraham on the Reltio MCP Server?

I have been playing  / hacking around with the MCP Server so can give you my take on it so long but am interested in what others have to say too.

I'm using the Claude desktop app on a Mac just so you know.  After install, the MCP Server is exposed in Claude under Settings -> Developer. If you click on the "reltio-mcp-server", it should show a state of "running". If not you may need to troubleshoot a bit based on looking at the logs.

The Reltio MCP Server exposes a series of "tools" (written in python) that act as wrappers to various Reltio API's (search entity, get entity, get matches etc).  By including markup in these python files along with the actual API calls, the LLM is able to intelligently figure out which tools to use based on your prompt.  It's actually very easy to extend the MCP Server with your own tools if you want to as part of learning how this works.  I have very little experience with python myself but with the help of the Claude app, I was able to generate some python very easily to add additional tools to the MCP Server.  However you certainly don't have to add additional tools in order to use the MCP Server as it is. 

Once it's running just start asking it some questions such as "search for a person with a phone number of bla bla" or "fetch the details about entity id bla bla".  It's amazing to watch how Claude will take your prompt and then use the tools (assuming your MCP Server is running of course). It may even try various tools if it doesn't get a result with one tool.  I've been stunned at times just to watch what it does and how it returns information.  Improved prompting will render better results so try different prompts and see how your output varies.  If you're interested, I can share a short video too about some new tools I added to my local version of the MCP Server in case you would like to try that too.

This is such a fascinating area and as we all know it's evolving almost daily. Let's all help each other get up to speed quickly and keep improving our knowledge in this area because this MCP Server concept coupled with the power of LLM's is an amazing capability!

Hi All,

I am trying to understand or get more information on the below items, would really appreciate if one can share some light into these areas.

1. How the recently introduced MCP server integrates Reltio and other AI tools/applications.
2. Does the MCP server uses  Reltio APIs to facilitate the integration between the two sides. 
3. Does the MCP server translates any prompts from AI tools to construct the response.
4. What is the level of details can be made available to AI tools/apps using the MCP server.
5. Apart from the MCP server, are there any options to integrate MDM with Gen AI/Open AI tools/apps.

Thanks!