How to retrieve Google search results into Mendix and evaluate them in a microflow (compliance screening use case)?

0
Hi all,I'm building a customer screening feature in my Mendix app (Studio Pro 10.x) and would like advice on the best approach.Use case:1. The app stores a Customer Name (e.g., "Hua Hong Semiconductor (Wuxi) Ltd").2. For screening, the customer name must be searched on Google together with predefined keywords from the database (e.g., "Huawei", "Hisilicon").3. Based on the search results, the app should record a conclusion per keyword: True Hit (genuine relationship found), False Hit (keyword appears but in an unrelated context), or No Match.What I have working so far:- A nanoflow with an Open URL action that builds the search URL: 'https://www.google.com/search?q=' + urlEncode($SearchRequest/CustomerName + ' ' + 'Huawei')- This opens Google in a new tab, but the user has to review the results manually and type the conclusion back into the app.What I want to achieve:- Remove the manual step: have Mendix retrieve the search results itself (via Call REST?) and auto-populate the conclusion and supporting URL.My questions:1. Is the Google Custom Search JSON API the recommended way to do this from a Call REST activity, or is there a marketplace module that already wraps this?2. Has anyone implemented snippet-level evaluation in a microflow (checking if both the company name and keyword occur in the same result snippet)? Any pitfalls with import mappings for the Custom Search response?3. For context-based classification (distinguishing a genuine relationship from coincidental co-occurrence), has anyone integrated an LLM step (e.g., Azure OpenAI) after the search call? Which connector did you use?Mendix version: Studio Pro 10.xAny pointers, marketplace modules, or example implementations would be much appreciated. Thanks!
asked
1 answers
0

Hi Yogeshwar,


To respond to your questions:


1. Is the Google Custom Search JSON API the recommended way to do this?

Not for a new implementation. Google has already announced that the Custom Search JSON API is closed to new customers and will be retired in 2027. I would avoid designing a new solution around a service that is already being phased out.


2. Has anyone implemented snippet-level evaluation in a microflow?

From an architectural perspective, I would use snippet-level evaluation as a filtering step rather than as the final decision criteria.


The main concern is false positives. A result may contain both the customer name and keyword, but that alone does not prove there is a meaningful relationship between them.


As long as the provider returns a consistent JSON structure, I would expect the import mapping to be relatively straightforward. The bigger challenge is defining reliable business rules for determining what qualifies as a true match.


3. Has anyone integrated an LLM step for context-based classification?

This is the approach I would recommend for the cases that remain after the initial filtering. Instead of checking only whether both terms appear in the result, an LLM can help evaluate whether the content actually indicates a relevant relationship.


Mendix already provides OpenAI/Azure OpenAI connectors, making this type of integration relatively straightforward. I would also request a confidence score and brief justification so reviewers can understand why a result was classified a certain way.


My recommendation would be:

  1. Retrieve candidate results from a search provider.
  2. Use simple rules to remove obvious false positives.
  3. Use an LLM to evaluate the remaining candidates and provide justification.


I hope this helps answer your question. If you need any additional details, feel free to reply below.


Best regards,

Oswaldo Valles

answered