Getting Results with AI Agents + Bing Grounding

With the rise of Generative AI technologies, web search has never been more important—or more overwhelming. Modern search engines like Bing excel at delivering vast amounts of information quickly, but quantity alone doesn’t always guarantee the best possible insights. Enter the AI Agent Bing Grounding search tool: a specialized tool that uses Bing to refine, curate, and tailor results for easy consumption. In this post, we’ll explore what makes the Bing Grounding tool special, and how curated results can transform the way you use search results in your application.

Taming the Overload with Curated Results

It’s no secret that information overload is a real challenge in today’s fast-moving digital landscape. Even casual browsing can feel time-consuming when sifting through pages of search results. The Bing Grounding search tool can help address this by:

• Combining the power of Bing’s broad indexing with refined filters and advanced algorithms.
• Pinpointing the most relevant sources and weeding out low-quality links.
• Offering a concise generated answer with a curated list of results—so you can focus on what matters, faster.

Saving Time

Conventional search approaches often require you to list results and for your user to skim through them, and decide what’s valuable. That manual effort eats into productivity. By contrast, the Bing Grounding tool adds an efficient workflow:

• Auto-summarizing the relevant sections from your results.
• Prioritizing the most important points and providing citations—so you can quickly decide whether the resource is worth a deeper dive.

Increasing Reliability and Consistency

Curated search results foster trust over time. By consistently showing links from reputable sources and verified authorities, the Bing Grounding tool sets itself apart:

• Building credibility for the user by avoiding spam, clickbait, or misleading articles.
• Bolstering user confidence with each successful result, creating a reliable research companion no matter your industry or field.

How it works:

The AI Agent Service deployed to our Azure AI Foundry uses a Bing Grounding Tool to perform Bing searches and return results in a summarized text with relevant source citations.  We use a simple Fast API server to serve user requests and execute the AI Agent Bing Tool in Foundry.

Using Swagger, we can submit API requests with a 'query' parameter:

Example curl:

curl -X 'GET' \ 'http://127.0.0.1:8000/search?query=msft%20stock' \ -H 'accept: application/json'

And retrieve the curated results from the Agent:

{ "response": { "type": "text", "text": { "value": "On April 4, 2025, the closing price of Microsoft (MSFT) stock was approximately $389.97【3:0†source】.", "annotations": [ { "type": "url_citation", "text": "【3:0†source】", "start_index": 87, "end_index": 99, "url_citation": { "url": "https://www.marketbeat.com/stocks/NASDAQ/MSFT/forecast/", "title": "Microsoft (MSFT) Stock Forecast and Price Target 2025 - MarketBeat" } } ] } } }

As you can see the summarized answer and source citations offer easy integration into applications looking to leverage up-to-date web search results.

Conclusion

This new approach signifies a shift in how we utilize search results to answer user questions and enhance application features. In a landscape where Generative AI acts as a force multiplier for content and digital information, the AI Agent Bing Grounding tool serves as a straightforward solution for providing relevant, summarized answers grounded in truth from reputable sources. Imagine it as having your own personal research assistant who not only gathers information but also curates it into a concise memo with all the appropriate references, rather than simply dropping a pile of papers on your desk.

Links

AI Foundry - What is Azure AI Foundry? - Azure AI Foundry | Microsoft Learn

AI Agent Service- What is Azure AI Agent Service? - Azure AI services | Microsoft Learn

Code Repository - jakeatmsft/bing_grounding_api