Agent Name: AI News Agent
AI News is an intelligent information summarization agent designed to help executives and decision-makers stay on top of the rapidly evolving world of artificial intelligence. By continuously scanning credible news sources, it identifies and summarizes the most impactful AI developmentsβespecially those relevant to the Microsoft ecosystem, healthcare, and legal industries.
The agent reduces information overload by prioritizing high-value updates and presenting them in a clear, digestible Markdown table format. Each summary includes a headline, date, key insight, strategic implications, and a verified source link.
AI News is not a replacement for human analysis or legal due diligence. Instead, it augments executive awareness by surfacing credible developments that may require further attention or discussion. All strategic decisions and compliance assessments remain the responsibility of the end user.
- Daily Team Briefing: Generate a morning digest of AI developments for Microsoft product teams.
- Industry Watch: Provide Healthcare or Legal compliance teams with up-to-the-minute AI news.
- Executive Snapshot: Deliver a concise AI highlights report to executives.
- Searches Bing News via the BingGroundedSearchTool for the latest AI developments (last 24 hours PT)
- Filters by publication date and credible domains (e.g., gartner.com, reuters.com, wired.com, mit.edu, stanford.edu)
- Prioritizes items in this order: Microsoft+Healthcare β Microsoft+Legal β Healthcare β Legal
- Summarizes each development and outlines implications for the three focus areas
- Outputs in a Markdown table with columns: Headline, Date, Summary, Implications, Link
The agent follows a system prompt that enforces:
- Source filtering against a predefined list of credible domains
- Publication date constraint within the 24βhour window
- Priority order for topic groups
- Consistent Markdown formatting for the output table
- Description: Queries Bing News for AI-related articles
- Auth: Configured via
BING_SEARCH_KEYenvironment variable (e.g., stored in Azure Key Vault) - Reference: OpenAPI schema provided in
tools/bing-grounded-search-schema.yaml
- "What are the most impactful AI developments for Microsoft in the last 24 hours?"
- "Summarize recent legal sector AI news."
- "Healthcare-related AI breakthroughs today?"
This agent uses Azure AI Projects with the BingGroundedSearchTool to query Bing News, applies filtering and prioritization logic in Python, summarizes via Azure OpenAI, and formats output as Markdown.
Refer to demo_ai_news_usage.ipynb for a guided walkthrough.
git clone https://github.com/your-org/azure-ai-agent-service-blueprints.git
cd azure-ai-agent-service-blueprints/ai-newsThis Bicep script will:
- Reference your existing Azure AI Agent Service project and model deployment
- Create a Bing Grounding resource and connect it to your Cognitive Services account
- Provision a Storage Account, App Service Plan, and Function App
Run:
az deployment group create \
--resource-group <rg> \
--template-file deploy.bicep \
--parameters functionAppName=aiNewsApp account_name=<your-cognitive-account-name>- Replace
<your-cognitive-account-name>with the name of your existing Cognitive Services account. - The script creates a Bing Grounding account named
bingsearch-<your-cognitive-account-name>and a connection.
In the project root, create .env containing:
PROJECT_ENDPOINT=<your-ai-agent-service-endpoint>
MODEL_DEPLOYMENT_NAME=<your-openai-model-deployment-name>
BING_CONNECTION_ID=<leave blank; Function App app setting will be injected>python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txtRefer to the main README for your prompt content.
python template.py- (Optional) Review and modify configuration in
template.py.
| Envar | Description |
|---|---|
PROJECT_ENDPOINT |
Azure AI Agent Service project endpoint |
MODEL_DEPLOYMENT_NAME |
Name of the OpenAI model deployment (e.g., gpt-4o-mini) |
BING_CONNECTION_ID |
Resource ID of the Bing Grounding connection (injected in FUNC) |
USE_SAMPLE |
Set to true to use local sample JSON instead of live calls |
- No input data is required; the agent runs on schedule and logs or prints the output.
User: What are todayβs top AI developments for Microsoft and healthcare?
Agent: | Headline | Date | Summary | Implications | Link |
|βββββ|βββ|βββ|ββββββ|β|
| ... | ... | ... | ... | ... |
User: Summarize the latest legal AI news.
Agent: | Headline | Date | Summary | Implications | Link | ... |
User: Run the AI Developments Aggregation Agent now.
Agent: (returns a Markdown table of up to 10 items)
- To route output to other targets (e.g., email, Teams), wrap the final table in a notification action.
- To change the timezone from Pacific, review the # Compute time window (Pacific) section and make the necessary adjustments
Marquee Insights is a specialized AI consultancy delivering cutting-edge solutions. Visit Marquee Insights.
For support, contact [email protected]
MIT License