ShapersOfAI.com · From Users to Shapers · AI Leadership · Future of Work · AI Skills

AI Use Case

Intranet News Publication Agent

Turns approved briefings and source materials into structured intranet news drafts, including headlines, teasers, article text and metadata.

Creating an intranet news article involves more than writing a text. Editors must review the briefing and source documents, identify the relevant facts, structure the story, formulate a headline and teaser, apply editorial guidelines and prepare the required metadata.

An Intranet News Publication Agent supports this process from briefing to publication-ready draft. Using approved source material, it creates the article text and the accompanying elements required by the intranet, such as headline alternatives, summary, category, keywords, image suggestion and review date.

In a more advanced implementation, the agent can transfer the approved content into a SharePoint news template and prepare the page for editorial review. The editor remains responsible for checking facts, selecting the final wording and approving publication.

The agent is not intended to research facts independently or publish content without human review. Its purpose is to make routine intranet production faster and more consistent.

Painpoints

Intranet articles are created through many repetitive manual steps
Briefings and source materials often arrive in different formats
Editors spend time extracting the relevant facts
Headlines, teasers and metadata are frequently created separately
Editorial quality varies between authors and departments
Important context or mandatory information can be overlooked
Content is sometimes copied manually into SharePoint
Publishing standards are not consistently applied
Routine articles require disproportionate editorial effort
Decentralised editors may have limited writing experience

Goals

Reduce the time required to create intranet news
Turn approved information into a complete article draft
Apply common editorial and structural standards
Produce headlines, teasers and metadata together with the article
Reduce repetitive copy-and-paste work
Support decentralised intranet editors
Improve consistency across departments and locations
Preserve facts and mandatory messages from the source material
Prepare content directly in the required SharePoint structure
Keep final editorial control with a human editor

Risks

The agent may add facts not contained in the approved sources
Important nuances may be lost during summarisation
Contradictory source documents may be combined incorrectly
Generated headlines may overstate the content
Incorrect metadata could reduce findability
Confidential information might be included unintentionally
Editorial guidelines may be applied too mechanically
Automated SharePoint page creation could use the wrong template or location
Editors may approve drafts without checking the original sources
Automatic publication could bypass necessary approvals

Implementation Idea

Advatera Community Examples

Define the intranet article types the agent should support, such as news, project updates, event announcements and leadership messages
Create a structured briefing form containing:
- Communication objective
- Target audience
- Key message
- Confirmed facts
- Required quotes
- Relevant dates
- Contact person
- Source documents
- Publication date
- Confidentiality or approval requirements
Store editorial guidelines, terminology, approved examples and page templates in SharePoint
Generate a reusable Intranet News Writing Skill containing the editorial rules and output structure
Use this skill directly in ChatGPT or Claude for a simple pilot
Implement the same instructions and knowledge sources as a Copilot Studio agent for use across the communications team
Restrict the agent to the briefing and attached approved sources for each article
Require the agent to flag missing or contradictory information before drafting
Instruct it never to complete missing facts from general AI knowledge
Generate the following output in one run:
- Three headline options
- Short teaser
- Structured article
- Suggested subheadings
- Relevant keywords
- Content category
- Suggested review date
- Image or visual recommendation
- List of missing information
Mark every quote, figure and date so the editor can verify it against the source
Add a second quality-control step that checks the draft against the briefing and editorial guidelines
Let an editor revise and approve the draft
Use Power Automate to transfer the approved fields into a SharePoint news-page template
Create the SharePoint page as a draft, not as a published page
Route sensitive articles to additional legal, HR or management approval
Publish only after explicit human approval
Save the final article, source briefing and approval status for documentation
Use feedback on edited drafts to improve the skill and examples over time

Practical Microsoft implementation

A practical Microsoft setup could combine:

Microsoft Forms, Power Apps or SharePoint Lists for the structured briefing
SharePoint for guidelines, terminology, examples and approved source documents
Copilot Studio for the shared publication agent
Power Automate for review, approval and transfer into SharePoint
SharePoint page templates for consistent layouts
Microsoft Graph or SharePoint APIs to create draft news pages
Teams as an optional interface for editors

A first pilot should focus on one predictable article type, such as event announcements or project updates. The agent should initially create a structured draft only. Automated SharePoint page creation can follow once the editorial output is reliable.