SharePoint AI Skills: What Works & Fails

Key insights

• Skills
  Skills are reusable, plain-text Markdown workflows that teach AI in SharePoint how to perform repeatable, multi-step tasks. They let teams capture procedures once and run them consistently across a site.
• Storage & authoring
  Skills are stored in the Agent Assets library as Markdown files. You create a skill from AI Chat with a natural-language prompt, save the draft, and edit it in the browser if needed.
• Hands-on use cases
  Preview demos include generating a pre-meeting briefing, checking project folder completeness with metadata write-back, and running a vendor insurance audit that creates and populates a SharePoint list.
• Permissions & governance
  Skills inherit SharePoint permissions, retention, sensitivity labels, and auditing. Users need Edit access to create skills and View access to run them, which supports compliance while allowing broad use.
• Current limitations
  The preview cannot call external systems or run custom code, cannot read permission or sharing details, and fails to create Planner tasks; the Markdown editor and governance controls also need improvement.
• Architectural questions
  AI in SharePoint uses a different chat paradigm than Copilot and lacks some agent features found in other products, raising questions about whether chat is the right interface for workflows that affect whole teams and what integrations are required before general availability.

Introduction

In a new YouTube video, Nick DeCourcy (Bright Ideas Agency) offers a hands-on walkthrough of the public preview for Skills in AI in SharePoint. The report describes how to enable the preview, create and store skills, and run practical demos that show real-world value. Importantly, DeCourcy balances praise with critique, noting both immediate benefits and the gaps that remain before broad adoption.

Overall, the coverage frames the feature as promising but incomplete, and it raises questions about where this capability sits within Microsoft’s broader AI stack. Consequently, organizations that follow this progress can weigh early experimentation against current limitations. The video aims to inform practitioners who plan deployments and governance teams responsible for compliance.

What Skills Do and How They Work

According to the video, a Skill is essentially a reusable, plain-text workflow saved as a Markdown file inside an Agent Assets library in a SharePoint site. DeCourcy demonstrates that users can author skills with natural language via the AI chat, refine the draft, then save it for team use; therefore, non-developers can create repeatable procedures without code. The agent then matches user intents to saved skills or allows teams to invoke them by name, producing consistent outputs.

As a result, skills can chain built-in actions like search, summarize, and list manipulation into a coordinated process. This design makes skills useful for tasks such as generating meeting briefings, checking metadata completeness with write-back, or running audits that populate SharePoint lists. However, this scope also defines clear boundaries: skills operate inside SharePoint's native capabilities rather than acting as generic automation scripts.

Demonstrations and Practical Use Cases

DeCourcy walks through three hands-on demos that highlight immediate value. First, a pre-meeting briefing skill extracts and summarizes relevant documents to prepare attendees quickly. Second, a metadata assessment skill checks folders and writes back metadata to improve library quality, which can save hours of manual auditing. Third, a vendor insurance audit skill creates and populates a SharePoint list, showing how skills can standardize repeated compliance tasks.

These examples show how skills can increase efficiency and consistency for team workflows that revolve around SharePoint content. Moreover, the markdown-based approach enables quick edits and iterations, helping teams refine procedures as requirements evolve. Yet, the demos also illustrate the tradeoff between ease of authoring and the current lack of integrations with systems beyond SharePoint.

Current Limitations and Tradeoffs

Despite the benefits, DeCourcy flags several limitations that should shape deployment decisions. For instance, skills cannot call external systems or run custom code, which restricts scenarios that require data from outside Microsoft 365. In addition, the agent cannot read SharePoint permission or sharing metadata, so governance-focused workflows remain constrained.

Furthermore, he notes issues with the built-in markdown editor and the inability to create Planner tasks directly from a skill, which forces teams to use Power Automate or other workarounds. These limitations reveal a tradeoff: Microsoft prioritizes accessible, safe tooling inside SharePoint, but that approach reduces flexibility for complex, cross-system processes. Therefore, organizations must weigh convenience and governance against integration needs and automation depth.

Architectural Questions and Where This Fits

DeCourcy also explores broader architectural questions about how AI in SharePoint relates to other Microsoft offerings such as Copilot and how it compares to external implementations like Claude from Anthropic. He points out that SharePoint's chat-based agent follows a different interaction model than Copilot, which raises design questions about the right modality for team-affecting workflows. In particular, chat may be useful for ad hoc queries but less well-suited for coordinated, auditable processes that affect many users.

Additionally, he contrasts Microsoft’s current implementation with Anthropic-style skills, suggesting that some capabilities remain less mature in the preview. Consequently, teams planning enterprise deployments should consider both the immediate productivity gains and the likelihood of future changes to integration, governance, and UX. This context matters because adoption strategies that lock into fragile workarounds may require costly rework later.

Outlook and Practical Recommendations

In conclusion, DeCourcy’s hands-on review finds a lot to like in the initial public preview while documenting clear gaps that slow broader adoption. For now, SharePoint skills make good sense for contained, document-centric workflows that benefit from consistency and simple metadata operations. Meanwhile, organizations needing cross-system automation or tight access-audit capabilities should plan hybrid approaches that pair skills with Power Automate or traditional development.

Ultimately, the preview invites early experimentation with careful governance and realistic expectations. As Microsoft iterates, teams should monitor improvements in integration, permission access, and editor experience before committing to large-scale rollouts. Nonetheless, the video presents a concrete step forward that helps non-technical teams capture and share repeatable procedures inside SharePoint with modest overhead.

Keywords

Skills in AI in SharePoint, SharePoint AI features, SharePoint Copilot skills, Hands-on SharePoint AI tutorial, AI in SharePoint search, SharePoint AI best practices, SharePoint AI limitations, Microsoft 365 AI integration