AI assistants for Swiss SMEs: verified capabilities to evaluate

Microsoft Copilot: Home, Code, and Autopilot for structured AI work

Microsoft Copilot now offers three new capabilities under its AI at Work platform: Home, Code, and Autopilot. These are designed to help SMEs manage AI work as part of their daily workflows, rather than as isolated experiments. Home unifies Chat and Cowork in a single starting point, where users can review recent activity, get suggestions on what Copilot can help with, and pick up work where they left off. Chat remains for instant, conversational interactions, while Cowork is for delegated, end-to-end tasks such as drafting an RFP response or building a financial close package. The vendor emphasizes that these capabilities are not static tools but part of a continuous improvement loop, where small improvements compound into real competitive advantage over time. This is grounded in Microsoft's long-term commitment to improving AI models and embedding them into familiar productivity apps like Word, Excel, and PowerPoint, which are now integrated into Copilot under "Office in Copilot".

Key considerations for SMEs:

• Does your team already use Microsoft 365, Dynamics 365, or Power Platform? If so, Copilot's integration into these apps may reduce context-switching.
• Are you comfortable with a predictable per-person subscription model for AI access, or do you prefer usage-based billing tied to specific tasks? Microsoft explicitly supports both models, noting that agents vary in their usage patterns.
• Can you define what "good work" looks like for your business, and measure AI agents against it? Microsoft suggests that this iterative approach is more effective than rigid upfront specifications.

[Image: Microsoft Copilot Home interface showing unified Chat and Cowork starting point],

Code: AI-powered solution building for non-developers

Microsoft's "Code" capability allows users to describe small, purpose-built solutions in natural language, and Copilot chooses an approach to build it. This includes desktop widgets, interactive dashboards, or cloud-hosted internal apps, all grounded in the user's context and data. The capability runs in a sandboxed environment within the Microsoft 365 tenant, governed by IT but accessible to non-developers. This is a shift from traditional developer-only tools, as Microsoft positions "Code" as a basic skill alongside writing a memo or modeling a budget. The vendor also notes that Code is powered by the same underlying technology as GitHub Copilot, which may appeal to teams already using GitHub for development.

Practical questions for SMEs:

• Do you have recurring tasks that could be automated, such as supplier reviews or customer briefings? Microsoft's "Autopilot" (previously called Scout) is designed for persistent, proactive work, like managing a supplier review process end-to-end.
• Would you benefit from a shared, cloud-hosted workspace for internal apps? Microsoft's "Copilot Managed Runtime" provides hosting infrastructure within the company's Microsoft 365 environment, enabling apps to connect to live data and be accessible from anywhere.
• Do you need to connect third-party data sources? Microsoft's platform supports Bring-Your-Own-Key (BYOK) for model inference and web search providers, allowing existing organizational setups to integrate with AI agents without additional markup.

[Image: Microsoft Copilot Code interface showing a natural language description of an app being built],

Autopilot: AI as a persistent teammate for long-running tasks

Microsoft's "Autopilot" is a cloud-hosted agent that can watch channels, follow up on threads, run recurring work, and pick up projects days later without waiting for a prompt. For example, it can set up and manage a full supplier review process, including building the schedule and workback plan, handling prep, meetings, and follow-ups, and reaching out to stakeholders for updates. This capability is designed to reduce overhead by automating tasks that traditionally require human intervention, such as monitoring and follow-ups.

Vendor-backed insights for SMEs:

• Autopilot is not a replacement for headcount but a capital investment to build capability, similar to how usage-based billing (UBB) is treated as capital rather than overhead.
• Microsoft emphasizes that the hardest part of AI adoption is not the technology but the change management required to rewire how work gets done and to build trust in AI agents.
• The vendor positions itself as a long-term partner, not just a tool provider, by committing to continuous improvement loops and sharing learnings from its own transformation and from other organizations across industries.

[Image: Microsoft Teams Rooms device joining a Google Meet meeting on Android (AOSP)],

Cloudflare AI Gateway: Web search and interoperability for grounded AI

Cloudflare's AI Gateway is designed to be the control plane for AI applications, with observability, unified billing, security, and access controls built-in. It partners with web search providers like Ceramic.ai, Exa, and Linkup to enable agents to browse the Internet dynamically, retrieving live, structured information from the web. This is particularly useful for tasks involving recent events, changing APIs, or evolving documentation, as the agent can access up-to-date context without relying on static, pre-trained data.

Key features for SMEs:

• AI Gateway supports programmatic web search via a standard REST endpoint, allowing existing backend systems to integrate with AI search without additional markup.
• The vendor identifies partners supporting Zero Data Retention (ZDR), meaning data is not stored after retrieval, which may appeal to teams with strict data privacy requirements.
• AI Gateway's unified billing system uses AI Gateway credits for both model inference and web search queries, simplifying cost management across different AI operations.

Practical implementation checklist:

• Grounding:** Does your AI assistant need to access live web data? If so, AI Gateway's web search API can provide dynamic context, reducing the risk of 404 errors from incorrect URL guesses.
• Transparency:** Does your team need visibility into how AI spends are allocated? AI Gateway logs all web search queries alongside model inference calls, giving complete visibility into spend and the ability to set limits before invoices arrive.
• Security:** Are you concerned about data retention? Cloudflare's AI Gateway identifies partners supporting ZDR, ensuring your data is not retained after use.
• Interoperability:** Do you use multiple video conferencing platforms? Cloudflare's AI Gateway supports interoperability between Google Meet and Microsoft Teams on Android (AOSP) devices, enabling hardware from one platform to join meetings on the other.

[Image: Cloudflare AI Gateway REST API endpoint for web search integration],

Web Search API: Structured, live data for AI applications

Cloudflare's Web Search API partners with verified web search providers to enable agents to retrieve live, structured information from the web. The vendor emphasizes that this is a net-positive for a fair Internet, as it requires compliance with Cloudflare's publicly stated requirements for "Verified bots" and includes a link to the source of crawled content. This transparency and control may help SMEs comply with regulatory frameworks like the EU's Cyber Resilience Act, which the vendor notes is a key consideration for open source projects and content management systems alike.

Vendor-backed standards for SMEs:

• The Web Search API is designed to integrate directly into existing AI pipelines, providing fresh, structured context without additional markup or overhead.
• The vendor positions itself as an advocate for open source and transparency, aligning with the Open Website Alliance's charter to ensure that everyone has the freedom to use, change, and share the tools they rely on.
• Cloudflare's AI Gateway is positioned as a step toward elevating industry standards for web search, ensuring that crawlers are honest, transparent, and respect all bot rules and preferences.

[Image: Cloudflare AI Gateway interoperability between Google Meet and Microsoft Teams on Android (AOSP) devices],

Google Workspace: Skills, interoperability, and AI voiceovers for SMEs

Google Workspace has introduced "skills" in the Gemini app and Workspace, which are reusable prompts tailored to specific tasks. Over the coming weeks, skills will roll out to Workspace users and eventually replace "Gems" as the tool for customizing instructions for tasks like AI voiceovers in Google Vids. Google Vids now supports AI voiceovers with an upgraded Gemini 3.8 Flash Lite TTS model, enabling more natural and expressive voiceovers for video content.

Verified capabilities for SMEs:

• Customization:** Google Workspace allows customization of AI voiceovers in Google Vids, including font, typography, screen placement, and animation styles, to align with brand or video aesthetics.
• Interoperability:** Google Workspace supports built-in interoperability between Google Meet and Microsoft Teams on Android (AOSP) devices, enabling users to join meetings on either platform from hardware running the other platform.
• Skills integration:** Google Workspace's "skills" feature is designed to bring specialized expertise into apps, such as financial skills in Google Sheets or legal skills in Google Docs, enabling organizations to scale best practices through reusable, role-specific workflows.

Practical questions for SMEs:

• Do you need to manage AI costs predictably? Google Workspace's skills feature is part of a broader shift toward predictable subscription models for AI access, similar to Microsoft Copilot's approach.
• Can your team benefit from reusable prompts for specific tasks? Google Workspace's "skills" feature is designed to replace "Gems" and provide a more scalable, role-specific approach to AI customization.
• Do you need to integrate AI into your existing workflows? Google Workspace's "skills" feature is designed to bring specialized expertise into familiar productivity apps, reducing the need to switch contexts.

[Image: Google Vids interface showing customizable AI voiceovers and captions],

Decision checklist: AI assistants for Swiss SMEs

1. Define your AI goals and tasks

Before evaluating AI assistants, clearly define what tasks or goals you want AI to accomplish. This could range from automating supplier reviews to handling customer enquiries or managing bookings. Microsoft Copilot's "Home" capability is designed to help users review recent activity and get suggestions on what AI can help with, while Google Workspace's "skills" feature is designed to bring specialized expertise into apps for specific tasks. Skills in Google Workspace are reusable prompts tailored,