Microsoft Copilot was officially released to business customers on 1 November 2023 and since then the take-up has been vast.
It is an AI-powered assistant designed to enhance productivity, streamline tasks, and provide intelligent solutions within the Microsoft 365 ecosystem. It combines the capabilities of the former Bing Chat and Bing Chat Enterprise tools, offering a seamless experience for users.
Pricing for Copilot is £296.40 + VAT per user per year, and requires a minimum of Microsoft 365 Business Premium or a Microsoft 365 Enterprise Plan. Pricing for Education and Non-Profit organisations is not yet available.
Microsoft has put together further information about Copilot’s key features and how to use it.
However, perhaps the best way to describe Copilot’s functionality, and to help put it into context, is to identify some real-life use scenarios.
Education - personalised learning plans:
Copilot can analyse student performance data to create customised learning plans, tailoring educational content to individual needs.
Healthcare - patient data management:
Copilot can organise and analyse patient data, aiding in quicker diagnoses and personalised treatment plans.
Retail - sales data Analysis:
The analysis of sales data to predict future trends and inventory needs.
Finance - automated financial reports:
Copilot can generate financial reports by extracting relevant data from spreadsheets and other sources.
Customer service - AI-powered chatbots:
The creation of chatbots to handle customer inquiries, answer FAQs, and escalate complex issues to human representatives.
Marketing - content creation and optimisation:
Copilot can assist in creating marketing collateral, suggesting improvements, and optimising content for better engagement.
Human Resources - CV screening and candidate matching:
Copilot can analyse CVs, match candidates to job descriptions, and streamline recruitment processes.
Like any new technology there are some limitations to Copilot, and the key ones are summarised below based on user feedback.
While Copilot aims to provide accurate and informative responses, it may not always be completely accurate. Responses are generated based on patterns and probabilities in language data, which can lead to occasional inaccuracies. So, users should verify critical information independently.
Despite its capabilities, Copilot does not support image generation using models like DALL-E, even within applications like Microsoft Word. Users cannot create custom images or visual content directly through Copilot.
Copilot's responses heavily rely on existing language data and patterns. If a query falls outside its training data, it may struggle to provide relevant answers. And users should be aware that Copilot's knowledge is based on historical data up to a certain point.
While users can provide feedback, the process is currently user-initiated. Administrators can supplement this by providing feedback on behalf of users, but it's not fully automated. The feedback system could be more streamlined and proactive.
Copilot processes user interactions and conversations. Organisations need to consider privacy implications, especially when sharing diagnostic logs with Microsoft. So, redaction of sensitive information is essential.
Building custom CoPilots using Copilot Studio requires technical expertise and effort. So, organisations may need additional resources for customisation as developing industry-specific CoPilots can be challenging.
Copilot does not facilitate real-time collaborative editing or simultaneous content creation with multiple users. Users cannot co-author documents directly within Copilot.
Users need time to adapt to Copilot's features and understand its capabilities fully. So, training and familiarisation are essential for optimal usage.
Copilot's understanding of context can sometimes be limited. It may not always grasp the nuances of a specific conversation or document. Users should review Copilot's suggestions carefully to ensure they align with the intended context.
Copilot's responses are generated by language models, which may not always capture domain-specific jargon or industry-specific terminology accurately. Users should review and validate responses, especially in specialised fields.
In summary, while Microsoft Copilot offers significant benefits, users should be aware of its limitations and use it prudently, complementing it with other tools and manual verification when necessary.