Home | Excel | Power BI | Copilot | Visualising Data
Here are some useful advanced capabilities of Copilot. We have a brief look at them now. There is an exercise later to get familiar with these.
A notebook contains a set of related documents e.g. financial performance, compliance policies. The content can be Word or PDF documents. Copilot just uses the content only to generate its responses - the responses are “grounded” in the content. It does not search the web or use it’s own training. Notebooks are very useful things with lots of great use-cases!
In this case study, we create a notebook titled “Jane Austen Novels” and import five of her novels
Let’s add some instructions to the notebook, for example:
Act as an expert in novels of Jane Austen. Adopt an engaging informal tone but keep your answers brief. Keep to the topic. If the user asks an unrelated question, politely decline to answer it reminding the user that this is outside your expertise and that you can only answer questions on the novels of Jane Austen.
Here are some starter questions to start our literary analysis.
How does Jane Austen portray social class and its impact on relationships in her novels?
In what ways do Austen’s female protagonists challenge or conform to the gender norms of the Regency era?
How does Austen use irony and satire to critique societal expectations and conventions?
What role does marriage play in Austen’s novels?
How do Austen’s narrative techniques shape readers’ understanding of character and perspective?
Copilot can create a 10 minute podcast based on the content - try this out.
Copilot’s Analyst agent will read a dataset, and then answer our questions related to the data; e.g., describe the data, build some charts, provide insights.
For example, we may want to analyse a dataset that has fictional details of SEGRO employees. Download from here. The initial prompt, to accompany the file could be simply
Describe this data.
Follow on prompts could be:
Create a column chart to show the age profile of the workforce.
Revise this chart to show by decade, not by year.
Copilot’s Researcher agent is useful for larger and more challenging tasks. Researcher uses a more powerful reasoning model. It may take a few minutes to respond but comes back with a more considered and comprehensive response.
For example we could ask Researcher to conduct a competitive analysis of SEGRO within the UK REIT market with this starter prompt.
Conduct a competitive analysis of SEGRO within the UK REIT market. Identify its main competitors, compare financial performance (e.g., market cap, revenue, dividend yield, total return), and assess its market positioning, strategic focus, and recent developments. Include data from the past 3–5 years and cite sources where possible.
Researcher will summarise its understanding and ask if we have any further instructions. We could for example add:
Compare in particular with Tritax Big Box
Copilot’s Visual Creator agent creates an image or video based on a text description. Here is an example prompt.
Create a corporate logo for a European REIT. Include the tagline “We create the space that enables extraordinary things to happen”.
A more ambitious prompt may be:
Create a corporate video to promote SEGRO. The message should be that SEGRO creates the space that enables extraordinary things to happen. It should be about 10 seconds long.
Visual Creator may or may not follow your instructions. At the time of writing this (July 2025), it sometimes ignores a prompt to create a video.