What you can ask

Before you start. Connect a client first (see the MCP Server overview). Name the deal by company or deal name; that is how the assistant finds it.

Read a deal

• "Pull up the Acme deal and walk me through the value breakdown." Returns the deal summary, each applicable driver with its calculated impact, and the readiness assessment from the latest story.
• "What are the risk factors in this deal based on what we know?" Returns the risks the copilot sees in the deal context, customer model and conversation history, with what to do about each.

Generate a business case

• "Build a business case for [company] using [product]." Creates a deal on that product and generates the customer model. This draws credits.
• "Write an executive summary for the economic buyer at [company]." Returns a summary in the register that audience expects, grounded in the deal's drivers.
• "Build the deck for tomorrow's meeting." Creates a Sales Deck on the deal and starts full generation. Exporting it as PPTX or PDF draws credits.

Coach an objection

• "They pushed back on price. Give me a response backed by the drivers." Returns an evidence-backed response anchored to the deal's quantified value drivers.
• "The buyer says the other vendor is cheaper. How should I respond?" Returns a response that uses the competitor comparison and pricing analysis on the deal, if you have attached them.
• "The prospect says their team handles this manually and it's 'good enough.' Based on our value model, help me quantify what 'good enough' is actually costing them." Returns the cost of the status quo, driver by driver.

Draft a follow-up

• "Draft a follow-up to their economic buyer, led with the payback number." Returns a draft that opens with the headline value, leads with the two drivers most relevant to that audience, and closes with a next step. It comes back for you to review and send; nothing is sent for you. Add tone, length and audience constraints to the prompt and the draft follows them.

Update a variable

• "Update their headcount to 1,200 and recalculate." Updates the customer variable, triggers a recalculation, and returns the new expected value and payback.
• "Update the deal notes: the champion confirmed budget is approved." Writes the note to the deal. Discovery details you mention (goals, pain points, stakeholders, timeline, budget) go into the deal's structured context.

Add a competitor

• "Add [competitor product] as an opposition model to the Acme deal." Attaches the competitor's value model to the deal and returns a side-by-side, driver-by-driver comparison. The competitor model must already exist in your workspace.

Browse products and benchmarks

• "Which products in my workspace are ready for deals?" Returns the products whose value model and market values are complete.
• "Show me the value drivers and equations for [product]." Returns each driver with its equation and the variables it uses.
• "What are the market values for [variable] on [product]?" Returns the low, typical and high figures from the market estimates.
• "List the deals in this workspace at the proposal stage." Returns the matching deals with stage and expected value.

Pull competitor pricing

• "Pull the competitive pricing analysis for the Acme deal." Returns structured data from the competitor's published pricing page: tiers, features, billing model and usage metrics, with competitive positioning. Analyses are cached for 90 days, so the data is current rather than real-time.

Get more from every prompt

• Give context, say what you need, and set constraints: who you are meeting, what stage the deal is in, what output you want, and the tone, length and audience.
• Name the persona. "The economic buyer" gets a better answer than "the buyer".
• Tell it what happened. The assistant builds on what you say, and what you say is written back to the deal.
• Save where you got to: "Save a checkpoint of this session." See Session checkpointing.

Related

• MCP Server overview
• Coach a stalled deal from Claude