// AI Advantage

Turn a Competitor Brief Into a Decision

Use one checked finding to decide what to change, what to investigate and what to leave alone.

A competitor announcement lands in Slack. By the next meeting, someone wants to change the roadmap. Between those two events, did anyone establish that your customers need the feature?

I would make the agent do one more piece of work before that brief circulates: explain what would justify a response.

In Give Dots and Muse the Same Job, we covered gathering official sources, checking claims and counting the cleanup. This edition picks up with one checked finding. The next job is deciding whether to change something, investigate further or leave it alone.

You can use Muse, Meta’s browser-capable AI agent, or another assistant to draft the decision. The worked example below is editorial analysis, not an output from a tested Muse run.

Turn one finding into a decision card

Use four fields:

Evidence: What does the source actually establish?
Relevance: Which customer, deal or product decision could it affect?
Response: What is the smallest useful action now?
Revisit trigger: What observable event would change that decision?

The second field is the one I would spend time on. “A competitor launched AI” is not customer relevance. “Three prospects asked whether we can do this exact workflow” is. If you do not have that evidence, the card should say so.

A worked example: announced does not mean available to everyone

Anthropic announced Claude Duet on October 21 as a proactive agent, with research-preview access beginning through a waitlist for U.S. Pro and Max subscribers.

Suppose you sell a tool for preparing daily work briefs. Here is an illustrative decision card:

Evidence: Duet has been announced for an initial waitlisted audience. The announcement does not establish that every prospect can use it today.

Relevance: It could affect buyers who want a morning brief across connected applications. We have not established how often our own prospects request that capability.

Response: Add a dated, sourced note to the internal competitor reference. Give sales one discovery question: “Which information would you want in a daily brief, and where does it live today?” Do not rewrite the roadmap solely because of the announcement.

Revisit trigger: Broader access relevant to our buyers, or three qualified prospects independently identifying this workflow as a purchase requirement. Three is an illustrative threshold for this team, not an industry benchmark.

Decision: investigate customer relevance; leave the roadmap unchanged. In this hypothetical team, the sales lead owns the discovery question and reviews the answers at the next weekly pipeline meeting. If prospects do not value the workflow, that weakens the case for building it—even if access expands.

Illustrative decision card for the hypothetical daily-brief business: investigate customer relevance, keep the roadmap unchanged and name a revisit trigger. Original AI-generated illustration for The AI Enterprise.

The card now has a decision, an owner and a point at which to look again. Nobody needs another five-page brief to know what happens next.

Give the agent the decision to support

Paste one verified finding and its supporting passage, then use this:

Our customer and product: [BRIEF DESCRIPTION]
Decision we might change: [SPECIFIC DECISION]
Checked finding and dated source: [FACT, PASSAGE, URL]
Customer evidence we already have: [EVIDENCE, OR NONE]

Draft one decision card:
1. What the source establishes, including availability limits.
2. Who it could affect, separating evidence from assumptions.
3. The smallest useful response now.
4. An observable trigger for reconsidering the decision.

Choose: act now, investigate, monitor, or no change.
Explain the choice. State what evidence would weaken it.
Do not invent customer demand, usage or measured performance.
Do not browse for more material; list any missing evidence.
Do not send, publish or update another system.
Return the card here for review.

Review the recommendation against the source and your customer evidence. Assign an owner and a review date only after you accept it. Keep the finding and source beside the decision so a colleague can reconstruct the reasoning.

“No change” is a useful answer when the finding does not affect your buyers or the evidence is too thin to justify action. Record why. Otherwise the same announcement can consume another meeting next week.

Try this with one finding already sitting in your inbox. Leave with an action and owner, or a reason to leave it alone.