Automate Prospecting and Call Prep with Claude as a Sales Manager

When your CRM, inbox, meeting notes, and pipeline all tell slightly different stories, sales management gets harder than it needs to be. You need to know which accounts deserve attention, what happened in previous conversations, where deals are stuck, and how to coach your reps—without piecing together every detail by hand.

Claude Cowork can help bring that information together when it has access to the approved folders and connected tools your task requires. The bigger win comes from turning one-off requests into reusable skills: saved instructions for recurring workflows. Here’s how to build practical skills for prospecting, follow-ups, call prep, coaching, and pipeline reviews—and keep the results grounded in evidence.

Connect the sales information you need

Claude’s output depends on the information it can access. Before you build a workflow, identify which systems hold the context for that task.

Depending on your setup, useful sources may include:

• Your CRM, such as Attio, for account details, deal stages, and ownership.
• Gmail and your calendar for conversation history and upcoming meetings.
• Fireflies.ai for meeting transcripts.
• Apify for supported prospect research and data collection.
• Apollo or Clay for lead data and enrichment.
• Slack or Google Docs for sharing reports.

Connector availability and permissions vary. Connect only what the task needs, and follow your organization’s privacy rules and platform terms when collecting or using prospect data.

Anthropic’s sales plugin includes starting points such as account research, call preparation, and pipeline review. Where available, these templates can help you get moving. Customize them around your team’s process so Claude returns information you can actually use.

Build prospecting workflows around your ICP

A prospecting skill can help you identify people engaging with relevant LinkedIn posts, enrich their profiles, and assess their fit against your ideal customer profile, or ICP.

In one reported workflow example, Claude used Apify to identify 127 unique engagers and qualified 17 against the supplied criteria. The output was a CSV with profile URLs, company information, and the posts those people had engaged with.

Those figures describe one particular run—not a promised qualification rate. Engagement alone also doesn’t establish buying intent.

For your team, shape the process around clear steps:

1. Identify a relevant post or profile.
2. Collect available engagement data through an approved tool.
3. Deduplicate the people identified.
4. Research their roles and companies.
5. Apply your ICP criteria.
6. Export a shortlist with supporting evidence for review.

Specify the evidence Claude needs. In another example, an initial collection found 154 unique people, but the workflow needed correction because it lacked the profile details required for qualification. Specifying fields such as the About section, company name, and website improved the process.

Try a prompt like this:

Research people who engaged with these LinkedIn posts using our approved Apify workflow. Deduplicate the results and assess each person against this ICP: [criteria]. Export a CSV with name, role, profile URL, company, company size, relevant post, qualification decision, and supporting evidence. Mark missing information as unknown rather than guessing. Do not send outreach.

Once the output meets your standards, ask Claude to save the process as a skill. Keep research and outreach separate so you can review the shortlist before anyone gets contacted.

Find follow-up opportunities in your CRM

Your next useful sales conversation may already be sitting in your CRM. Lost or neglected opportunities can be worth a second look when something has changed—or when the original record needs a closer review.

A prospect-mining skill can combine CRM records, current account research, and email history to help you decide which conversations may be worth reopening.

In one reported example, the workflow reviewed 160 records in a lost-lead column. It researched the contacts, checked previous communications, and produced a prioritized CSV based on company size, ICP fit, and conversation history. That result describes research, not evidence that those leads later converted.

For each record, ask Claude to surface:

• The original reason the opportunity stalled or closed.
• The last meaningful interaction.
• Any current information that could justify reconnecting.
• A priority level and its rationale.
• A draft follow-up grounded in the previous conversation.

For larger lists, ask Claude to use parallel subagents where supported. One example assigned 15 leads per subagent and used the approach for lists of roughly 100–200 leads. Treat these as workflow examples, not capacity guarantees. Connector limits and research depth still matter.

Review every suggested message before sending, especially if a contact previously declined or opted out.

Prepare sharper briefs before sales calls

Call preparation is a useful place to start: the output is easy to inspect and immediately helpful. Ask Claude to combine the account’s CRM record, relevant emails, previous Fireflies.ai transcripts, and current company research into a short brief.

A practical brief can include:

• Company overview: what the business does and why it may fit.
• Account snapshot: stage, stakeholders, and open questions.
• Interaction history: previous discussions and commitments.
• Suggested agenda: what the meeting should accomplish.
• Discovery questions: what you still need to learn.

Customize the skill to match your team’s sales methodology. You might ask Claude to separate confirmed facts from assumptions and highlight unresolved objections.

Where scheduled tasks are available, you could request briefs for the day’s meetings at 7:00 a.m. Check the execution requirements first: a scheduled desktop workflow may depend on Cowork being open and the necessary connections being available.

Turn sales analysis into useful coaching

Claude can combine CRM outcomes, transcripts, and email history to help you prepare a win-loss report. Give it a clearly defined period and deal population, then ask it to examine:

• Win rate and outcomes by customer profile or company type.
• Common objections and pricing concerns.
• Differences in follow-up and stakeholder engagement.
• Examples of won and lost deals.
• Potential early disqualification signals.

Ask for evidence, not just conclusions. If Claude says deeper engagement is associated with wins, ask it to identify the supporting deals and note any missing data. A pattern in your records does not prove what caused the outcome.

You can also use these sources to support rep coaching. Ask for strengths, improvement areas, and specific examples related to discovery, demonstrations, objection handling, pricing discussions, and follow-up.

Treat any AI-generated grade as a coaching aid, not an objective verdict. Use a consistent rubric and account for differences in territories, deal complexity, and record completeness.

Make pipeline review a daily management habit

A pipeline-review skill can give you a focused brief on the deals that need attention today, delivered in chat or through Slack where connected. Ask Claude to flag stalled opportunities, missing next steps, unanswered messages, and mismatches between recent conversations and CRM records.

Start with recommended updates, not automatic changes. Once the workflow proves reliable, define which low-risk updates Claude can make and which still need your approval.

Do this next: Choose one recurring sales task, run it with Claude, inspect the evidence, and refine the instructions. Save it as a skill only when the output is dependable. That’s how you build a repeatable process—not just another report to read.

Ready to build practical AI skills for your work? Explore Tixu.ai, a beginner-friendly AI learning platform.