Sales Command Center

Claude
Built on Claude

One place Sales can use to research, prep, and advance their accounts. They name the account they're working and what they need, in plain English. It reads the live CRM and call transcripts, does the work, and writes back only what they approve.

Think of it as an analyst who already knows the pipeline, the methodology, and the competitors, and never forgets to check the CRM first.

How it's built · the Claude features behind it

What it's made of

Claude Project
Each salesperson gets their own Project, connected to their own CRM login, so it only ever sees their accounts and writes as them.
Shared setup
The skills and reference library are shared across every Project and managed centrally. One person maintains them and an update lands for the whole team at once. Individual workspaces, one common playbook.
Project instructions
The operating rules Claude loads into every conversation: how to start a session, when to ask before writing, how to weigh evidence, what never to do. Edited once, in force everywhere.
Project knowledge
The reference library Claude reads from: methodology, competitor battlecards, product proof points, territory map, win/loss history, plays, objection handling, pricing. Facts live here so the rules stay short.
Skills
One file per repeatable output, holding its template, the questions to ask, and the data to pull. Claude picks the right one from plain English, so the same request produces the same structure for everyone. Fix a skill once, everyone gets the fix.
Connectors
Claude's live, permissioned links to the CRM and the meeting-transcript tool, using each person's own login. Nothing is cached between sessions, so every run reads current data. Without this, an assistant is guessing from whatever got pasted in.
Web search
External research on companies and people, with a rule to prefer primary sources and flag conflicting numbers rather than quietly picking one.
File creation
Claude produces real deliverables, like a filled interactive model or a finished document, not just text to copy out of a chat window.
Claude Design
Anything customer-facing is generated as content plus a design brief, then rendered through the shared design system so branding stays consistent without anyone policing it.

What happens when you hit enter

1 · Identify
Claude reads who you are from your CRM connection, so "my pipeline" is always a live query scoped to you, never a stored list that goes stale.
2 · Resolve
Matches the account, and stops to ask if more than one record could match. Working the wrong company is worse than a slower answer.
3 · Pull
Live records, call transcripts where they exist, and anything you paste in. Claude holds an entire deal at once, so it reconciles these against each other instead of summarizing one at a time.
4 · Run
Executes the matching skill. Same steps, same structure, every time.
5 · Propose
Shows proposed changes with their evidence, and waits. Deal data is never written without a yes.
6 · Log
Records the run and stamps a cycle-stage field, so the team can see what is used and where deals actually sit.

Why this shape

Rules vs. facts
Instructions carry the rules; knowledge files carry the facts. When a fact changes, swap one file instead of rewriting the rules.
One skill per job
Every output type has exactly one file behind it, shared across all Projects. That is what makes quality consistent across people instead of dependent on who wrote the best prompt.
Guardrails that hold
The rules that matter most here are behavioral: never write without asking, tag every claim, say plainly when evidence is missing, and push back when the record is not supported by the calls. Claude holds these across long sessions rather than drifting.
State lives in the CRM
Chat history is deliberately not the memory. Everything durable is written back to the record, so nothing depends on finding an old conversation.

What it looks like

A real exchange you type the account · it loads everything and offers what's useful
Northside Industries
Northside Industries · Springfield, IL · Customer · Owner: you · CS lead: J. Rivera
Open dealStageAmountClosesLast touch
Platform ExpansionNegotiation$92,0969/30/2668 days ago
Regional RolloutProof$97,87912/31/264 days ago
  • Deal qualification: 6 of 8 areas partially covered, 2 unknown. The budget holder isn't confirmed and legal hasn't started.
  • Your main contact is supportive but hasn't pushed this upward, which is a risk on a deal this size.
  • The expansion has been quiet for 68 days. The incentive you offered expired in July.
Most useful from here. 1. score the deal and update the CRM · 2. full risk read · 3. draft the re-engagement email
You typed 2 wordsIt found the account, both open deals, the history, and the transcripts on its own.
It read the callsThe 68-day gap and the expired incentive came from meeting transcripts, not from anyone's notes.
It waitsIt offers next steps and stops. It doesn't run anything or change anything until you pick.

How to start

Northside Industries

Just the account. What you saw above. Best when you want to see where a deal stands before deciding anything.

prep me for the Northside demo

Account plus what you need. Fastest. Plain English, so you never have to memorize a command.

go

Nothing in mind. Asks which account and lists everything it can do. You can paste a CRM link instead of typing a name.

Things you can ask

"Where does this deal really stand?"
Grades every qualification area against what was actually said on calls, then offers to update the CRM.
"Is this going to close?"
Ranked risks, what's missing, the highest-leverage next move, and a forecast confidence with reasoning.
"I'm demoing Thursday, prep me."
Who's attending and what each cares about, their pain in their words, goals, questions to weave in.
"Write the recap from today's call."
Next steps with dates, the value points they reacted to, and their questions answered.
"What do we know about this company?"
Research before a first call: people, business lines, systems in use, connections, questions to ask.
"Who's the actual decision maker here?"
Maps the stakeholders, flags who has authority, and says plainly when it's a guess.
"They're comparing us to someone else."
Head-to-head positioning aimed at the criteria this buyer actually said they care about.
"Build the business case."
Fills the ROI model, starting from numbers the customer already agreed to rather than guesses.
"This has gone quiet. What now?"
Reads the gap since last contact, finds what lapsed, and drafts the re-engagement.
"Is this inbound lead worth my time?"
Who they are, how qualified, how urgent, and who should own it.
"Where should I be prospecting?"
The best accounts to work in a territory, ranked, with a reason for each.
"What's on my plate this week?"
Your open deals pulled live, with what needs attention and why.

Ask in your own words. You never have to name a job or remember a command.

Everything it can do

Ask for…WhenWhat comes back
top accountsPlanningThe best accounts to work in a territory, ranked, with a reason for each
segment an accountPlanningEnterprise or Commercial, and who should own it
account briefResearchCompany research before a first call: people, business lines, systems in use, connections, questions to ask
outboundProspectingEmail, LinkedIn message, call opener, and a follow-up bump, specific to that company
event follow-upProspectingPost-event email built from the conversation you actually had
qualify an inboundInboundIs this lead real, how urgent, and who should own it
score the dealAny dealAll 8 qualification areas graded with evidence, a fix-list, and a proposed CRM update
risk readAny dealRanked risks, gaps, one next move, forecast confidence
demo prepDemoInternal prep guide: attendees, their pain, technical fit, goals, questions to weave in
follow-up emailDemo · ClosePost-demo or post-pricing recap, ready to send
ROI estimateProofThe ROI calculator filled in, starting from numbers the customer already agreed to
ROI slidesProofCustomer-facing business case with every number traceable
competitor compareAny dealHead-to-head against whoever's in the deal, aimed at what they care about

How it behaves

It asks before writing

Any change to your deal shows up first as a table: what's there now, what it wants to change it to, and why.

Say push to apply it, pick some of it, or say no. Do nothing and nothing happens.

It shows its sources

Every claim is marked: confirmed on a call, assumed, or an internal benchmark.

When it doesn't know, it says so instead of filling the gap with something that sounds right.

It'll disagree with the CRM

If a contact is marked as a strong advocate but never actually pushed internally, it'll say so and explain why.

You can overrule it. It's your deal, and it just won't pretend the evidence is there.

Runs get logged

Each time a job runs, it records which one, on what account, and when.

That's about improving the tool, not grading you. Logging never changes your deal data.

Before your first run: connect your own CRM and meeting-transcript accounts inside your workspace. They're per-person, so nobody can do it for you.

One account per chat. New deal? New chat. It keeps accounts from bleeding into each other.