How to Automate Pre-Call Account Research and ABM Scoring with Claude

Sales reps spend roughly 30 percent of their time actually selling, according to Salesforce, with the rest consumed by admin, data entry, and prospect research. One GTM team we spoke with found their SDRs were spending the first 30 minutes of every call block manually researching prospects with AI chat tools, and the data came back wrong often enough to reduce their trust on those tools. The reps were putting in the effort, but the research architecture underneath was broken.

This guide shows how to build a Claude workflow that pulls live account data through MCP, scores each account against your ICP criteria using natural-language reasoning, and delivers a scored pre-call brief before every scheduled meeting. You can set it up in Claude Desktop without writing code, or build a fully automated pipeline in Claude Code that runs on a schedule. Both paths use the same enrichment APIs and produce the same output: a brief that tells the rep whether the account fits, what signals matter, and what to talk about.

Why Every LLM Fails at Account Research Without a Live Data Provider

Every large language model, Claude included, produces unreliable account research when it relies on web search or training data alone. The most common failures are hallucination and displaying outdated funding rounds, wrong job titles, and headcounts that reflect a company's size from months ago.

The architectural fix is connecting Claude to a data provider that returns structured, fresh records through an API. However, B2B data decays at roughly 2.1 percent per month according to Martal, so any data provider that does not update its records frequently will return wrong information on a meaningful share of your accounts. It's imperative that you connect a good quality B2B data provider to Claude.

For example, when conncting Crustdata to Claude, it gets live data. When Claude calls the Company Enrichment API through MCP, it gets 250+ live company datapoints from 15+ sources per company, not a web-scraped summary. When it calls the People Enrichment API, it gets 90+ structured datapoints per person, including current role, tenure, work history, and skills. The model handles reasoning and synthesis. The data provider handles accuracy and freshness. That separation is what makes the workflow reliable enough to trust before a call.

What the Workflow Looks Like End to End

The full pipeline runs in six steps, from a calendar event to a delivered brief. Each step maps to a specific MCP tool or API call.

Step 1: Calendar trigger. A new meeting with external attendees appears on the calendar. In Claude Code, a Python script reads the Google Calendar API on a schedule (cron every 15 to 30 minutes) and kicks off enrichment for any new meetings it has not processed yet, or a Google Calendar push notification fires a webhook that spawns the workflow in real time. In Claude Desktop, you open a conversation and ask Claude to check your calendar using its built-in Google Calendar integration, which is the one manual step in an otherwise automated pipeline.

Step 2: Attendee identification. Claude reads the meeting invite, extracts attendee email addresses and names, and resolves each email domain to a company. Personal email domains (gmail.com, outlook.com, etc.) are filtered out, and the remaining corporate domains are used to look up companies.

Step 3: Company enrichment. For each company, Claude calls the Company Enrichment API to pull firmographics (employee count, headquarters, industry, revenue), funding history (last round type and amount, total raised, investors), headcount growth (3, 6, and 12-month rates by department), hiring signals (open roles, posting velocity), tech stack, and web traffic trends.

Step 4: People enrichment. For each attendee, Claude calls the People Enrichment API to pull their current role, seniority, tenure, past companies, education, and skills.

Step 5: Signal scan. Claude checks for recent signals that change the account's priority: a funding round in the last 90 days, executive hires, a spike in job postings, or recent social posts from decision makers at the company. The Job Listing API provides hiring velocity, and the Posts API surfaces what key people at the account are publishing and engaging with.

Step 6: Score and deliver. Claude evaluates the enriched data against your ICP criteria, assigns a score with reasoning, generates a structured brief, and delivers it to Slack, CRM, email, or a shared file.

The result is a scored account brief that covers six data categories (firmographics, funding, hiring, headcount, tech stack, and people) plus an ICP score with explanation, all generated from live data. For a deeper look at what each of those data categories should contain and how they map to API endpoints, the account-based intelligence guide covers the framework in detail.

How to Set This Up in Claude Desktop

Claude Desktop lets you run this workflow without writing code. You connect the Crustdata MCP server for enrichment and use Claude's built-in Google Calendar integration to read your meetings. Claude Desktop does not support background automation, so you open the app and trigger the workflow yourself each morning or before each call block. Once triggered, everything from calendar reading to brief generation runs automatically within the conversation.

Connect the Crustdata MCP Server

Open Claude Desktop settings and add the Crustdata MCP server to your configuration. Once connected, Claude gains access to company enrichment, people search, people enrichment, job search, social posts, and web search as callable tools. You will need a Crustdata API token, which you can get from your Crustdata dashboard.

The MCP server exposes the same endpoints as the REST API, but Claude calls them as tools within the conversation. When you ask Claude to "enrich this company," it calls the Company Enrichment API behind the scenes and returns structured data directly into the chat.

Connect Google Calendar

Claude Desktop has a built-in Google Calendar integration. Connect it through the integrations menu, which requires OAuth authentication with your Google account. Once connected, Claude can read your upcoming meetings and extract attendee names and email addresses directly in the conversation, without you copying and pasting meeting details.

Run the Research Workflow

With the Crustdata MCP server and Google Calendar connected, start a new conversation and use a prompt like this:

Check my Google Calendar for all meetings scheduled today. For each meeting
with external attendees:

1. Extract the attendee email addresses and resolve each domain to a company.
2. Enrich each company with firmographics, funding history, headcount growth
   (3, 6, and 12-month rates), hiring signals (open roles and posting velocity),
   tech stack, and web traffic.
3. Enrich each attendee with their current role, seniority, tenure at their
   current company, and past employers.
4. Check for recent signals: funding rounds in the last 90 days, executive
   hires, job posting spikes, and recent social posts from decision makers
   at each company

Check my Google Calendar for all meetings scheduled today. For each meeting
with external attendees:

1. Extract the attendee email addresses and resolve each domain to a company.
2. Enrich each company with firmographics, funding history, headcount growth
   (3, 6, and 12-month rates), hiring signals (open roles and posting velocity),
   tech stack, and web traffic.
3. Enrich each attendee with their current role, seniority, tenure at their
   current company, and past employers.
4. Check for recent signals: funding rounds in the last 90 days, executive
   hires, job posting spikes, and recent social posts from decision makers
   at each company

Check my Google Calendar for all meetings scheduled today. For each meeting
with external attendees:

1. Extract the attendee email addresses and resolve each domain to a company.
2. Enrich each company with firmographics, funding history, headcount growth
   (3, 6, and 12-month rates), hiring signals (open roles and posting velocity),
   tech stack, and web traffic.
3. Enrich each attendee with their current role, seniority, tenure at their
   current company, and past employers.
4. Check for recent signals: funding rounds in the last 90 days, executive
   hires, job posting spikes, and recent social posts from decision makers
   at each company

Claude will read your calendar through the built-in integration, then call the Crustdata MCP for each company and person. Depending on your meeting load, this takes 30 seconds to a few minutes.

Add Scoring

After the research step, add scoring to the same conversation:

Now score each account on a scale of 1 to 100 against these ICP criteria.
Explain your reasoning for each score.

Our ICP:
- B2B SaaS companies with 200 to 2,000 employees
- Series B or later funding
- Actively hiring sales or marketing roles
- Using Salesforce or HubSpot as their CRM
- Headcount growth above 15 percent in the last 6 months
- Headquartered in North America or Western Europe

Now score each account on a scale of 1 to 100 against these ICP criteria.
Explain your reasoning for each score.

Our ICP:
- B2B SaaS companies with 200 to 2,000 employees
- Series B or later funding
- Actively hiring sales or marketing roles
- Using Salesforce or HubSpot as their CRM
- Headcount growth above 15 percent in the last 6 months
- Headquartered in North America or Western Europe

Now score each account on a scale of 1 to 100 against these ICP criteria.
Explain your reasoning for each score.

Our ICP:
- B2B SaaS companies with 200 to 2,000 employees
- Series B or later funding
- Actively hiring sales or marketing roles
- Using Salesforce or HubSpot as their CRM
- Headcount growth above 15 percent in the last 6 months
- Headquartered in North America or Western Europe

Claude reads the enriched data it already pulled, evaluates each criterion against the data, and writes a scored assessment. The output includes the score, which criteria the account matched, which it missed, and any additional context that affects fit. The Natural-Language ABM Scoring section below covers the methodology in detail.

Format and Deliver the Brief

Ask Claude to format the results as a structured brief:

Format each account as a pre-call brief with:
- Account name, ICP score, and tier (Tier 1: 80+, Tier 2: 50-79, Tier 3: below 50)
- Scoring reasoning (2-3 sentences)
- Key firmographics and funding data
- Hiring signals and headcount growth
- Attendee profiles (role, tenure, recent activity)
- Three suggested talking points based on the signals
- Any risk factors or concerns

Format as markdown I can paste into Slack

Format each account as a pre-call brief with:
- Account name, ICP score, and tier (Tier 1: 80+, Tier 2: 50-79, Tier 3: below 50)
- Scoring reasoning (2-3 sentences)
- Key firmographics and funding data
- Hiring signals and headcount growth
- Attendee profiles (role, tenure, recent activity)
- Three suggested talking points based on the signals
- Any risk factors or concerns

Format as markdown I can paste into Slack

Format each account as a pre-call brief with:
- Account name, ICP score, and tier (Tier 1: 80+, Tier 2: 50-79, Tier 3: below 50)
- Scoring reasoning (2-3 sentences)
- Key firmographics and funding data
- Hiring signals and headcount growth
- Attendee profiles (role, tenure, recent activity)
- Three suggested talking points based on the signals
- Any risk factors or concerns

Format as markdown I can paste into Slack

Copy the output into your team Slack channel, CRM account notes, or wherever your reps access pre-call context.

How to Set This Up in Claude Code

Claude Code runs the same workflow but adds full automation. You tell Claude Code what to build, it writes the integration code for your specific stack, and you can schedule it to run before your first call every morning. No pre-written scripts to copy and modify.

Run the Workflow from a Prompt

With the Crustdata MCP server configured in your Claude Code environment, you can run the entire pipeline in a single prompt:

Read my Google Calendar for today. For each meeting with external attendees,
enrich every attendee's company and profile using Crustdata. Pull firmographics,
funding, headcount growth, hiring signals, tech stack, and web traffic for each
company. Pull current role, seniority, tenure, and past companies for each
person. Check for recent funding rounds, executive hires, job posting spikes,
and social posts from decision makers at each company.

Score each account 1-100 against this ICP:
- B2B SaaS, 200-2,000 employees
- Series B or later
- Hiring sales or marketing roles
- Using Salesforce or HubSpot
- Headcount growth above 15% in 6 months
- HQ in North America or Western Europe

Write each brief to a markdown file in ./briefs/YYYY-MM-DD/.
Post a summary of all briefs to the #sales-prep Slack channel

Read my Google Calendar for today. For each meeting with external attendees,
enrich every attendee's company and profile using Crustdata. Pull firmographics,
funding, headcount growth, hiring signals, tech stack, and web traffic for each
company. Pull current role, seniority, tenure, and past companies for each
person. Check for recent funding rounds, executive hires, job posting spikes,
and social posts from decision makers at each company.

Score each account 1-100 against this ICP:
- B2B SaaS, 200-2,000 employees
- Series B or later
- Hiring sales or marketing roles
- Using Salesforce or HubSpot
- Headcount growth above 15% in 6 months
- HQ in North America or Western Europe

Write each brief to a markdown file in ./briefs/YYYY-MM-DD/.
Post a summary of all briefs to the #sales-prep Slack channel

Read my Google Calendar for today. For each meeting with external attendees,
enrich every attendee's company and profile using Crustdata. Pull firmographics,
funding, headcount growth, hiring signals, tech stack, and web traffic for each
company. Pull current role, seniority, tenure, and past companies for each
person. Check for recent funding rounds, executive hires, job posting spikes,
and social posts from decision makers at each company.

Score each account 1-100 against this ICP:
- B2B SaaS, 200-2,000 employees
- Series B or later
- Hiring sales or marketing roles
- Using Salesforce or HubSpot
- Headcount growth above 15% in 6 months
- HQ in North America or Western Europe

Write each brief to a markdown file in ./briefs/YYYY-MM-DD/.
Post a summary of all briefs to the #sales-prep Slack channel

Claude Code reads the calendar, calls the Crustdata MCP tools for each company and person, scores the enriched data against your ICP, and writes the output to disk. If it needs to build a helper script for calendar access or Slack delivery, it writes that too.

Save It as a Reusable Skill

Running this from a prompt every morning works, but saving it as a skill means any team member can invoke it without re-prompting. Create a file in your project:

# /daily-precall-research

## About Us
[Your company, what you sell, who you sell to]

## ICP Definition
- B2B SaaS companies with 200 to 2,000 employees
- Series B or later funding
- Actively hiring sales or marketing roles (5+ open positions)
- Using Salesforce or HubSpot as their CRM
- Headcount growth above 15% in the last 6 months
- Headquartered in North America or Western Europe

## Workflow
1. Read today's Google Calendar meetings
2. Filter to meetings with external attendees. Skip attendees on personal
   email domains (gmail.com, outlook.com, yahoo.com, etc.)
3. For each company domain, enrich using Crustdata: firmographics, funding
   history, headcount growth (3, 6, 12-month), hiring signals, tech stack,
   web traffic
4. For each attendee, enrich using Crustdata: current role, seniority,
   tenure, past companies, recent social posts
5. Score each account 1-100 against the ICP definition above. For each
   criterion, state match, partial match, or miss with reasoning
6. Generate three talking points grounded in the enrichment data
7. Flag risk factors (ICP criteria that scored low)

## Output
- Write each brief as markdown to ./briefs/YYYY-MM-DD/company-name.md
- Post a summary table to #sales-prep via Slack webhook

## Tier Definitions
- Tier 1 (80-100): Full brief with personalized talking points
- Tier 2 (50-79): Standard brief with key signals highlighted
- Tier 3 (below 50): Quick summary

# /daily-precall-research

## About Us
[Your company, what you sell, who you sell to]

## ICP Definition
- B2B SaaS companies with 200 to 2,000 employees
- Series B or later funding
- Actively hiring sales or marketing roles (5+ open positions)
- Using Salesforce or HubSpot as their CRM
- Headcount growth above 15% in the last 6 months
- Headquartered in North America or Western Europe

## Workflow
1. Read today's Google Calendar meetings
2. Filter to meetings with external attendees. Skip attendees on personal
   email domains (gmail.com, outlook.com, yahoo.com, etc.)
3. For each company domain, enrich using Crustdata: firmographics, funding
   history, headcount growth (3, 6, 12-month), hiring signals, tech stack,
   web traffic
4. For each attendee, enrich using Crustdata: current role, seniority,
   tenure, past companies, recent social posts
5. Score each account 1-100 against the ICP definition above. For each
   criterion, state match, partial match, or miss with reasoning
6. Generate three talking points grounded in the enrichment data
7. Flag risk factors (ICP criteria that scored low)

## Output
- Write each brief as markdown to ./briefs/YYYY-MM-DD/company-name.md
- Post a summary table to #sales-prep via Slack webhook

## Tier Definitions
- Tier 1 (80-100): Full brief with personalized talking points
- Tier 2 (50-79): Standard brief with key signals highlighted
- Tier 3 (below 50): Quick summary

# /daily-precall-research

## About Us
[Your company, what you sell, who you sell to]

## ICP Definition
- B2B SaaS companies with 200 to 2,000 employees
- Series B or later funding
- Actively hiring sales or marketing roles (5+ open positions)
- Using Salesforce or HubSpot as their CRM
- Headcount growth above 15% in the last 6 months
- Headquartered in North America or Western Europe

## Workflow
1. Read today's Google Calendar meetings
2. Filter to meetings with external attendees. Skip attendees on personal
   email domains (gmail.com, outlook.com, yahoo.com, etc.)
3. For each company domain, enrich using Crustdata: firmographics, funding
   history, headcount growth (3, 6, 12-month), hiring signals, tech stack,
   web traffic
4. For each attendee, enrich using Crustdata: current role, seniority,
   tenure, past companies, recent social posts
5. Score each account 1-100 against the ICP definition above. For each
   criterion, state match, partial match, or miss with reasoning
6. Generate three talking points grounded in the enrichment data
7. Flag risk factors (ICP criteria that scored low)

## Output
- Write each brief as markdown to ./briefs/YYYY-MM-DD/company-name.md
- Post a summary table to #sales-prep via Slack webhook

## Tier Definitions
- Tier 1 (80-100): Full brief with personalized talking points
- Tier 2 (50-79): Standard brief with key signals highlighted
- Tier 3 (below 50): Quick summary

A team member types /daily-precall-research in Claude Code and the workflow runs end to end. Claude Code handles the calendar integration, the API calls, the scoring logic, and the file output. If your environment changes (different CRM, different Slack channel, different ICP), you edit the skill file, not a codebase.

Schedule It to Run Automatically

Save the prompt to a file and add a cron entry to run it before your team's first call block:

0 7 * * 1-5 cd /path/to/project && claude --prompt "$(cat prompts/daily-research.txt)" --allowedTools "mcp*"

0 7 * * 1-5 cd /path/to/project && claude --prompt "$(cat prompts/daily-research.txt)" --allowedTools "mcp*"

0 7 * * 1-5 cd /path/to/project && claude --prompt "$(cat prompts/daily-research.txt)" --allowedTools "mcp*"

This generates briefs for every meeting on the day's calendar before 7 AM. Most meetings are booked at least a day in advance, so a morning batch covers the majority of cases.

For real-time coverage, Google Calendar supports push notifications that fire a webhook whenever a meeting is created or updated. Ask Claude Code to build a lightweight webhook receiver that listens for new events and triggers the enrichment workflow. That way briefs are generated within minutes of a meeting being booked, regardless of when it was scheduled.

Set Up Signal Monitoring with Watchers

Use the Watcher API to monitor accounts that are already on your calendar. If a company raises a funding round or hires a new VP of Sales the night before your meeting, the webhook fires and the brief updates with the new signal. This layers real-time signal coverage on top of the calendar trigger, so the brief reflects changes that happened after the initial enrichment.

What Claude Code Needs from You

Claude Code writes the integration code, but there are a few things it cannot figure out on its own:

Natural-Language ABM Scoring with Claude

Traditional ABM scoring uses rigid point systems: 10 points for more than 500 employees, 5 points for the SaaS industry, 3 points for a recent funding round. These models encode assumptions as fixed numbers, and they break in predictable ways. A 200-person company that just raised a Series C and is growing headcount at 40 percent might be a better fit than a 2,000-person company with flat headcount and no recent funding, but a rigid model scores the larger company higher because the employee count bracket is worth more points.

Most teams face a choice between full-stack ABM platforms that cost $50,000 or more per year and manual CRM-based scoring that relies on whatever data reps enter. Claude with a live data provider fills the gap between those two options.

How Claude Scores with Natural Language

Instead of assigning point values, you define your ICP criteria in plain English and let Claude evaluate each account against them. The prompt looks like this:

Score this account 1-100 based on the following ICP criteria.
For each criterion, state whether the account matches, partially matches,
or does not match, and explain why. Then provide an overall score
with a 2-3 sentence summary.

ICP Criteria:
1. B2B SaaS company with 200-2,000 employees
2. Series B or later funding stage
3. Actively hiring sales or marketing roles (5+ open positions)
4. Using Salesforce or HubSpot as primary CRM
5. Headcount growth above 15% in the last 6 months
6. Headquartered in North America or Western Europe
7. Not acquired or in acquisition process

Account Data:
[paste enriched company data from Crustdata]

Score this account 1-100 based on the following ICP criteria.
For each criterion, state whether the account matches, partially matches,
or does not match, and explain why. Then provide an overall score
with a 2-3 sentence summary.

ICP Criteria:
1. B2B SaaS company with 200-2,000 employees
2. Series B or later funding stage
3. Actively hiring sales or marketing roles (5+ open positions)
4. Using Salesforce or HubSpot as primary CRM
5. Headcount growth above 15% in the last 6 months
6. Headquartered in North America or Western Europe
7. Not acquired or in acquisition process

Account Data:
[paste enriched company data from Crustdata]

Score this account 1-100 based on the following ICP criteria.
For each criterion, state whether the account matches, partially matches,
or does not match, and explain why. Then provide an overall score
with a 2-3 sentence summary.

ICP Criteria:
1. B2B SaaS company with 200-2,000 employees
2. Series B or later funding stage
3. Actively hiring sales or marketing roles (5+ open positions)
4. Using Salesforce or HubSpot as primary CRM
5. Headcount growth above 15% in the last 6 months
6. Headquartered in North America or Western Europe
7. Not acquired or in acquisition process

Account Data:
[paste enriched company data from Crustdata]

Claude reads the enriched data, evaluates each criterion, and produces output like this:

ICP Score: 82/100 (Tier 1)

1. Employee count: Match. 450 employees, within the 200-2,000 range.

2. Funding stage: Match. Series C ($40M raised in January 2026, $85M total).

3. Hiring activity: Match. 12 sales roles and 4 marketing roles posted in the last 30 days.

4. CRM: Partial match. Tech stack detection shows Salesforce, but the specific CRM version is unconfirmed.

5. Headcount growth: Match. 22% growth over the last 6 months, with sales headcount up 35%.

6. Geography: Match. HQ in Austin, TX.

7. Acquisition status: Match. Privately held, no acquisition signals.

Summary: Strong ICP fit. The company matches on all core criteria, with particularly strong hiring velocity in sales. The recent Series C gives them budget for new tooling. The only uncertainty is CRM confirmation, but Salesforce detection in the tech stack is a strong signal.

A rep reading this brief knows exactly why the account scored high and can tailor their prep accordingly. Companies with predictive scoring report 23 to 30 percent faster sales cycles according to Demandbase, and the natural-language approach delivers a similar outcome without a dedicated scoring platform.

Tiering Accounts from Scores

Scores map to tiers that drive prep depth:

Tiering helps reps allocate their limited prep time. 82 percent of top-performing sellers always research before calls compared to 49 percent of the rest, according to HubSpot. When the brief and score are already generated, even a Tier 3 account gets more prep than most reps would do manually.

Iterating on Your Scoring Criteria

When you update your ICP definition, you change the prompt text and Claude immediately scores differently. There is no scoring model to rebuild, no engineering sprint to update point weights, and no platform configuration to modify. If you notice Tier 1 accounts are not converting, tighten the criteria. If too few accounts qualify, loosen them. The feedback loop is a prompt edit, not a project.

What the Pre-Call Brief Looks Like

Here is a complete scored brief for a fictional account, showing how the six data categories plus scoring plus talking points come together in a single output. This is the final product your reps receive.

DataFlow Systems | ICP Score: 78/100 (Tier 2) Meeting: May 7, 2026 at 2:00 PM with Sarah Chen (VP Engineering) and Mark Torres (Head of RevOps)

Scoring Reasoning Matches on funding stage (Series B, $28M raised), headcount growth (18% over 6 months), and active hiring (8 engineering roles, 3 sales roles posted). Missed on employee count (currently 165, below the 200 minimum) and CRM (uses Pipedrive, not Salesforce or HubSpot). Geography matches (HQ in Denver, CO).

Company Snapshot

People in This Meeting

• Sarah Chen | VP Engineering | 14 months at DataFlow | Previously at Stripe (Sr. Engineering Manager, 3 years). Posted on LinkedIn last week about scaling data pipelines for enterprise clients.

• Mark Torres | Head of RevOps | 8 months at DataFlow | Previously at Gong (RevOps Manager, 2 years). No recent social activity.

Suggested Talking Points

• DataFlow is hiring 3 sales roles while growing engineering at 2x that rate, which suggests they are building product capacity ahead of a sales push. Ask whether they are planning to expand their outbound motion.

• Sarah Chen's recent post about enterprise data pipelines may indicate an upmarket move. If so, their data infrastructure needs will grow with their deal sizes.

• Mark Torres joined 8 months ago from Gong, a company known for strong RevOps tooling. He likely has opinions about data quality in the CRM and may be the internal champion for a data enrichment solution.

Risk Factors

• Employee count (165) is below the 200 minimum for Tier 1. The company may not have the internal resources for a full implementation without support.

• Uses Pipedrive instead of Salesforce or HubSpot, which means standard CRM integrations may not apply.

This brief took Claude roughly 45 seconds to generate using MCP-connected enrichment. A rep doing this manually would spend 30 to 40 minutes across five or six browser tabs and still miss the headcount growth rates, the job posting velocity, and the attendee's recent social activity.

Where the Brief Goes

The brief is only useful if it reaches the rep without them asking for it. Four delivery options, from simplest to most automated:

Slack. Post each morning's briefs to a dedicated channel like #sales-prep, or send per-meeting DMs 30 minutes before each call. In Claude Code, use a Slack webhook to post the markdown output directly. In Claude Desktop, copy and paste into Slack manually.

CRM. Write the brief to the account notes field in Salesforce or HubSpot. If you are using Claude Code with a CRM MCP server or API, this step is part of the automated pipeline. The brief becomes part of the account record and is visible to anyone who opens the account before the call.

Email. Send the brief to the rep's inbox as a morning digest. A simple SMTP call at the end of the Python script handles this. Less elegant than Slack but universally accessible.

Shared files. Write briefs to a shared Google Drive folder or Notion database organized by date. Useful for teams that want a searchable archive of past research.

The delivery method matters less than the automation. If the rep has to request the brief or go find it, most will skip it on busy days. When the brief arrives automatically, every meeting gets at least a Tier 3 summary, and the consistency of preparation across the team improves even if no individual rep changes their behavior.

What Comes Next

Pre-call research and account scoring typically run as separate workflows, with research in one tool and scoring in another while the rep stitches them together manually before each call. This workflow collapses both into a single Claude conversation that pulls live data from a provider, reasons over it against your ICP criteria, and outputs a brief the rep can act on.

If you want to see the MCP workflow in action, book a demo and we will walk through a live calendar-triggered research and scoring session. If you prefer to start building, the API documentation covers every endpoint used in this guide, and the Crustdata MCP server is available for both Claude Desktop and Claude Code.