Bionic Business
Posts
Issue #91: AI Automations for Boring But Valuable Work

Issue #91: AI Automations for Boring But Valuable Work

Sam Woods
December 15, 2025

Good morning.

Here's what I've noticed from the hundreds of messages I get each month:

Most entrepreneurs are stuck in a loop.

They know AI can help. They've seen the possibilities. But they're paralyzed by complexity and every new shiny release.

So let's break that loop today.

I'm giving you two automations you can implement in a single work session.

Each one solves a real problem, creates immediate value, and runs without your attention.

These are a few of the many I've seen work across dozens of businesses.

One more thing: I'm including Claude Skills setups for each automation. If you haven't used Skills yet, they're Anthropic's new way to give Claude specialized capabilities through structured instructions and scripts. Think of them as custom training for specific tasks. They load only when needed, keeping Claude fast while making it dramatically better at specialized work.

— Sam

IN TODAY’S ISSUE 🤖

Quick Win #1: Meeting Intelligence Extractor (turns every call into organized action items)
Quick Win #2: Lead Qualification Accelerator (scores and prioritizes leads while you sleep)
Implementation Options: Manual (20 min) or Automated (45-60 min) for each

Let’s get into it.

Quick Win #1: The Meeting Intelligence Extractor

Never lose another action item, decision, or follow-up from your calls

The Problem: You finish a 45-minute client call. You remember the energy was good. But what exactly did you commit to? What did they commit to? What decisions were made?

Two days later, you're scrambling to reconstruct the conversation from scattered notes.

The Solution: A system that transforms meeting transcripts into organized intelligence—action items, decisions, commitments, and follow-ups—in seconds.

Time to implement: 20 minutes (manual) or 60 minutes (automated)
Tools needed: Claude/GPT-4 + your meeting recordings (Fathom, Otter, Grain, or built-in Zoom transcription)
Saves you: 5-8 hours per week of meeting follow-up chaos

Option 1: The Manual Method (20 minutes)

The Meeting Intelligence Prompt

Copy this exactly:

MEETING INTELLIGENCE EXTRACTION

You are a precise executive assistant specializing in extracting actionable intelligence from meeting transcripts.

Analyze this transcript and extract:

1. DECISIONS MADE
   - What was decided (be specific)
   - Who made or approved the decision
   - Any conditions or dependencies

2. ACTION ITEMS
   For each action item:
   - Task description (specific and actionable)
   - Owner (who committed to do this)
   - Deadline (stated or implied)
   - Dependencies (what needs to happen first)

3. COMMITMENTS GIVEN
   - What we committed to them
   - What they committed to us
   - Timeline for each commitment

4. FOLLOW-UPS REQUIRED
   - Items needing clarification
   - Information to be sent
   - Meetings to be scheduled

5. KEY INSIGHTS
   - Unstated concerns detected
   - Opportunities mentioned
   - Risks or objections raised

6. NEXT MEETING PREP
   - Topics for next conversation
   - Questions to prepare

Format with clear headers. Be specific—vague action items are useless.
Use their exact words for important commitments.

[PASTE YOUR MEETING TRANSCRIPT HERE]

Manual Implementation Steps

Export transcript from your meeting tool (most download as .txt or .doc)
Open Claude (Pro or Team tier for best results)
Paste the prompt + transcript
Copy output into your project management tool
Assign action items to team members within 30 minutes of call

Pro Tip: Run this within 1 hour of your call. Fresh context helps you catch anything the AI missed.

Option 2: Using Claude Skills

Claude Skills let you create specialized capabilities that Claude loads only when relevant. Here's how to build a Meeting Intelligence Skill:

Creating Your Custom Skill

In Claude, go to Settings > Capabilities > Skills. Create a new custom skill:

SKILL.md contents:

---
name: meeting-intelligence
description: Extract action items, decisions, and commitments from meeting transcripts. Use when processing meeting recordings or notes.
---

## Meeting Intelligence Extraction Protocol

When given a meeting transcript:

### Phase 1: Quick Scan
- Identify all participants and their roles
- Note the meeting's stated purpose
- Flag any time-sensitive items

### Phase 2: Deep Extraction
Extract each category with precision:

**Decisions:** Look for phrases like "let's go with," "we've decided," "agreed," "confirmed"
**Action Items:** Look for "I'll," "we'll," "you'll," "can you," "please," "by [date]"
**Commitments:** Track promises made in both directions
**Concerns:** Note hesitations, "but," "however," qualified agreements

### Phase 3: Prioritization
Rank action items by:
1. Deadline proximity
2. Dependency on other items  
3. Business impact
4. Owner bandwidth

### Output Format
Use structured headers. Include owner names in ALL CAPS for scannability.
Include exact quotes for critical commitments.
Flag any ambiguous items requiring clarification.

Once saved, Claude will automatically use this skill when you mention meeting transcripts or ask for meeting analysis.

Option 3: Gumloop Automation

Workflow Overview:

Meeting Recording Upload → Transcript Extraction → AI Analysis → Notion/Asana Update → Slack Summary

Step-by-Step Setup:

Node 1: Trigger

Type: Webhook or Manual Trigger
Purpose: Receives meeting recording or transcript
Alternative: Email trigger if your meeting tool emails transcripts

Node 2: Transcript Handling

Type: Conditional Router
If audio file: Route to Whisper/transcription node
If transcript: Pass directly to analysis

Node 3: AI Analysis Node

Type: Claude or GPT node
Model: Claude Sonnet 4.5 (recommended for accuracy)
Prompt: Use the Meeting Intelligence prompt above
Output: JSON structure for downstream processing

Parse the AI response into structured JSON:
{
  "decisions": [...],
  "action_items": [...],
  "commitments": [...],
  "follow_ups": [...],
  "next_meeting_prep": [...]
}

Node 4: Task Creation Loop

Type: Iterator/Loop node
Purpose: For each action item, create a task
Integration: Notion, Asana, Linear, or Monday.com

Node 5: Slack Summary

Type: Slack integration
Channel: Your team channel or DM
Format:

📋 Meeting Intelligence: [Meeting Title]

📌 Decisions Made: [count]
✅ Action Items: [count] 
⏰ Urgent Follow-ups: [count]

Top 3 Action Items:
1. [ACTION] - Owner: [NAME] - Due: [DATE]
2. [ACTION] - Owner: [NAME] - Due: [DATE]
3. [ACTION] - Owner: [NAME] - Due: [DATE]

Full breakdown: [Link to Notion page]

Cost: ~$37/month (Gumloop Solo plan with 10,000 credits)

Option 4: Lindy AI Agent

Lindy excels at autonomous, context-aware tasks. Here's the agent setup:

Agent Name: "Meeting Intelligence Agent"

Agent Type: Document Processing + Task Automation

Configuration:

You are my Meeting Intelligence Agent. Your job is to process meeting transcripts and ensure nothing falls through the cracks.

TRIGGERS:
- New transcript uploaded to Google Drive folder: "Meeting Recordings"
- Email received with subject containing "transcript" or "meeting notes"
- Slack message with meeting recording link

PROCESSING WORKFLOW:

1. EXTRACT all action items, decisions, and commitments
2. CREATE tasks in [Asana/Notion/your tool]:
   - Set owner based on names in transcript
   - Set due dates based on mentioned timelines
   - Add meeting context to task description
3. SEND summary to #meetings Slack channel
4. FLAG any commitments without clear owners or deadlines
5. SCHEDULE reminder for any follow-ups mentioned

QUALITY RULES:
- Never create vague tasks like "follow up on project"
- Always include specific context
- Quote exact words for important commitments
- Default deadline: 7 days if not specified
- Tag urgent items if deadline < 48 hours

ESCALATION:
If transcript is unclear or missing key info, DM me asking for clarification rather than guessing.

Lindy Integrations Required:

Google Drive or Dropbox (for transcript storage)
Slack (for summaries)
Asana, Notion, or your project management tool
Gmail (optional, for transcript emails)

Cost: Starting at $29.99/month

Why This Works

Every meeting contains valuable intelligence that decays rapidly. Within 24 hours, you've forgotten 70% of the details. Within a week, 90%.

This system captures everything while it's fresh and routes it to where action happens.

The compound effect: After one month, you'll have:

Zero dropped commitments
Clearer accountability across your team
A searchable archive of every decision made
Dramatically improved client relationships (you never forget what you promised)

Real Example Output

From a 38-minute client strategy call:

DECISIONS MADE:

Launch email sequence to segment A first (approved by Sarah)
Budget: $2,500/month for paid ads starting January 15
Weekly check-ins moving to Mondays at 10am

ACTION ITEMS:

JOHN: Send revised proposal by Friday 5pm
SARAH: Approve final ad creative by Wednesday EOD
MARCUS: Set up tracking pixels before Jan 10

COMMITMENTS GIVEN:

We committed to: "First draft of landing page by next Monday"
They committed to: "Internal team review within 48 hours of receiving materials"

FOLLOW-UPS REQUIRED:

Get Sarah's assistant's email for calendar coordination
Clarify: Does "budget approval" mean from marketing or finance?

Quick Win #2: The Lead Qualification Accelerator

Score and prioritize leads automatically so you only talk to buyers

The Problem: Leads come in. Some are ready to buy. Some will never buy. You can't tell the difference without a 30-minute discovery call.

You're wasting hours talking to tire-kickers while real buyers wait.

The Solution: An AI system that scores leads instantly, identifies high-priority opportunities, and routes them correctly—before you spend a minute on them.

Time to implement: 20 minutes (manual) or 60 minutes (automated)
Tools needed: Claude + your lead data
Saves you: 10+ hours per week on unqualified leads

Option 1: The Manual Method (20 minutes)

The Lead Qualification Prompt

LEAD QUALIFICATION ANALYSIS

You are an expert sales qualification specialist. Analyze this lead data and score their likelihood to buy.

MY IDEAL CUSTOMER PROFILE:
- Company size: [your ICP]
- Industry: [your target industries]
- Budget range: [typical deal size]
- Key pain points: [problems you solve]
- Decision maker: [typical buyer title]

LEAD INFORMATION:
[Paste lead data: form submission, email inquiry, LinkedIn profile, company info]

SCORE AND ANALYZE:

1. FIT SCORE (1-10)
   How well do they match your ICP?
   - Company size fit
   - Industry fit
   - Role/title fit
   - Budget indicators

2. INTENT SCORE (1-10)
   How ready are they to buy?
   - Urgency signals in their language
   - Specificity of request
   - Research depth (what pages they visited, questions asked)
   - Timeline mentions

3. QUALIFICATION SIGNALS
   Green Flags (buy signals):
   - [List specific positive indicators]
   
   Red Flags (concerns):
   - [List specific negative indicators]

4. PRIORITY CLASSIFICATION
   HOT (8-10 combined): Schedule immediately
   WARM (5-7): Nurture sequence
   COLD (1-4): Low priority or disqualify

5. RECOMMENDED APPROACH
   - Best channel to reach them
   - Key talking points based on their signals
   - Questions to ask in discovery
   - Potential objections to prepare for

6. PERSONALIZATION DATA
   Information to reference that shows you researched them:
   - Recent company news
   - Specific challenges mentioned
   - Relevant case study to share

Manual Workflow:

Lead comes in via form/email
Pull available data (LinkedIn, company website, form responses)
Paste into Claude with your ICP details
Get instant score and routing recommendation
Act accordingly (call immediately, add to nurture, deprioritize)

Option 2: Claude Skills Setup

SKILL.md for Lead Qualification:

---
name: lead-qualifier
description: Score and qualify inbound leads against ICP criteria. Use when processing new leads, form submissions, or sales inquiries.
---

## Lead Qualification Framework

### Required Context
Before qualifying, need:
- Ideal Customer Profile criteria
- Historical conversion data (what did past winners look like?)
- Current lead capacity (how many can sales handle?)

### Scoring Methodology

**Fit Score (50% weight):**
| Criteria | Points |
|----------|--------|
| Exact ICP match | 10 |
| Adjacent industry | 7 |
| Company size match | 10 |
| Budget indicators | 10 |
| Decision maker confirmed | 10 |

**Intent Score (50% weight):**
| Signal | Points |
|--------|--------|
| Explicit budget mentioned | 15 |
| Timeline stated | 10 |
| Specific pain point described | 10 |
| Competitor mentioned | 5 |
| Pricing page visited | 5 |
| Multiple touchpoints | 5 |

### Qualification Tiers

**HOT (80-100 points):** 
Route to sales immediately. Respond within 1 hour.
These leads have clear fit, budget, and urgency.

**WARM (50-79 points):**
Add to high-touch nurture sequence.
Schedule follow-up within 24-48 hours.

**COOL (25-49 points):**
Automated nurture sequence.
Monthly check-in only.

**COLD (<25 points):**
Low priority. Automated content only.
Re-evaluate if behavior changes.

### Enrichment Actions
For any lead scoring 50+, automatically:
1. Pull LinkedIn company page data
2. Check for recent news/funding
3. Identify mutual connections
4. Find relevant case study match

Option 3: Gumloop Automation

Build an automated lead scoring pipeline:

Workflow Structure:

Node 1: Trigger

Type: Webhook from your form tool (Typeform, HubSpot, etc.)
Alternative: Email trigger for inquiry emails
Captures: All form data

Node 2: Data Enrichment

Type: HTTP request to enrichment API
Services: Clearbit, Apollo, or LinkedIn API
Pulls: Company size, industry, funding, tech stack
Cost: Varies by enrichment provider

Node 3: Web Research (Optional)

Type: Web scraper node
Target: Lead's company website + recent news
Extracts: Key context for personalization

Node 4: AI Scoring

Type: Claude node
Prompt: Lead Qualification prompt with enriched data
Output: JSON with scores and classification

{
  "fit_score": 8,
  "intent_score": 7,
  "total_score": 75,
  "classification": "WARM",
  "green_flags": ["Mentioned specific pain point", "Decision maker role"],
  "red_flags": ["No budget mentioned", "Long timeline"],
  "recommended_action": "Add to high-touch nurture, follow up in 24 hours",
  "personalization_notes": "Reference their recent Series B, mention similar fintech case study"
}

Node 5: CRM Update

Type: HubSpot/Salesforce/Pipedrive integration
Actions:
- Create/update contact record
- Set lead score field
- Add qualification notes
- Assign to appropriate owner based on score

Node 6: Routing Logic

Type: Conditional router

If classification = "HOT":
  → Slack alert to sales team
  → Create urgent task in CRM
  → Send immediate personalized auto-response
  
If classification = "WARM":
  → Add to high-touch nurture sequence
  → Schedule follow-up task for 24 hours
  → Send educational content email

If classification = "COOL" or "COLD":
  → Add to automated nurture
  → No immediate action required

Node 7: Slack Notification (for HOT leads)

Immediate alert with:
- Lead name and company
- Score breakdown
- Key talking points
- Link to CRM record

Option 4: Lindy AI Agent

Agent Name: "Lead Intelligence Agent"

Configuration:

You are my Lead Intelligence Agent. Your job is to qualify every inbound lead instantly and ensure we never miss a hot opportunity.

TRIGGERS:
- New form submission on website
- New email to sales@ inbox
- New LinkedIn connection request mentioning business inquiry
- New HubSpot contact created

IMMEDIATE ACTIONS:

1. ENRICH the lead:
   - Pull company data from LinkedIn/Clearbit
   - Check for recent news or funding
   - Identify decision maker status
   - Look for mutual connections

2. SCORE the lead against our ICP:
   [Insert your ICP criteria]
   
   Calculate Fit Score (1-10) and Intent Score (1-10)

3. CLASSIFY and ROUTE:
   - HOT (16-20 combined): Slack alert @sales-team immediately
   - WARM (10-15): Add to priority nurture, create follow-up task
   - COOL (5-9): Automated nurture sequence
   - COLD (1-4): Low priority list

4. PERSONALIZE outreach:
   - Draft personalized first response email
   - Identify relevant case study to share
   - Note specific talking points for discovery call

5. UPDATE CRM:
   - Log all qualification data
   - Set appropriate tags and scores
   - Assign owner based on territory/capacity

QUALITY RULES:
- Never let a HOT lead sit more than 1 hour without contact
- Always include personalization data in handoff
- Flag any leads from target accounts immediately
- If unclear about qualification, default to WARM (don't miss opportunities)

Why This Works

Research shows responding to leads within 5 minutes makes you 21x more likely to qualify them. But you can't respond to every lead in 5 minutes—unless you know which ones matter.

This system creates instant triage. Hot leads get immediate attention. Everyone else goes into appropriate sequences.

The compound effect:

Sales team focuses only on qualified opportunities
Response time to hot leads drops from hours to minutes
Conversion rate improves as fit improves
Pipeline quality increases, forecasting becomes reliable

Real Example Output

Lead: Sarah Chen, VP Marketing

FIT SCORE: 9/10

Company size: 50-200 employees ✓
Industry: B2B SaaS ✓
Role: Decision maker ✓
Recent funding: Series A ($12M in March) ✓

INTENT SCORE: 8/10

Mentioned specific pain point: "Can't scale content production"
Viewed pricing page 3x in past week
Downloaded 2 resources
Timeline: "Looking to implement Q1"

CLASSIFICATION: HOT (17/20)

GREEN FLAGS:

Decision maker with budget authority
Specific, urgent pain point matching our solution
Active research behavior
Clear timeline

RED FLAGS:

None significant

RECOMMENDED ACTION: Call within 1 hour. Lead with content scaling case study from similar SaaS company. Reference their recent funding—they likely have budget allocated for growth initiatives.

PERSONALIZATION DATA:

Recent company news: Launched new product line in October
LinkedIn shows: Previously at competitor's client company
Mutual connection: James Park (your investor)
Relevant case study: DataFlow Inc (similar size, 3x content output in 60 days)

That should do it for this issue.

The best automations are invisible.

They run quietly in the background. They catch things you'd miss. They do the work you shouldn't be doing.

Each Quick Win I've shared today solves a specific problem:

Meetings become organized action (instead of forgotten promises)
Content patterns become visible (instead of mysterious)
Lead quality becomes obvious (instead of discovered too late)

Pick one. Implement it this week. Feel the relief of something just... working.

Then come back and build the other two.

One more thing on Claude Skills:

If you haven't explored Skills yet, start simple. The pattern is always the same:

Identify a task you do repeatedly
Document your exact process and criteria
Create a SKILL.md that teaches Claude your method
Let it load automatically when relevant

They're the fastest way to make Claude genuinely useful for your specific workflows.

ChatGPT does not yet have an equivalent feature but perhaps it will soon.

Until next time,
Sam Woods
The Editor