Client Discovery

Take a client's raw list of requests and produce a structured scoping breakdown with categories, hours, pricing, dependencies, and recommended phases.

When to use

• Client sends a list of automation/AI tasks they want built
• "analyze requests from [client]"
• "scope this project"
• "estimate hours for [client]"
• "create proposal breakdown"
• Before a discovery/scoping call — to come prepared with estimates

Dependencies

• Other skills: query-leads (CRM data), client-workspace (for shared docs)
• External: none (this is an analysis skill, no scripts)

How to execute

Step 1: Gather inputs

1. Client's raw request list — from TG, email, call notes, or shared doc
2. Client's tech stack — CRM, ATS, tools they use (from CRM notes or questionnaire)
3. Company context — from CRM: size, industry, budget signals

Step 2: For each request item, analyze

For every item in the client's list, produce:

Field	Description
Name	Short name (2-5 words)
Category	agent / automation / integration / knowledge-base / product
What client wants	Plain language — what outcome they expect
What needs to be built	Technical: APIs, triggers, LLM prompts, data flows
Key questions	What we need to clarify before building
Integrations	Which tools/APIs: CRM, ATS, LinkedIn, Bluedot, etc.
Complexity	low (prompt eng, 4-6h) / medium (integration, 6-10h) / high (multi-system, 10-15h)
Hours estimate	Range: low-high
Dependencies	Other items that should be built first

Step 3: Prioritize

Group items into:

1. Quick wins (low complexity, high impact) — do first, show value fast
2. High ROI (medium complexity, core business impact) — second phase
3. Strategic (high complexity, long-term value) — third phase
4. Can skip / already exists — tools like NotebookLM that solve it out of the box

Step 4: Check for off-the-shelf solutions

Before estimating custom build hours, check if an existing tool already does it:

• NotebookLM for knowledge bases
• Zapier/Make for simple automations
• Existing SaaS (Fireflies for transcription, Clay for signal tracking, etc.)

Flag these as "buy vs build" decisions with the client.

Step 5: Produce summary table

| # | Request | Hours | $ | Phase | Notes |
|---|---------|-------|---|-------|-------|
| 1 | Job posting AI | 4-6 | 400-600 | Quick win | Few-shot prompting |
| 2 | CRM automation | 8-12 | 800-1200 | Phase 2 | Needs API access |
...
| TOTAL | | 60-90 | $6K-9K | | |

Step 6: Generate discovery questions

Based on gaps in the analysis, generate a pre-call questionnaire:

• Questions about tech stack and data
• Questions about priorities and budget
• Questions about team and users

Use client-workspace skill to create a shared Google Doc with these questions.

Rate Card

Service	Rate
Consulting / implementation	$100/hr, 15-min increments ($25 min)
Quick win (4-6h)	$400-600
Medium project (6-12h)	$600-1200
Complex project (10-15h)	$1000-1500

Output Format

The analysis should be saved as:

1. CRM activity — summary in activities.csv
2. Shared doc — if questionnaire created, in client's Discovery folder
3. Text summary — shown to Ivan for review before the call

Checklist

• All client request items analyzed and categorized
• Hours and pricing estimated for each item
• Off-the-shelf alternatives checked
• Items prioritized into phases
• Discovery questions generated for unknowns
• Summary table produced
• CRM activity logged

Examples

Client J (2026-03-04)

Client: Diana Prince, Client J (IT recruiting, 22 years experience) Stack: Recruitee (ATS), Streak (CRM), Bluedot (call recording) 11 automation requests → analyzed into 4 blocks:

• Block 1: CRM & Sales (18-27h, $1.8-2.7K)
• Block 2: Recruiting process (16-22h, $1.6-2.2K)
• Block 3: Knowledge base (13-20h, $1.3-2K) — partly solved by NotebookLM
• Block 4: Client-facing products (14-22h, $1.4-2.2K) Total: 61-91h, $X-YK

Related skills

• client-workspace — create shared docs for discovery
• call-prep — prepare for the discovery/scoping call
• query-leads — CRM data lookup