|
1
|
--- |
|
2
|
name: ai-sdr |
|
3
|
description: Find relevant companies and leads for B2B sales with ICP definition and qualification frameworks. |
|
4
|
--- |
|
5
|
|
|
6
|
# AI SDR |
|
7
|
|
|
8
|
Find relevant companies and leads for B2B sales. Define ideal customer profiles, identify target accounts, qualify prospects, and organize research for outreach. |
|
9
|
|
|
10
|
## When to Use |
|
11
|
|
|
12
|
- User wants to find companies that match their ICP |
|
13
|
- User needs to build a prospect list for sales outreach |
|
14
|
- User wants to research target accounts |
|
15
|
- User needs lead qualification analysis |
|
16
|
|
|
17
|
## When NOT to Use |
|
18
|
|
|
19
|
- Writing outreach sequences (use sdr-outreach or cold-email-writer skills) |
|
20
|
- Recruiting candidates (use ai-recruiter skill) |
|
21
|
- General market research (use deep-research skill) |
|
22
|
|
|
23
|
## Methodology |
|
24
|
|
|
25
|
### Step 0: Interview the User About Their Business |
|
26
|
|
|
27
|
**Before doing any research, interview the user.** Most users won't give you enough context unprompted — they'll say "find me customers" without explaining what they sell or who buys it. Use multi-option quizzes to make this fast and low-friction. Ask one question at a time. |
|
28
|
|
|
29
|
**Question flow:** |
|
30
|
|
|
31
|
1. **What do you sell?** |
|
32
|
- A) Software / SaaS |
|
33
|
- B) Physical product |
|
34
|
- C) Professional services / consulting |
|
35
|
- D) Marketplace / platform |
|
36
|
- E) Something else (describe) |
|
37
|
|
|
38
|
2. **Who buys it?** |
|
39
|
- A) Other businesses (B2B) |
|
40
|
- B) Consumers (B2C / DTC) |
|
41
|
- C) Both |
|
42
|
|
|
43
|
3. **(If B2B) What size companies?** |
|
44
|
- A) Startups / small teams (1-50 people) |
|
45
|
- B) Mid-market (50-500) |
|
46
|
- C) Enterprise (500+) |
|
47
|
- D) Not sure yet |
|
48
|
|
|
49
|
4. **What's your price point?** |
|
50
|
- A) Under $100/mo |
|
51
|
- B) $100-$1,000/mo |
|
52
|
- C) $1,000-$10,000/mo |
|
53
|
- D) $10,000+/mo or custom pricing |
|
54
|
- E) One-time purchase |
|
55
|
|
|
56
|
5. **Who inside the company makes the buying decision?** |
|
57
|
- A) Engineering / technical (CTO, VP Eng, developers) |
|
58
|
- B) Marketing (CMO, growth, content) |
|
59
|
- C) Sales / revenue (CRO, VP Sales, RevOps) |
|
60
|
- D) Operations / finance (COO, CFO) |
|
61
|
- E) Founder / CEO directly |
|
62
|
- F) Not sure |
|
63
|
|
|
64
|
6. **Do you have existing customers?** |
|
65
|
- A) Yes, paying customers — I can describe who they are |
|
66
|
- B) A few early users / pilots |
|
67
|
- C) No customers yet |
|
68
|
|
|
69
|
7. **(If they have customers) What do your best customers have in common?** (free text — this is the most valuable answer) |
|
70
|
|
|
71
|
8. **Any industries or verticals you're focused on?** (free text or skip) |
|
72
|
|
|
73
|
**If the user provides a detailed prompt upfront**, skip the questions they already answered. Don't re-ask what's obvious. But if key info is missing (who buys, what size, what price), ask before proceeding — the ICP will be wrong without it. |
|
74
|
|
|
75
|
### Step 1: Define ICP (Ideal Customer Profile) |
|
76
|
|
|
77
|
An ICP describes accounts where three things are true: they **can buy** (budget/size fit), they **will buy** (pain exists now), and they **will stay** (retention profile). Pick 6-10 attributes — more than 10 and nothing qualifies. |
|
78
|
|
|
79
|
| Attribute | How to define it | Example | |
|
80
|
|-----------|------------------|---------| |
|
81
|
| Headcount | Hard range, not "SMB" | 50-500 employees | |
|
82
|
| Revenue | Estimate from headcount if private | $10M-$100M ARR | |
|
83
|
| Industry | NAICS/SIC codes or named verticals | SaaS, fintech, digital health | |
|
84
|
| Geography | Where you can legally sell + support | US, UK, Canada | |
|
85
|
| Tech stack | Tools that signal fit (technographics) | Uses Salesforce + Segment + AWS | |
|
86
|
| Funding stage | Proxy for budget + growth pressure | Series A-C, raised in last 18mo | |
|
87
|
| Hiring signals | Job posts reveal priorities | Hiring "RevOps" or "Head of Data" | |
|
88
|
| Negative signals | Disqualifiers — the sharpest filter | <20 employees, agency model, on-prem only | |
|
89
|
|
|
90
|
**ICP vs Buyer Persona:** ICP = which company. Persona = which human inside it. A Series B fintech (ICP) has a VP Eng who cares about velocity and a CFO who cares about cloud spend (two personas, different messaging). |
|
91
|
|
|
92
|
### Step 2: Source Accounts |
|
93
|
|
|
94
|
**Free sources (use `webSearch` + `webFetch`):** |
|
95
|
|
|
96
|
- **Crunchbase** — `site:crunchbase.com "series a" fintech 2025` for funding events |
|
97
|
- **LinkedIn** — `site:linkedin.com/company [industry] "11-50 employees"` (headcount filter leaks into page text) |
|
98
|
- **BuiltWith / Wappalyzer lookups** — `webFetch` a prospect's homepage, then scan source for tech signatures (Segment snippet, Intercom widget, Shopify checkout) |
|
99
|
- **Job boards** — `site:linkedin.com/jobs "[target company]" "data engineer"` reveals what they're building; `site:greenhouse.io` and `site:lever.co` for startup hiring |
|
100
|
- **G2 / Capterra category pages** — companies reviewing competitors are in-market |
|
101
|
- **GitHub orgs** — public repos reveal tech stack and eng team size for dev-tool ICPs |
|
102
|
- **SEC EDGAR** (public cos) — 10-K "Risk Factors" sections list the exact problems they're worried about |
|
103
|
|
|
104
|
**Paid sources the user likely has (shape output for these):** |
|
105
|
|
|
106
|
- **Apollo** (~210M contacts, $49+/mo) — best value for SMB/mid-market, filters on headcount growth + job postings + intent |
|
107
|
- **LinkedIn Sales Navigator** (~1B profiles) — most accurate job-change data, but no email export |
|
108
|
- **ZoomInfo** — strongest US enterprise coverage + intent data (tracks content consumption across the web) |
|
109
|
- **Clay** ($134+/mo) — waterfall enrichment: chains Apollo → Hunter → Cognism to maximize match rate. Best for teams with RevOps capacity. |
|
110
|
- **Cognism** — best EU/UK data + phone-verified mobiles (GDPR-compliant) |
|
111
|
|
|
112
|
### Step 3: Buying Signals (Trigger Events) |
|
113
|
|
|
114
|
Prospects with an active trigger convert 3-5x higher. Rank by signal strength: |
|
115
|
|
|
116
|
| Signal | Why it matters | How to find it | |
|
117
|
|--------|----------------|----------------| |
|
118
|
| New exec in target persona | New VPs buy tools in first 90 days | `site:linkedin.com "[company]" "I'm excited to join"` or Sales Nav job-change alerts | |
|
119
|
| Funding round | Budget just unlocked | Crunchbase, `webSearch: "[company] raises"` | |
|
120
|
| Hiring spike in relevant role | Building the team that needs you | LinkedIn Jobs count, `site:greenhouse.io/[company]` | |
|
121
|
| Tech stack change | Migration = pain = budget | BuiltWith historical, job posts mentioning "migrating from X" | |
|
122
|
| Competitor displacement | Negative G2 review of competitor | `site:g2.com "[competitor]" 1-star OR 2-star` | |
|
123
|
| M&A / new product launch | Org chaos creates tool gaps | Press releases, TechCrunch | |
|
124
|
| Earnings call mentions | Public co priority signals | `webFetch` SeekingAlpha transcripts, Ctrl-F for your problem space | |
|
125
|
|
|
126
|
### Step 4: Qualify & Tier |
|
127
|
|
|
128
|
**Fast disqualification (do this first):** Before researching, kill accounts that fail any hard constraint — wrong geo, below headcount floor, competitor customer under contract, recent layoffs (no budget). |
|
129
|
|
|
130
|
**Qualification frameworks:** |
|
131
|
|
|
132
|
- **BANT** (Budget / Authority / Need / Timeline) — fine for transactional/SMB |
|
133
|
- **MEDDPICC** (Metrics / Economic buyer / Decision criteria / Decision process / Paper process / Identify pain / Champion / Competition) — use for enterprise deals >$50k. The extra P's (paper process, competition) matter because enterprise deals die in legal/procurement, not in the pitch. |
|
134
|
|
|
135
|
**Tiering:** |
|
136
|
|
|
137
|
- **Tier 1** — ICP match + active trigger in last 30 days → full personalization, multi-channel, SDR owns it |
|
138
|
- **Tier 2** — ICP match, no trigger → lighter-touch automated sequence, monitor for triggers |
|
139
|
- **Tier 3** — Partial fit → newsletter/nurture, revisit quarterly |
|
140
|
|
|
141
|
### Step 5: Scale with Parallel Agents |
|
142
|
|
|
143
|
**Target minimum 40 prospects.** A single sequential search won't get there fast enough. Use `startAsyncSubagent` to run **5 parallel research agents**, each focused on a different search angle: |
|
144
|
|
|
145
|
1. **Industry/vertical search** — companies in the target vertical via Crunchbase, G2 category pages |
|
146
|
2. **Funding/growth search** — recently funded companies matching the ICP |
|
147
|
3. **Hiring signal search** — companies hiring for roles that indicate they need the user's product |
|
148
|
4. **Competitor customer search** — companies using competitors or reviewing them on G2/Capterra |
|
149
|
5. **Lookalike search** — competitors and alternatives to any strong-fit companies already found |
|
150
|
|
|
151
|
Each agent should return 10-15 prospects with all columns filled in. Use the `wait_for_background_tasks` tool to wait for them, then deduplicate and merge into the final spreadsheet. |
|
152
|
|
|
153
|
### Step 6: Output as a Spreadsheet |
|
154
|
|
|
155
|
**Build a real spreadsheet** using the excel-generator skill or write a CSV file — not a markdown table. The output should be something the user can import directly into a CRM, Clay, or Apollo. |
|
156
|
|
|
157
|
**Columns:** |
|
158
|
|
|
159
|
| Company | Domain | Headcount | Fit Score (1-5) | Trigger | Trigger Date | Target Contact Name | Target Title | LinkedIn URL | Email (if found) | Why Now (1 sentence) | |
|
160
|
|
|
161
|
- **LinkedIn URL** — search for the target persona at the company: `site:linkedin.com/in "[company]" "[title]"`. Include direct profile links. |
|
162
|
- **Email** — look for email patterns via `webSearch("[company] email format" OR "[name] [company] email")`. Common patterns: `first@company.com`, `first.last@company.com`. If not found, leave blank and note the likely format. |
|
163
|
- **Why Now** — the most valuable column. It's the first line of the cold email. |
|
164
|
|
|
165
|
The spreadsheet should be downloadable and ready to import — not just displayed as text. |
|
166
|
|
|
167
|
### Deep Research for Complex ICPs |
|
168
|
|
|
169
|
For industries or markets you don't know well, pull in the **deep-research** skill to build context before prospecting. This is especially useful when: |
|
170
|
|
|
171
|
- The user sells into a niche vertical you don't have strong priors on (e.g., "construction tech", "veterinary SaaS") |
|
172
|
- You need to understand market landscape, key players, and buyer behavior before defining the ICP |
|
173
|
- The user wants competitor analysis as part of the prospecting process |
|
174
|
|
|
175
|
Use deep-research to gather industry context, then return here to build the prospect list with that knowledge. |
|
176
|
|
|
177
|
## Agent Tactics |
|
178
|
|
|
179
|
**Tech stack detection:** |
|
180
|
Use `webSearch("[company] tech stack" OR "[company] built with")` to find BuiltWith/Wappalyzer/StackShare profiles. Common signatures to look for in search results: |
|
181
|
|
|
182
|
- Segment, Intercom, Stripe, Shopify, Google Analytics |
|
183
|
- Job postings often reveal stack: `webSearch("[company] careers engineering")` |
|
184
|
|
|
185
|
Note: `webFetch` returns markdown content, not raw HTML — script tags and asset URLs are stripped. Use search-based detection rather than HTML source scanning. |
|
186
|
|
|
187
|
**Waterfall search pattern:** If one query returns nothing, don't stop — try synonym variants. "VP Engineering" OR "Head of Engineering" OR "Engineering Lead" OR "CTO" all map to the same persona at different company sizes. |
|
188
|
|
|
189
|
**Lookalike expansion:** Once you find 5 good-fit accounts, search for their direct competitors — `webSearch: "[good-fit company] vs"` or `"[good-fit company] alternatives"` surfaces the category. |
|
190
|
|
|
191
|
## Best Practices |
|
192
|
|
|
193
|
1. **50 researched > 500 sprayed** — reply rates on researched lists run 3-5x higher |
|
194
|
2. **Disqualify before you qualify** — negative filters are cheaper to check |
|
195
|
3. **Trigger freshness decays fast** — a funding round is a 60-day window, a job change is a 90-day window |
|
196
|
4. **Enrich once, cache the result** — don't re-research the same account every sequence |
|
197
|
5. **B2B data decays ~30%/year** — any list older than 6 months needs re-verification |
|
198
|
|
|
199
|
## Limitations |
|
200
|
|
|
201
|
- Cannot log into LinkedIn Sales Navigator, Apollo, ZoomInfo, or Clay — builds search strategies the user executes |
|
202
|
- Cannot verify email deliverability (user should run through NeverBounce/ZeroBounce before sending) |
|
203
|
- Cannot detect intent data (Bombora/6sense-style content consumption signals require paid platforms) |
|
204
|
- Company headcount/revenue estimates from public web are approximate — private company data is inherently fuzzy |