I'm an enterprise AE at Freshworks. I've won President's Club seven times and closed a Seagate deal that was the largest TCV in company history. But when I decided I wanted to move into AI sales — specifically targeting Anthropic and OpenAI — I had the same problem every sales rep has: I didn't know the right people.
So I built a system.
Instead of scrolling LinkedIn manually and sending generic connection requests, I used Claude Code and Apify to scrape the right GTM contacts at both companies, discover their email formats, draft personalised cold emails using my actual credentials, and track everything in Obsidian automatically. The whole pipeline runs from a single slash command.
The meta point: Using an AI tool to apply to AI companies isn't just clever — it's the right signal. Anthropic and OpenAI want AEs who understand the product from the inside. Showing up with a scraper and a pipeline is the application.
The Problem With Normal Job Applications
Cold applications to Anthropic and OpenAI go into an ATS black hole. Both companies receive thousands of applications for every GTM role. Data from Glassdoor and Blind shows Anthropic responds to sales applicants within 5–14 days if they're interested — meaning if you're past 4 weeks with no response, the cold application is dead.
The move is direct outreach to recruiters and sales leaders before or alongside the application. But that requires knowing who to contact, their emails, and what to say. That's the problem I automated.
The System I Built
I created a Claude Code skill called /job-prospect that runs the full pipeline end-to-end:
LinkedIn Scrape via Apify
Runs 3 search queries against harvestapi/linkedin-profile-search: recruiters, enterprise AEs, and sales leaders. Filters to TIER 1 (recruiters) and TIER 2 (sales leaders). No manual searching.
Email Format Discovery
Web searches RocketReach and Hunter.io to find the confirmed email pattern. Anthropic is firstname@anthropic.com (61% confidence). OpenAI is firstname@openai.com (41% confidence).
Personalised Email Drafts
Drafts one email per TIER 1 recruiter and one per TIER 2 sales leader. Each email uses my real credentials: 7× Club, Seagate Fortune 500 close, $2.3M ACV at 200% of quota. No generic templates.
Obsidian Note + Tracker
Saves everything to a structured Obsidian note with frontmatter properties (email_sent, reply_received, call_booked) that feed into a Base tracker. Every future session can see exactly where each company is in the pipeline.
Memory Across Sessions
A memory file in Claude Code's project memory stores all contacts, sent status, and next steps. In the next session, I just say "I sent an email to Debora Floriano at Anthropic" and the system updates everything automatically.
What I Found at Anthropic
The scrape returned 9 contacts across TIER 1 and TIER 2. The key insight: Anthropic's recruiting team is split by geography and function. GTM recruiting (Nick B.) handles sales roles separately from Applied AI recruiting (Anne Marie Riffle, New York) and SF Bay Area recruiting (Lynda Galliano). Emailing the right person matters.
For US roles specifically: Debora Floriano is EMEA recruiting — skip her if you're applying stateside. That's the kind of thing you only know if you look at their LinkedIn profiles, not just the company recruiter list.
What I Found at OpenAI
OpenAI calls the role "Account Director" not "Account Executive." This matters more than it sounds — job postings, ATS keywords, and the recruiters who own the role all use AD language. Applying for "AE" roles when they're listed as "AD" means your application might not surface in their search.
OpenAI's enterprise motion is more mature than Anthropic's. They have vertical-specific ADs (Healthcare, Financial Services, Retail, Digital Natives), territory design, and a more structured SDR/BDR function. The comp is higher base today; Anthropic equity is likely the bigger long-term bet.
The Email That Actually Works
After testing multiple approaches, the email that gets responses leads with one specific number, not a list of credentials:
"I closed Seagate as a net new Fortune 500 logo — largest TCV deal in our company's history. I'd love 15 minutes to learn more about how you're building the enterprise team."
Recruiters at AI companies see hundreds of "passionate about AI" emails. One specific, verifiable number cuts through.
How to Build It Yourself
The whole system lives in two Claude Code skills. Here's exactly how it works:
Step 1: Open a terminal and type claude to start Claude Code. Then run:
/job-prospect [Company] | [Role]
Examples: /job-prospect Anthropic | Enterprise Account Executive or /job-prospect OpenAI | Account Director. The skill runs the full pipeline: LinkedIn scrape → email format → ranked contacts → personalised emails → Obsidian note. Takes 3–4 minutes.
The skill searches LinkedIn for people by keyword (not company URL — that returns zero results on the Apify actor I use). It queries three separate searches: "[Company] GTM recruiter", "[Company] enterprise sales", and "[Company] revenue". Then it dedups, ranks, and removes irrelevant roles.
One thing that took me two tries to figure out: Anthropic's recruiting team is split. Nick B. handles GTM roles (sales, marketing, partnerships). Anne Marie Riffle handles Applied AI roles (solutions engineers, technical AEs). Debora Floriano covers EMEA. If you're applying for a US-based enterprise AE role, Nick is your person — not the generic "recruiting" contact.
The Obsidian Tracker
Every company gets its own Obsidian note with YAML frontmatter that acts as a database:
email_sent / reply_received / call_booked / contacts_emailed
These feed directly into an Obsidian Base view — a table that shows every company, contact, and status in one place. When I open the terminal the next day and say "I sent an email to Debora Floriano at Anthropic", Claude reads the note, updates the fields, and opens the tracker. No manual data entry.
The second skill — /job-update — handles the ongoing pipeline. You just describe what happened in plain English. Claude figures out which note to update, which fields to change, and what the next step should be. It's the kind of CRM update I'd normally spend 10 minutes on.
The Numbers That Got Responses
I sent 8 emails across Anthropic and OpenAI. The emails that performed best were built around one specific, verifiable number rather than a list of achievements. The opening that worked:
"I closed Seagate as a net new Fortune 500 logo — largest TCV deal in our company's history. Year two, I hit $2.3M ACV at 200% of quota — first rep to do it. I built a prospecting pipeline in Claude Code to find the right people at Anthropic before sending this. Happy to walk you through how it works on a 15-minute call."
The last line — about building the pipeline in Claude Code — generated the most curiosity. Recruiters at AI companies get hundreds of "passionate about AI" emails. Showing up with something you actually built is a different signal.
Want to Run It for Your Target Company?
The system works for any company. You need:
- Claude Code installed (
npm install -g @anthropic-ai/claude-code) - An Apify account (free tier covers ~50 scrapes/month)
- Your credentials written out once in a memory file — Claude reuses them in every email automatically
The full skill files (/job-prospect and /job-update) live in my ~/.claude/commands/ folder. I'll publish them properly when I've run the full pipeline through a few more companies — Cohere, Mistral, and Salesforce AI are next.
What's Next
The pipeline is live. Emails are sent to Nick B. and Ryan Ball at Anthropic. The tracker is running. I'll write up what happens — whether I hear back, what the interview process looks like, and what I'd change if I were starting over.
If you're an enterprise AE trying to make the same move and want to talk shop — I'm on LinkedIn. Or subscribe below and I'll send you the update when it comes.
