Priscilla Philavong
- ○ Why this profile excites you
- ○ Any yellow or red flags you'd want to validate
- ○ What you'd prioritize confirming in an early conversation
- ○ What archetypes of profiles would you go after if you could choose 1–3 types in buckets
- ○ In your view, what are the key ingredients to building and sustaining a high-density technical organization at Series A?
- a. If you could snap your fingers and build 3 automations to supercharge your team's workflow, what would they be?
My Responses
This section contains my detailed responses and is password protected. Please enter the access code provided.
Priscilla Philavong
1. Three candidate profiles I'd pursue
Profile archetype
Former engineer at Stripe, Zapier, Retool, or Cloudflare who has built customer-facing integrations against enterprise APIs and knows how to translate customer pain into product decisions. Proficient in Python and TypeScript. Has done discovery sessions, demos, and iterated on product based on real usage. Comfortable in a room with a CISO and an engineering lead in the same conversation.
Why this profile excites me
Runlayer's Product Engineer sits at the intersection of building and customer work. This person has already lived that loop — they know the most important thing isn't the elegance of the code, it's whether the customer actually uses what you shipped. They won't need to be taught customer obsession; it's already how they work.
Yellow / red flags
- Only internal tooling experience — hasn't shipped customer-facing product
- Strong technical depth but avoids customer conversations — wrong orientation for this role
- No LLM or agent framework experience — given Runlayer's product, a real gap
Confirm early
- Walk me through the last integration you shipped end-to-end. Who was the customer?
- How do you decide what to build vs. what to say no to after a customer conversation?
- What AI or agent frameworks have you shipped production code with?
Profile archetype
Engineer from OpenAI solutions, Anthropic tooling, Langchain, or Modal who has helped enterprise customers implement AI workflows. Deeply familiar with LLM APIs, agent frameworks, and the real-world problems that come up when AI connects to production systems. Bonus: has built something with MCP, FastMCP, or Vercel AI SDK directly.
Why this profile excites me
They already understand the problem Runlayer is solving from the customer's perspective. They've been in the room when an enterprise security team pushed back on an AI integration. They know what "secure by default" means to a CISO and what "zero additional friction" means to an engineer. That dual fluency is hard to teach.
Yellow / red flags
- Strong on AI theory but hasn't shipped production integrations against real enterprise systems
- Only internal tooling — limited customer-facing experience
Confirm early
- What's the most complex enterprise integration you've shipped? What broke in production?
- Have you worked with MCP or any agent framework? What did you find limiting?
Profile archetype
Recent CS grad or early-career engineer who has been building in public — contributed to MCP-adjacent repos, built their own MCP server, active in Claude/Cursor/Windsurf communities. Has shipped something independently that real people use. Not a passive AI user — a builder on top of the stack Runlayer secures.
Why this profile excites me
Zero bad habits, genuine passion for the protocol layer, and will grow with the company. David Soria Parra on the cap table means something to this person. They onboard fast because they already understand the product's why.
Yellow / red flags
- All frontend, no systems or API depth — Runlayer needs integration builders
- Open source contributions are docs PRs only, no shipped code
- Interested in AI as a user, not as a builder
Confirm early
- Show me something you built. GitHub, side project, anything in production.
- Have you worked with MCP? What problem were you trying to solve?
Candidate archetypes — 3 sourcing buckets
1. Integration engineers from developer tools companies — Zapier, Retool, Stripe, Cloudflare. Know how to build against enterprise APIs at scale, customer-facing already, used to translating customer pain into product decisions. Maps directly to the JD's emphasis on API/integration experience and enterprise systems like Salesforce, ServiceNow, Jira.
2. AI solutions engineers from LLM companies — OpenAI solutions, Anthropic tooling, Langchain, Modal, Replicate. Deep LLM and agent framework experience, understand the MCP ecosystem from the inside. The JD explicitly calls for MCP or agent framework experience — this archetype often has it natively.
3. Founding engineers from adjacent AI infra startups — shipped 0-to-1, high ownership, work directly with customers without a PM buffer. Cursor team, GitHub Copilot team, or any AI dev tools startup where engineers own the customer relationship. These people understand that their users are other engineers — exactly Runlayer's buyer.
2. Startups with high technical talent density
~40 people, hundreds of millions in ARR. Ex-OpenAI engineers who shipped a product developers are obsessed with. High bar, fast shipping, engineers own outcomes fully. Directly relevant: their engineers use and understand MCP clients daily — Cursor is one of 300+ clients Runlayer supports.
Small team, extreme craft standards. Every engineer is a generalist who ships end-to-end. Known for one of the most deliberate hiring processes in tech. Their product reflects their team's quality directly.
$0 to $100M ARR in 18 months. Team built around Unit 8200 alumni — engineers who have worked in high-stakes real-world security environments. Directly relevant talent pool: security engineers who ship fast and understand enterprise buyers.
Consistently strong engineering culture in data and AI infrastructure. Their ML/platform engineers are among the most sought-after in the market. Jeff Settle came from here — that network is already a Runlayer asset.
This narrative is uniquely Runlayer's and no competitor can steal it. Andrew and Tal built the first MCP server at Zapier. David Soria Parra — the creator of MCP — is on the cap table. Travis McPeak, head of security at Cursor, is an advisor. Every employer brand touchpoint should reinforce this provenance: the people who shaped how AI connects to the world are now building the security standard for it.
Activation: Tal and Andrew publish first-person pieces about what they saw at Zapier — the blind spots, the attack vectors, what kept them up at night. Engineers want to work with people who have been inside the problem. This story attracts them before a recruiter ever reaches out.
The people identity (Runners), the values (Say it, Ship it, Scale it), and the operating principle of intellectual humility all point to the same place. We hire exceptional people and then get out of their way. 9 products shipped in 8 months. What you build here runs at Gusto, Instacart, and Ramp.
Activation: Engineering blog posts from Runners about what they shipped, how fast, and what they learned. Technical write-ups of real problems solved. The brand isn't what Runlayer says about itself — it's what Runners say about their work.
MCP is the connective tissue of the agentic web. Runlayer is the security and governance layer for that connective tissue. The narrative for engineers who think in decades: the infrastructure you help build here will be foundational to how AI operates in enterprise for the next 10 years. This isn't a feature — it's a new category being defined in real time.
Activation: Conference presence. Vitor already speaks internationally on MCP security. Get the team on stage at AI infrastructure and security conferences. The engineers who attend those conferences are exactly who Runlayer needs to hire.
The EVP framework — Runners SHIP
Give / Get — the EVP pillars
Pillar 1 · Protocol Builders
Give: Deep technical ownership at protocol layer. You're not contributing to a feature — you're defining a category that didn't exist 18 months ago.
Get: Direct access to the team that built MCP. David Soria Parra and Travis McPeak on the cap table. You work with the people who wrote the spec.
Pillar 2 · Ship It Culture
Give: End-to-end ownership. No PM buffer, no committee, no permission needed. A Runner defines the problem and delivers the solution.
Get: Your work runs at Gusto, Instacart, Ramp. 5M+ secured MCP calls. Real stakes, real customers who depend on what you built.
Pillar 3 · AI-Native Environment
Give: You work inside the stack you're securing. AI runs 20+ hours a day at the company. Every new employee gets an AI agent on day one.
Get: A workplace where AI fluency is the default. You grow faster here because the environment compounds your output.
Top 3 ways I'm currently using AI as a talent leader
Takes raw Brighthire transcripts from recruiter phone screens and converts them into structured hiring manager review documents — automatically. The output includes a takeaway, summary, strengths and weaknesses, and a role alignment section that cross-references the career ladder, job description, salary, salary budget, and market benchmark compensation for the candidate's geography, company size, and industry.
Once generated, the summary is automatically submitted on behalf of the interviewer in Ashby and triggers an Ashby notification to the hiring manager. No recruiter admin, no formatting, no delay — the HM sees it the moment the screen ends.
Why it matters: Most recruiter screen notes are information-dense but inconsistently formatted, creating friction for hiring managers. This closes the loop automatically and surfaces compensation alignment before the conversation moves forward — eliminating one of the most common sources of late-stage offer failure.
Takes a candidate profile — GitHub, conference talks, open-source contributions, LinkedIn — and generates a personalized outreach message that references their specific work. Adapts by persona and seniority: the message to a Staff security engineer reads differently than the one to an early-career MCP contributor.
Result: 46.6% InMail response rate over the last 6 months — nearly 2x the LinkedIn benchmark of ~25%. The lift comes from specificity, not volume.
After interview panels close, synthesizes all written feedback to: identify conflicting interviewer signals, surface weak or unsupported opinions, flag scoring inconsistencies across the same competency, and produce a hire/no-hire recommendation with reasoning. Separates "signal" from "confidence" — the distinction most calibration meetings miss.
Result: Debrief conversations are more honest and faster. Panelists come in having seen where they diverge, not ratifying a consensus that formed before the meeting started. Directly improves hiring bar consistency across functions.
My full AI recruiting stack — 5 agents in production
Brighthire transcript → structured HM review document → auto-submitted in Ashby → HM notification triggered. Full loop, zero admin. Includes comp alignment against market benchmarks, career ladder, and job description.
Personalized outreach by persona and seniority. Drives 46.6% InMail response rate — 2x LinkedIn benchmark. Built on Claude, adapted per role and candidate archetype.
Synthesizes panel feedback, flags scoring inconsistencies, surfaces unsupported opinions, produces hire/no-hire recommendation. Most strategically important agent — directly affects hiring quality and consistency.
EVP and recruiting messaging copilot. Writes outbound messages, LinkedIn content, and candidate-facing narratives in Abnormal's voice. Includes persona modes (Staff SWE, AI Infra, Security Engineering) and objection handling for common candidate pushback.
Generates offer packets, compensation framing, level alignment summaries, and closing narratives. Standardizes approvals and formats rationale clearly — reduces offer-stage admin and directly improves close rates and candidate experience.
3 automations I'd build for Runlayer
Weekly digest flagging: candidates who've gone cold, candidates at risk of losing to a competitor, and candidates ready for an offer push. CRM intelligence for the candidate pipeline — the account management a great sales Runner does for their book, applied to recruiting.
When a candidate mentions a competing offer, surfaces public comp data for that company/role/level, the candidate's stated motivations, and draft talking points for the close. Turns every competing offer into a prepared negotiation — especially important at a seed-stage company competing on mission and equity over cash.
The same Brighthire → Ashby loop I've built at Abnormal, adapted to Runlayer's interview process and ATS. Auto-submits structured recruiter screen summaries to hiring managers with role alignment against Runlayer's compensation benchmarks and career ladder. The goal: zero time between screen and HM review, zero formatting overhead for the recruiter.