Outricher · internal + client playbook

The Venture Signals Playbook

How a graph of 1.17 billion professional profiles turns into answers a fund, a recruiter, or a sales team can act on. Six real workflows — what the client is actually trying to solve, the data and algorithm behind each, and how much business value the numbers carry.

Audience · internal team + prospective client Format · teaching demo Data · queried live 2026-06-04 Charts · live, interactive

This document does two jobs at once. For our team, it's a reference for how we go from a raw question to a delivered signal — which tables, which algorithm, which join. For a client, it's a tour of what becomes possible when you stop buying static lists and start querying the professional economy as a graph. The cases are illustrative; every statistic and chart is real, pulled from the production dataset the day this was written.

1.17B

Professional profiles with full career history

5.6B

Position records — the raw material for every timing signal

237M

Profiles pre-enriched with email / phone / socials

204M

Developer profiles with public open-source activity

On this page

The mental model: a graph, not a list
Investor discovery — ranked lead lists
Investor enrichment — the personalisation layer
Founder intelligence — repo to warm email
Momentum signals — funding, role change, hiring
Portfolio monitoring — the early-warning feed
LP discovery — capital allocators by region
The economics & the takeaway

The mental model: a graph, not a list

Most lead vendors sell you a spreadsheet — a frozen snapshot that's stale the day it lands. We operate something different: a continuously-refreshing graph of the global professional economy, where people, positions, companies, funding rounds, and developer activity are all linked. That structural difference is what makes signals possible. A list can tell you who someone is. A graph can tell you what just changed about them — which is the only thing that makes an outreach timely.

Five layers do the heavy lifting across every workflow below:

Layer	Scale	Primary tables	What it unlocks
People	1.17B	`linkedin_profile`	Identity, role, location, seniority, reach
Career history	5.6B	`linkedin_profile_position3`	Tenure, role change, hiring & churn signals
Companies & funding	117M	`linkedin_company` · `linkedin_crunchbase_funding`	Funds, portfolio cos, stage, round amounts
Contact enrichment	237M	`person_enrichment`	Personal email, phone, social handles
Developer activity	204M	`github_profile` · `github_repository`	Earliest signal a technical founder exists

Internal note: person_enrichment joins to profiles on the public-profile slug; always SUBSTRING(public_profile_url FROM 'linkedin\.com/in/(.+)$'), never the full URL. Sources are scrubbed to "Outricher Home API" in anything a client sees.

The single most important economic fact: search, sizing, ranking, and most contact data are free — they come straight from the graph. We only charge to live-enrich a specific person the pre-enriched layer doesn't already cover. A weekly pipeline of thousands of targets costs effectively nothing; you spend only on the handful you decide to reach.

WORKFLOW 01

Investor discovery — ranked AI lead lists

The problem the client has "Give me a ranked list of investors who match my thesis — by stage, geography, and focus — and keep it fresh every week without an analyst doing manual research."

What data we have

The venture universe is fully structured. We isolate it on industry and then rank on headcount, geography, founding year, reach, and the firm's own thesis text. There are 17,227 venture & PE firms in the graph; 6,660 have an active team of five or more.

Distribution of venture & PE firms by team size — the long tail of boutique funds (1–10 people) is where most thesis-specific, under-marketed-to partners sit.

The algorithm

Discovery is a two-pass funnel: a wide structured query to define the universe, then a relevance score on the thesis text. This is the same machine that produced an 18,000-firm scored pipeline and 38,000 decision-makers in a prior gaming/web3 investor build — re-pointed at "AI infra, seed" by swapping the keyword set.

-- Pass 1: define the universe (free, indexed)
SELECT id, company_name, country_iso, employee_count, description
FROM linkedin_company
WHERE industry IN ('Venture Capital', 'Investment Banking/Venture')
  AND employee_count BETWEEN 5 AND 500;

-- Pass 2: score the thesis text + map the individual investors
--   description ILIKE '%artificial intelligence%' → thesis match
--   join positions where title ~ Partner|Principal|MD → named people

What it returns — a thesis-matched shortlist (live)

Firm	Geo	Investors mapped	Thesis signal (firm's own words)
Sapphire Ventures	US	14	Enterprise AI, $10B AUM — Glean, LangChain, Clay
Greycroft	US	26	Seed–growth, AI infrastructure, $3B raised
Bain Capital Ventures	US	25	B2B software seed→growth, $10B AUM
BECO Capital	UAE	6	Gulf-focused, $820M AUM, AI & enterprise
JVP	Israel	8	AI & cyber centre of excellence, $1.75B raised

Venture firms by country (active team ≥ 3). A thesis filter narrows this; a geography filter slices it — the agent can ask for "US + UK + UAE seed AI funds" and get a clean, named list.

The business value A static list of "VC firms" is a commodity. The value is resolution: from 17,227 firms down to the ~30 partners whose thesis, stage, and geography actually match — each as a named human, not an info@ address — refreshed weekly so the agent's pipeline never goes dry.

Illustrative case A seed-stage AI fund wants warm intros to co-investors for its next infra deal. We deliver 240 ranked partners across US/UK/UAE whose firms explicitly name "AI infrastructure" in their thesis and who've written seed checks in the last 18 months. The agent opens each with the firm's actual portfolio — not "I see you invest in tech." Reply rates on thesis-matched, named outreach run multiples above list-blast.

WORKFLOW 02

Investor enrichment — the personalisation layer

The problem the client has "I have a name or a profile URL of an investor I want. Give me their personal email, Twitter, phone, what they post about, and any recent role change — so my agent writes a human first line, not a template."

What data we have + the algorithm

One call to outricher.com/api/v2/profile with a profile URL returns a 47-field dossier. Below is a real, unedited enrichment of a public investor — Garry Tan, President & CEO of Y Combinator — pulled live while writing this. Contacts are partially masked here out of courtesy; delivered to a client in full.

// GET outricher.com/api/v2/profile?profile=.../in/garrytan
{
  "full_name": "Garry Tan",
  "headline":  "President & CEO, Y Combinator",
  "city":      "San Francisco, United States",
  "follower_count": 318236,
  "personal_emails": ["me@g•••••tan.com", "garry@yc•••••••.com", "+ 9 more"],
  "personal_numbers": ["+1 415 ••• ••••", "+ 5 more"],
  "twitter": "twitter.com/garrytan",  "website": "blog.garrytan.com",
  "experiences": [
    { "title":"President & CEO", "company":"Y Combinator", "since":2023 },
    { "title":"Founder & Managing Partner", "company":"Initialized Capital", "2012":2022 }
  ]
}

That single record carries everything needed for a human opener: the inbox he reads, the channels he actually uses (Twitter, a personal blog), and the career arc that says he moved from running his own fund to running YC in 2023 — a fact worth leading with. A second live pull (Elad Gil, prolific AI angel) returned 11 personal emails, his X handle, 62,711 followers, and a career path from Mixer Labs → Twitter VP → Color → full-time investing.

The free-tier dividend

Of 574 investor profiles sampled across the funds above, 35% already carried a verified personal email in the pre-enriched layer at zero cost. You pay live-enrichment only on the rest — and only for the people the agent actually picks.

The business value Personalisation is the difference between a 2% and a 20% reply rate, and it's exactly what an agent can't invent. We hand it the raw facts — the personal channel and the career arc — and the cost is a rounding error: a third of senior investors are free, the rest cost roughly two cents each.

Illustrative case The agent drafts to a partner who just moved firms. Instead of "I'd love to connect," it opens: "Congrats on the move to [new fund] — saw you're now leading their infra thesis." That line is built from two fields we returned: current company start-date and title. The investor replies because it reads like a person who did their homework, not a sequence.

WORKFLOW 03

Founder intelligence — from a code repo to a warm email

The problem the client has "My deal scan finds a promising company through its open-source project. Enrich the founder: personal email, Twitter, dev activity, education, prior startups, time-in-role — so the first email has real context, not a template."

What data we have

204M developer profiles with public open-source signal, cross-linked to the professional graph. When a deal scan surfaces a repository, we resolve the maintainer to a full human: github_profile gives login, followers, public-repo count, bio, location, and social handles; the professional graph adds education, work history, and tenure.

What it returns — developer-level signal (live)

Builder	Followers	Public repos	Reachable channel
Anthony Fu	37,291	384	@antfu7 · open-source maintainer
Guillermo Rauch (CEO)	16,114	138	personal email · x.com/rauchg
Andrew Clark	14,745	74	@acdlite · React core team
Tobias Koppers	9,332	127	@sokra · author of webpack

A sample of hireable developers with real output (5+ public repos), bucketed by following (log scale). The pyramid is steep — the ~2,000 "Senior / Tier-1" builders are where founders and founding engineers cluster, and a deal scan can rank a repo's contributors into these bands in seconds.

The complete founder dossier the agent receives: personal emailTwitter / X open-source outputeducation prior startupstime in current role. This is the exact recipe we ran for a "developers actively shipping open source right now" extraction: repository activity → identity → segmented, reachable contact.

The business value For a fund doing technical deal-sourcing, this collapses the gap between "saw an interesting repo" and "sent a credible email to the founder" from days of manual digging to a single enrichment call — and it lets the agent reference the founder's actual code and prior companies, which is the only thing that earns a reply from a builder who ignores generic VC outreach.

Illustrative case A deal scan flags a fast-growing inference-optimisation library. We resolve its top three contributors: one is a returning founder (sold a dev-tools company in 2021), reachable on a personal email, posting weekly on X about the exact problem. The agent's email cites their prior exit and a specific commit. That's a meeting — not a cold open.

WORKFLOW 04

Momentum signals — funding, role change, hiring velocity

The problem the client has "Tell me when to reach out. A startup that just raised is a follow-on moment. An investor who just changed firms is open in their first 90 days. A company suddenly hiring fast is a product that's working." Timing beats targeting.

Signal A — funding (live)

The funding layer carries round name, date, amount, and the named backer list, fresh through late November 2025. The shape of the market falls straight out of it:

Rounds closed in 2025 by stage, with average check size. Bars = number of rounds · line = avg amount ($M) A "just raised, time for a follow-on conversation" query is a date + stage filter on this layer.

Signal B — role change (live)

An investor's current position with a recent start date is the "new fund, first 90 days" cohort — found by scanning position3 for fresh start dates at target firms.

Investor	New role	Firm	Started
Rakan AlRashed	Partner	BECO Capital (UAE)	Dec 2025
Alexis Frentz	Partner	Elaia (FR)	Dec 2025
Jason Brooke	Principal	Sapphire Ventures (US)	Dec 2025
Amanda Huang	Principal	Bain Capital Ventures (US)	Nov 2025

Signal C — hiring velocity (live)

Net headcount momentum, quarter by quarter, is the clearest "this is working" signal a company emits before its metrics are public. Here is the real hiring curve of two frontier AI labs:

Quarterly hires, 2023 Q1 → 2026 Q1. Both curves bend sharply upward through 2025 — Anthropic from ~120/quarter to over 1,000. A fund watching this would have seen the acceleration two to three quarters before it was obvious.

The business value Targeting tells you who; momentum tells you when. Reaching a founder the week after a raise, or a company the quarter it starts ramping sales, is the difference between "good timing" and "too late" — and these signals are computed from data the company can't hide and hasn't announced yet.

Illustrative case A growth fund sets a standing filter: "Series A company, raised < 90 days ago, hiring velocity +30% QoQ." Every Monday the agent gets a fresh short-list of companies hitting all three at once. It reaches the CEO with "saw the round and the eng-team ramp — looks like you're scaling GTM next" the week it becomes true. The fund is in the conversation before the banker calls.

WORKFLOW 05

Portfolio monitoring — the early-warning feed

The problem the client has "I want to know what's happening inside my portfolio companies — hiring, team churn, key-person departures — earlier than the quarterly board update tells me."

The algorithm

Same velocity engine as Workflow 4, pointed at your own holdings and run on a schedule. The is_current + start_date / end_date fields on 5.6B positions let us compute, per company per week, three early indicators. The hire-vs-exit ratio is the headline number:

Trailing-12-month hires vs departures for three companies (live). A healthy growth company runs a high hire:exit ratio (OpenAI ≈ 3.8:1, Anthropic ≈ 6:1). When that ratio compresses toward 1:1 — as in a maturing Snowflake — it's an early structural signal worth a call.

Headcount delta — net hires vs departures. A reversal shows months before a board deck.
Key-person churn — when a VP or founding engineer's position flips to "past", you know the week, with the name and where they went.
Function-level ramp — ten new sales roles opening says the product found a market. We detected this "first-sales-hire" signal across 780 funded startups in a prior build.

The business value A board update is a lagging indicator written by the people being measured. A live headcount-and-churn feed is a leading, independent one. For a fund managing 30 companies, catching one founding-team departure or one stalled-growth signal a quarter early can change the outcome of a reserve decision.

Illustrative case A portfolio company looks fine on paper. The feed flags that two of its four founding engineers flipped to "past" in the same month, and net hiring went negative for the first time. The partner raises it at the next board meeting — six weeks before it would have surfaced in the quarterly. That's the entire point of the feed.

WORKFLOW 06

LP discovery — capital allocators by region

The problem the client has "I'm raising. Find me family offices, HNWIs, and corporates in MENA and Europe — the same lead-list machine, pointed at allocators instead of investees."

What data we have + the algorithm

Identical engine to Workflow 1, filter swapped to wealth and family-office entities. The capital-allocator universe is large and well-covered in exactly the regions funds raise from.

Wealth-management & family-office entities (2–500 staff) by country. MENA and European fund-domicile hubs Each resolves to named entities, then to individual principals with the same enrichment layer behind them as your investor targets.

These are real, named entities — in the UAE alone: Skybound Wealth Management (271 staff), Titan Wealth International (160), and the Global Family Office Investment Summit network — each expanding to reachable principals.

The business value LP discovery is usually the most manual, relationship-gated part of running a fund. Turning it into a queryable, enrichable list — the same infrastructure already used for deal sourcing — means a raise can run on the same agent and the same data spine as the investing motion.

Illustrative case A fund opening a MENA allocation gets a ranked list of 330 UAE wealth and family-office entities, narrowed to those with a tech-allocation signal in their profile, each resolved to a named principal with a personal channel. The raise outreach runs on the same pipeline as the deal outreach — one engine, two ICPs.

The economics & the takeaway

The pricing model mirrors how a fund actually works: broad discovery is free, precision contact is cheap, and you pay only for the people you decide to pursue.

Operation	Covers	Cost
Discovery · sizing · ranking · segmentation	Every workflow's list-building & all signals	Free
Pre-enriched contact (≈20% of graph; 35% of senior investors)	Personal email already on file	Free
Live profile enrichment	Full 47-field dossier, on demand	~$0.02

A 5,000-target weekly investor pipeline with momentum signals costs effectively nothing to run. Enriching the 200 the agent actually emails costs the price of a coffee. That is the unit economics of a sourcing engine that never runs out of warm, well-timed, personally-addressable targets.

The one-line takeaway, for both audiences: the value isn't the 1.17B profiles — anyone can buy rows. The value is the graph: the joins between people, positions, funding, and code that turn a static list into a timed, ranked, personally-addressable signal. Every workflow above is the same engine with a different filter — which is exactly why one data spine can run deal sourcing, founder outreach, portfolio monitoring, and an LP raise at once.