GTME / Eric Fleshman
Receipts / the scoreboard, without the story

Every number I claim,
in one place.

Some people want the story. The case studies are one click away for them. This page is for everyone else: the numbers, the system each one came from, and how it was counted. If a number here does not have a source line, it does not belong on this page.

No narrative Every figure sourced Producible on request
01

Kizen: first GTM engineering hire, 90 days

A GTM function that did not exist, built from an empty CRM to a running revenue system. Full build: the $50M inbound engine.

$50M+

Pipeline processed through inbound automation I built

How counted: total pipeline value flowing through the routing and enrichment path, across web forms, event lists, and Apollo activity.

72h to 10s

Speed to first touch on an inbound lead

Before: inbound landed in a VP inbox and got hand-typed, roughly 72 hours to first touch. After: agent-driven routing delivered an enriched, assigned contact in about 10 seconds.

$10.91M

Marketing-sourced pipeline the attribution chain surfaced

How counted: pipeline sitting in unattributed records before the Contact to Lead to Account to Opp inheritance chain shipped, then correctly credited after. 4 production automations, one 8-value source taxonomy.

600+

Documented changes shipped in 90 days

How counted: rows in the changelog I kept myself, from day one, with zero onboarding and no ticketing system to inherit.

40+

Production automations, each monitored with failure alerts to Slack

How counted: distinct automations created February to May 2026, per the same changelog. Monitored means a failure pages a human, not that it merely ran.

2,000

Accounts in the insurance TAM I built and validated

Method: a multi-source enrichment waterfall across LinkedIn, government filings, ZoomInfo, and Perplexity. The account tiering framework was adopted by the CEO as the company standard.

400+ / $38M+

Event contacts routed through the same pipe, and the TAM they represented

Source: the Medicarians 2026 conference list, loaded through the production inbound path rather than a side spreadsheet.

4 objects

Core CRM objects designed from scratch, with bidirectional relationships

Scope: the data model everything else in this section sits on. There was no prior model to extend.

4 syncs

Bidirectional Apollo integration: contacts, emails, calls, tasks

Build: direct API to API with no middleware, 10 minute scheduler intervals, activity and stage transitions moving both directions.

6 gates

Production stage gates with dynamic field validation

Enforcement: a custom Slack bot that DMs a rep the exact fields they are missing, plus Timeline API stage reverts. Data quality without manual policing.

2 apps

Custom Slack apps built from scratch for GTM operations

Shipped: GTM Alerts for signals and lead notifications, Pipeline Police for stage gate enforcement. Scoped permissions, bot tokens, incoming webhooks.

48 deals

Lapsed deals closed on the first run of the auto-close scheduler

Build: a daily job that audits open pipeline, requires a loss reason, and DMs reps when close dates need updating.

11 views

Pipeline waterfall reporting surfaces: 7 leadership widgets, 4 per-rep homepages

Underneath: a custom data object tracking every opportunity value change, with automated delta math across all stages.

02

AlphaForge: Clay's GTM engineering program, cohort 2

Four weeks, graded on working systems rather than slide decks. Full arc: the AlphaForge page.

50 of 1,000+

Builders selected into the cohort

Source: Clay's own published cohort 2 numbers, June to July 2026.

12

Builds shipped in 4 weeks

How counted: one graded build per prompt, all twelve indexed on the AlphaForge page with what each one actually did.

Architect Award

The coaching staff's pick at graduation

Selected by: the program's coaches, not a vote and not self-nominated.

631

Heavy-civil contractors in the qualified TAM, after the fit gate

Method: disqualification logic that rejects dealers, design-only firms, and trade media, which look identical to real buyers in a raw list.

25 sent / 24 delivered

Approved outbound sends and confirmed deliveries

Path: Clay list to n8n send engine to Gmail, with a human approval gate before any send.

1

Real buyer reply, caught live and routed back into Clay

Why it counts: an equipment leader at a top ENR heavy-civil contractor. The reply watcher initially dropped it on a case-sensitive filter bug, found and fixed in production. A loop is not closed until you have watched it close.

03

Shipped in public

Things anyone can go run, read, or break right now.

719

Company job boards CandidateZero reads directly, every night

Not a scraper of aggregators: it reads the systems employers post with. 20,684 postings in a single run, 33 clearing every filter. The build.

Live agent

A working agent on the homepage: a company name in, a GTM systems hypothesis out

Design: Perplexity researches the public web, Claude writes the hypothesis under my rules. Key-locked, daily-capped, no person lookup, no visitor record, no stored input. Run it.

7

Public pages on this site, all deployed from a source repo anyone can read

Pipeline: git push to GitHub, Vercel deploys main. The repo is public and privacy-audited.

$0

Monthly run cost of EatTailor, a live app rebuilt by directing AI agents

How: 7 agent sprints across 1 weekend, on free infrastructure tiers. The build.

04

The eight years before GTM engineering

Seed to Series D, on the demand side. Marketing taught me what pipeline is worth before engineering taught me to build it.

$100M+

Annual pipeline target I owned marketing's share of

Where: Convene, Regional Marketing Manager, NYC.

$15M

ARR influenced through account-based marketing

Where: Augury.

$1.5M

Pipeline from launching an AI chatbot, years before it was fashionable

Where: Podium. The first time I shipped an AI system into a revenue motion.

$200K to $5M+

ARR growth as employee #6 and founding marketing hire

Where: strongDM, later acquired for roughly $275M.

On sourcing. Every number above came out of a system I built or ran, and I kept the record: changelogs, dashboards, attribution reports, run histories. I will walk through the underlying report for any figure on this page in an interview.

On what is missing. Numbers I cannot source do not appear here. That is the whole point of the page.

Want the reasoning instead of the scoreboard?

The numbers say what happened. They do not say why I picked those problems. That argument lives on the thesis page.

More on this site: Home · Thesis · AlphaForge · CandidateZero · EatTailor · Agent Anatomy