Evan Lemoine

Aesthetic Management Partners capital medical devices

The problem

Reps were spending two to four hours manually building call lists that were unenriched — searching for practices to target, and then manually dialing each number.

What I built

An end-to-end pipeline that turns a natural-language request into a dialable, enriched call list, and then closes the loop after the call without anyone typing into the CRM.

The design decision that matters is the trigger. Reps will not open a workflow builder, and they will not file a request in a queue. They will type a sentence into an app already on their phone — so that's the interface, and everything else runs behind it.

Why it mattered

• Rep output doubled to 4,000+ calls per month
• Booked meetings went from 25 to 45 per week, per rep
• List building and CRM logging stopped consuming selling hours entirely

Nearly doubling meetings without hiring anyone is the whole argument for treating GTM as an engineering problem: the constraint was never how hard reps were dialing, it was how much of their day never reached a phone.

• Claude Code
• Telegram
• Serper.dev
• Firecrawl
• Claude API
• JustCall
• HubSpot

cross-client capability ReplyGen AI & Manage Digital

Scale operated

• positive replies · 17.7%
• everything else — wrong person, no thanks, auto-responders

ReplyGen AI & Manage Digital · managed cold email infrastructure run end to end for B2B clients

Two numbers do the work here. Sub-2% bounce across 500+ mailboxes is what keeps a sending operation alive — above that band, mailbox providers stop trusting the domains and the whole channel quietly dies. 99% sender reputation is the same discipline measured from the other side.

Sustaining that at 250K sends a month is a different problem from doing it on one domain. It requires domain strategy, staged warmup per inbox, volume caps tied to inbox age, list verification before anything sends, and monitoring that catches placement decay before reply rates fall off.

Why it mattered

• 54 booked meetings in a single month — a company record
• An average of 15 meetings per month, per client, across SaaS, recruitment, IT, wholesale, and defense tech
• Replies from companies including Tesla and SpaceX — deliverability good enough to reach inboxes that reject most cold mail

Verified sample

One SmartLead workspace, pulled directly from the platform, showing what the discipline looks like at campaign level:

source: SmartLead export, one workspace of several · corroborates the bounce and reply rates above at a volume that can be independently checked

Reply rates within that sample ranged from under 1% on early untuned campaigns to 7.7% on the best-performing ones — the same operator and tooling, an order of magnitude apart on list quality, segmentation, and sequence design. That spread is the argument for treating outbound as infrastructure rather than copywriting.

Open rates are absent from most of this because open tracking is switched off. Tracking pixels measurably hurt inbox placement, and an open rate inflated by bot scanners is a vanity metric anyway. Reply rate is the honest top-of-funnel signal, so that's what gets measured.

The problem

Sustained deliverability requires engineering underneath the copy: domain strategy, inbox warmup, DNS authentication, sending-volume discipline, and monitoring. Sequences and copy are the visible part of outbound. Infrastructure is the part that decides whether any of it lands — and the part most teams discover only after their domain is already burned.

The stack underneath

• SmartLead
• Instantly
• SPF / DKIM / DMARC
• domain & inbox warmup
• sequence architecture
• CRM sync

ReplyGen AI internal tooling

The problem

Every new client started the same way: sit on a discovery call, then spend half a day turning what you heard into a positioning statement, an ICP definition, a Clay table that actually returns the right accounts, and copy that matches the offer. It was judgement work, but it was also the same judgement work every time — and it delayed the first campaign by days.

What I built

A library of Claude Skills that takes the transcript and the lead form as input and produces the whole first-campaign artefact set, then builds the test campaign itself.

Why it mattered

It compressed roughly four hours of setup into under thirty minutes, which changes what a client engagement can look like: a campaign goes live the same day as the discovery call, while the conversation is still fresh, rather than the following week.

The wider point is that the leverage isn't in generating copy — anyone can do that. It's in encoding the sequence of judgements an experienced operator makes between a sales call and a launched campaign, so the reasoning is repeatable rather than living in one person's head.

• Claude Code
• Claude Skills
• Claude API
• Clay
• Instantly
• MillionVerifier

commercial real estate investment & sourcing 5.0 rating

The problem

Deal sourcing was almost entirely manual: pulling property and owner data from CRE databases by hand, filtering in spreadsheets, looking up contacts one at a time, retyping into a CRM, and sending outreach individually.

Automating it meant solving heterogeneous ingestion — the interesting engineering problem here. Some sources expose APIs (Lightbox); others offer only platform exports or CSVs (CoStar, Crexi). Everything had to be normalized into one schema, owner contact data resolved through enrichment APIs, clean deduplicated records landed in the CRM, and automated outreach run on top.

What I built

Two ingestion paths feeding one schema. The structured path pulls the CRE databases; the proprietary path pulls signal nobody else is subscribed to — deal news and blog coverage via RSS, read and classified by a Claygent agent before the opportunity shows up anywhere purchasable.

Why it mattered

Sourcing from databases means competing with everyone else who bought the same list. Reading deal signal out of news the moment it publishes means reaching the party before the opportunity is broadly known — which in deal sourcing is the entire advantage.

“A very patient and cooperative human.”

client · 5.0 · endorsed: Collaborative

• Clay / Claygent
• n8n
• RSS
• CoStar
• Crexi
• Lightbox
• Apollo.io
• Hunter.io
• Close
• Instantly.ai

Curaleaf · employer, not client Wholesale Account Manager · Nov 2023 – Nov 2024

The problem

The wholesale account base stood at 50 active buyers. Growing it meant systematic acquisition across a fragmented dispensary market, and it had to be built and operated by one person with no added headcount.

What I built

An outbound system whose differentiator was the creative, not just the cadence: each prospect received a personalized video with their own website on screen. In a market where every buyer gets the same templated email, a video that visibly could not have been mass-produced is what earns the reply.

Outcome

• Wholesale account base grew 5x in twelve months — 50 to 250+ active buyers
• $1M average monthly revenue on a personal book — roughly $12M annual production
• Quota hit 12 of 12 months, ranked #1 rep of 100 company-wide
• Delivered with no added headcount

Why it's in a technical portfolio

It proves the systems I build are grounded in a motion I've personally run at scale. I'm not automating a process I read about.

• list building
• segmentation
• sequenced cold email
• call cadence
• CRM discipline

B2B services company active, ongoing engagement

The problem

Lead generation was a chain of manual steps: researching prospects by hand, building lists in spreadsheets, sending outreach one at a time. The requirement was an engine — web scraping to source prospect data, a CRM as the structured database layer, and cold email at volume on top, connected with no manual handoff between stages.

Volume cold email made this a deliverability engineering problem as much as an automation one: an engine that sources perfectly and then lands in spam has solved nothing.

The approach

n8n orchestrating the whole chain, GoHighLevel as the structured database layer, and Instantly handling sequenced sending, with scraped and enriched records syncing back so reporting reflects reality.

• n8n
• GoHighLevel
• Instantly
• web scraping
• CRM sync

self-built and operated · proprietary platform

What it does

Sources target accounts, scrapes and enriches them programmatically, applies LLM reasoning for scoring and qualification, and stores structured output in Postgres.

• Next.js
• Supabase / PostgreSQL
• Firecrawl
• Serper
• Claude API

medical software company · sells to healthcare providers HIPAA · SOC 2 · BAA-backed contracts

The problem

The entire revenue lifecycle ran as manual handoffs across eleven disconnected systems. The failures compounded at every stage:

• Legal exposure. Prospects were seeing the product before NDAs were executed.
• Contracting by hand. Agreements were assembled manually against a pricing matrix living in a shared drive.
• Broken reconciliation. Payments landed in Stripe but reached QuickBooks without line-item revenue recognition, so subscription revenue and one-time onboarding revenue were never separated.
• Reactive customer success. No usage telemetry, no renewal runway — churn risk surfaced only after it was too late to act.

And none of it could be solved with a generic automation: HIPAA meant no PHI in automation logs or test data, SOC 2 meant real access control and audit logging, and Slack had to remain the live operational hub.

What I built

One orchestrated lifecycle across the whole stack, with each stage gated so the next can't begin until the prior one is legitimately complete. Compliance was a design input, not a review step at the end.

Why it matters

This is the exact shape of the work a RevOps team owns at a Series B–D company: quote-to-cash, revenue recognition that survives an audit, and multi-system integration inside real compliance boundaries rather than a sandbox.

Contract completed and paid in full.

engagement outcome

• GoHighLevel
• Make.com
• Stripe + Stripe Tax
• QuickBooks Online Advanced
• Slack
• Zoom
• Google Workspace

entrepreneur · 3 brands real estate · modular development · financial services 4.0 rating

The problem

Four fundamentally different processes were being run by one person: a property transaction, a development project cycle, a service onboarding, and a lending underwriting flow. Leads arrived from five form types, two Facebook pages, three Instagram accounts, and multiple websites. Nothing was structured, attributed, or automated.

The hard constraint: two of the websites were platform-controlled franchise sites that could not be edited, so standard form embeds were simply impossible.

What I built

A single GoHighLevel environment with separated brand workspaces and four custom pipelines, each modeling its own real sales process rather than forcing one generic funnel onto four businesses.

The franchise-site problem is the part I'd point to in an interview. Rather than fighting a platform I couldn't access, I built standalone branded landing pages on short URLs, so lead capture worked on properties whose code could never be touched.

Outcome

Four business lines live on one attributed system — every lead routed to the correct pipeline, tagged by source and brand, and followed up automatically. I then trained the client to independence with recorded Loom videos, a quick-reference guide, and a live walkthrough. They operate the platform themselves.

“Good experience and solid work, will use again.”

client · 4.0 · endorsed: Committed to Quality, Clear Communicator, Detail Oriented, Accountable for Outcomes

• GoHighLevel
• multi-brand workspaces
• custom pipelines
• fields & tags
• forms & funnels
• landing pages
• Meta Business Suite
• FB/IG Lead Ads
• SMS + email automation
• Loom

sales consultancy · end client's HubSpot same-day-urgent brief 5.0 rating

The problem

Clay and HubSpot needed connecting in both directions on a same-day deadline. Pulling contact and company objects out of HubSpot was the easy half.

The real requirement was write-back with matching: before Clay created anything in HubSpot, it had to query for an existing record, decide whether it matched, and branch accordingly. Without that lookup layer, every single sync run would manufacture duplicates — and the native integration couldn't handle the field mapping or conditional logic this needed.

What I built

Four production Clay tables orchestrated through Make.com and direct HTTP calls to HubSpot's API — two for full import, two implementing the lookup-and-match layer that makes the write-back safe.

Outcome

A same-day-urgent brief delivered end to end in roughly 31 hours from contract start, with duplicate-safe bidirectional sync live in the client's production HubSpot instance.

“Evan was great to work with, knowledgeable, fast, and clearly experienced with Clay and Make.com. He quickly set up complex HubSpot integrations exactly as needed. Highly recommend for any Clay work.”

client · 5.0

• Clay
• HubSpot Contacts / Companies / Search API
• Make.com
• direct HTTP/JSON

marketing agency · inbound + outbound 11-week engagement

The problem

Lead sources were labeled inconsistently, so one channel appeared under multiple names and none could be measured. There was no standardized campaign tracking structure. Pipedrive's fields and mapping didn't match the business, forms and CRM were disconnected, and leadership had no recurring view of performance across channels.

What I built

Audited the existing form and CRM setup, then standardized the lead source taxonomy and campaign tracking structure — the unglamorous work that makes every downstream number trustworthy. Configured Pipedrive custom fields and mapping to match the actual business, automated form-to-CRM capture, and built a weekly performance report.

Delivered with full documentation and a recorded Loom walkthrough, so the team could operate and extend the system without me.

Eleven-week engagement, completed and paid in full.

engagement outcome

• Pipedrive
• Make.com / Zapier
• web forms
• Google Sheets
• UTM / campaign taxonomy
• Loom

multi-brand digital publisher · advertising & placements 5.0 rating

The problem

Leads arrived through forms on multiple websites with no source attribution and no brand-correct response path. Getting that right meant solving three things at once:

• Brand identity. Replies had to come from the right publication's inbox in the right voice, and follow-up had to differ by lead type — resellers, businesses, and personal brands don't get the same sequence.
• Deliverability. Separate sending domains, each needing DKIM, SPF, and custom-domain authentication kept intact. Multi-brand sending is where deliverability quietly dies.
• Knowing when to stop. Prospects asked the same pre-purchase questions endlessly and a human answered every one — but legal questions and pricing exceptions had to escalate to a person rather than be improvised by an AI.

The full recommended stack had to run under $500/month, and the client wanted architecture advice, not just execution.

The approach

An AI conversational layer handling the repetitive pre-purchase questions, sitting on top of per-brand email infrastructure with its own authenticated sending domains, source-tracked forms feeding a central CRM, and multi-touch sequences branched by lead type — with explicit escalation rules routing legal and pricing questions to a human.

“Evan was really easy to work with, and I would hire him again in the future!”

client · 5.0

• LLM conversational layer
• per-domain DKIM / SPF
• multi-brand email infrastructure
• source-tracked forms
• CRM + central reporting
• sequences by lead type

self-built and operated

The problem

Dialer platforms report activity, not performance. There was no way to analyze call outcomes against rep behavior — connect rates by time of day, talk-time patterns, disposition trends — because the data lived inside JustCall with no structured export.

The approach

Webhook-driven capture of call records into Postgres, giving the structured layer the dialer never exposed, so rep behaviour can finally be measured against outcomes.

Why it matters

This is a revenue analytics build — the thing a RevOps team does with a warehouse. The data model gets designed first; the automation serves it.

• JustCall API
• Supabase / PostgreSQL
• webhooks