( Selected case studies ) Tamara Omomo / AI GTM Engineer

The
case studies.

Seven systems from the seam between marketing, sales and code. Each one replaced a manual, leaky process — and moved a number I can point to.

( 07 — and counting )
07 · Latest

Synthetic-persona outbound engine

15minPer campaign, from days of copywriting

Business problem

Every outbound campaign took days of manual copywriting — and still sounded generic. What actually convinces a CIO versus a CFO versus a Head of Procurement lived only in scattered sales calls. That knowledge never reached the emails prospects received, and campaign reviews showed it: generic first messages got ignored.

The solution

I took transcripts from the team's real sales conversations — pain-point calls with 7 AEs and SDRs — and synthesised them into 5 synthetic buyer personas: pains, buying triggers, objections and counters, per role, size band and industry. Not guessed from a whiteboard; drawn from what buyers actually said. New sales-call transcripts sync from Google Drive hourly and refresh the personas automatically. Upload a contact CSV and every lead is matched to a persona, enriched, and handed a unique email + LinkedIn sequence written in that persona's language (German Sie-form and English), grounded in real transcript quotes. A human review queue, a compliance blocklist and a canary test-send gate every campaign before a one-click push to Lemlist.

Business impact

Every first line opens with the recipient's role-specific pain in their own language — a CFO reads about budget predictability, a CIO about ticket volume. Per-campaign effort collapses from days to a 15-minute review, scaling the team's best hand-written outreach to 500+ contacts. The playbook compounds: every new call makes the next campaign sharper, even as people move on.

Workflow live run
Sales-call transcripts · Drive Synthetic persona synthesis Contact CSV · persona-match + enrich AI writes email + LinkedIn Human review · compliance · canary Lemlist
Tech used
ClayPerplexityLemlistAnthropicClaude CodeCustom Tool
01

Always-on outbound engine

Daily leadsNew conversations every day, on autopilot

Business problem

Right-fit companies were visiting the site and leaving anonymously. There was no way to spot them, find the right people and reach out before the intent went cold.

The solution

An always-on system that tracks which companies visit the site, pulls the domain, identifies the right contacts by job title, and pushes them into parallel LinkedIn + email sequences across multiple senders — with a human approval gate before anything sends.

Business impact

The campaign generates fresh leads and new conversations every day — a steady, predictable flow of replies and booked calls instead of one-off blasts. A Slack channel surfaces each company and contact in real time.

Workflow live run
Website visitor Company identified Contacts found by title Human approval LinkedIn + email sequences Booked calls
Tech used
ClayLemlistSlackClaude CodeCustom ToolLinkedIn
02

Inbound routing engine

11minMedian response, from 4 hours

Business problem

Leads arrived from everywhere — website inbound, paid search, LinkedIn ads — into different tools and inboxes. Round-robin couldn't account for fit, source or rep capacity, so high-intent demos went cold before anyone replied.

The solution

A routing engine that unifies every lead source into one flow. It scores each lead, round-robins across the sales team by territory and live capacity, enrolls them in the right drip sequence based on score, and hands the rep full context — source, enrichment and history.

Business impact

Median first response dropped from 4 hours to 11 minutes, the speed-to-lead SLA held above 95%, and no lead slipped through the cracks regardless of where it came from.

Workflow live run
Every lead source Score + qualify Route by territory + capacity Score-based drip sequence Rep with full context
Tech used
HubSpotNodeSlackRevenue Ops
03

AI research agents

+38%SQL → opportunity conversion

Business problem

Reps burned hours on unqualified accounts. CRM records were thin and stale, and off-the-shelf enrichment couldn't answer the nuanced, research-style questions that actually decide fit.

The solution

Custom Clay AI agents that research each account the way a human SDR would — reading sites, news and signals to pull firmographic, technographic and intent data, qualify against our ICP, and write a clean fit score back to the CRM automatically.

Business impact

60k accounts agent-enriched and continuously refreshed; SQL-to-opportunity conversion up 38%. Reps only touch accounts worth touching.

Workflow live run
Account list AI research agents Fit + intent scoring CRM write-back Prioritised outreach
Tech used
ClayAI AgentsPythonEnrichmentScoring
04

Event-driven campaigns

Pre-bookedExec calendars, filled before each event

Business problem

Conferences are full of right-fit buyers, but exec time on the ground is scarce. Leadership showed up without booked meetings or context on who was even worth meeting.

The solution

A two-part engine. Before the event, scrape the attendee list, identify the right accounts and run targeted campaigns to book meetings onto exec calendars. Then a custom briefing tool gives each exec full context per contact — slot, background, and whether the meeting happened.

Business impact

Execs walked in with pre-booked, well-briefed meetings instead of cold floor-walking — and because the tool tracked what happened, follow-up campaigns fired automatically, turning conversations into pipeline.

Workflow live run
Attendee list scraped Target accounts identified Pre-event campaigns Exec briefing tool Meeting tracked Automated follow-up
Tech used
LovableClaude CodeClayCustom ToolLemlist
05

Multi-touch attribution

$1.1MSpend reallocated to what closes

Business problem

Spend decisions ran on last-touch guesses. Marketing and finance argued endlessly over what actually drove pipeline.

The solution

A SQL/dbt multi-touch attribution model that stitches every touch to closed revenue in one source of truth, surfaced in a live dashboard everyone can query.

Business impact

$1.1M of spend reallocated toward the channels and plays that actually close — with the data to defend every call.

Workflow live run
Every touchpoint dbt models Multi-touch weighting Live dashboard Budget decisions
Tech used
dbtSQLHockeyStackAnalytics
06

In-house tooling

Self-serveGTM team, off the ops backlog

Business problem

The GTM team waited on ops and eng for every dashboard, list pull or one-off report — a constant bottleneck.

The solution

Internal dashboards, Slack apps and lightweight data tools made in-house — so the team can answer its own questions in seconds.

Business impact

GTM self-serves on the data it needs, and the ops backlog shrank — freeing engineering for the systems that matter.

Workflow live run
GTM question Internal tool / Slack app Live data Self-serve answer
Tech used
Claude CodeReactPythonSlack Apps

Currently Building

{ The lab — live experiments }
In progress

AI Deal Desk

An agent that assembles pricing, legal and approval context the moment a deal hits stage three.

In progress

Pipeline Copilot

Weekly pipeline review, automated — surfaces stalled deals, coverage gaps and next actions.

Exploring

Customer Health Agent

Reads product usage, support tickets and CRM signals to flag churn risk before the QBR does.

Exploring

AI Account Research

Deep-research briefs on any target account, generated on demand for reps before every call.

( Let's ship better GTM systems )

Let's ship.

Book a conversation
OMOMO