HomeCase Studies
Case Studies

From week one to an agent factory.

Every engagement below started with one system shipped in days. Some stayed focused. One grew into 25 workstreams across an entire GTM org. Clients stay anonymous by design, so the industries, the systems, and the outcomes tell the story.

Select an Engagement
Life Sciences QMS Software Mid-market Fractional retainer 25 workstreams

An agent factory at portfolio scale.

The Problem

The GTM org was scaling on manual effort. Reps transcribed their own call notes, marketing tracked competitors by hand, customer success read the news account by account, and leadership waited on weekly status someone had to assemble. Plenty of AI curiosity, zero production systems.

The Solution

An agent factory built in waves. The first wave was marketing-led: competitive intelligence, executive LinkedIn engagement, brand-voice content skills. The second moved into demand gen with ICP scoring, lead routing, dormant-account re-engagement, and LinkedIn outbound. The third opened the AE, CS, and executive layer: after-call notes, per-rep call prep, account watchers, account planning, and a chief-of-staff brief for the CMO chair.

01Pick the painOne team, one painful workflow, chosen together.
02Ship week oneFirst system live in days, not after discovery.
03Name an ownerA person on the team reviews and inherits it.
04Expand the waveAdjacent teams see it working and ask for theirs.
05Aggregate upDaily and weekly rollups ride on the agents below.
What Shipped

ICP scoring and routing

Every inbound lead scored on a five-dimension rubric with sanctions and export-control checks layered in. High-value leads skip the queue on a fast-track lane.

After-call notes to CRM

A call recording becomes a structured note in Slack for rep review, then writes back to the CRM. Branches per role, including a SPICED variant for AEs.

Competitive intelligence

A weekly brief covering 11 competitors with urgency-ranked findings, plus just-in-time kill sheets triggered by signals from sales calls.

Executive LinkedIn engagement

Posts worth engaging identified daily, comments drafted in the executive's voice, approved, and published. Running continuously since January.

BDR re-engagement

Dormant accounts filtered from the CRM, researched by parallel agents, and delivered as a Monday hit list with dossiers attached.

Customer success watchers

Daily, weekly, and monthly account briefs across 34 accounts, covering FDA actions, leadership changes, clinical milestones, and funding events.

Account planning

A 2 to 3 page plan per top renewal account, built from CRM data, a year of call transcripts, and public research, rendered for the role reading it.

The executive layer

Weekly reporting, a daily lead-activity pulse, and a chief-of-staff pattern: morning briefs, meeting agendas, decision logs, and an action tracker.

The Outcome
25
Distinct workstreams in flight across the GTM org
14
Agents in production, each with a named owner on the team
6
Functions covered, from outbound to the executive chair

The maturity signal: new asks increasingly land as aggregation skills on top of agents that already exist, instead of requiring new builds. The team owns every system.

Tool Chain
ClaudeCoworkMakeSalesforceGongHubSpotHeyReachSlack
Quote-to-Cash SaaS Early stage Project + retainer

An AI SDR motion with a human gate.

The Problem

Pipeline targets without SDR headcount. Lead lists sat unworked, follow-ups were template blasts, and competitive intel lived in screenshots passed around Slack.

The Solution

An outbound system that does the research and the writing while humans keep the judgment, surrounded by a weekly competitive scan, visitor intelligence, and an ad engine with brand QA.

01Resolve the listLeads matched against the CRM and enriched.
02Write per leadFive-email persona sequence plus LinkedIn touches.
03Human reviewEvery sequence staged in a workbook for approval.
04Push and stampSequencer enrollment with the CRM stamped back.
05Handle repliesReplies classified, responses drafted in the AE's voice.
The Outcome
5-touch
Researched sequences written per lead, not per template
14
Competitors scanned weekly with deduplicated findings
100%
Of outbound human-approved before it sends

The motion runs daily without an SDR hire. Visitor intelligence classifies flagged site visitors as buyer, competitor, or partner, and a Friday report rolls the week up for leadership.

Tool Chain
ClaudeHubSpotClayLemlistFirecrawlSlack
Mobile App Security Mid-market Signal plays

Signal plays, and a Friday report that writes itself.

The Problem

Re-engagement timing was calendar guesswork, displacement messaging came from static battlecards, and the Friday status report ate someone's afternoon every week.

The Solution

Four signal-driven plays from the play library run here: closed-lost re-engagement, micro-campaigns, champion tracking with warm intros, and competitor displacement. Stacked signals decide the timing. Reps decide the send. A weekly executive report assembles itself from the week's GTM activity.

01Collect signalsJob changes, funding, postings, reviews, intent.
02Detect the stackMultiple signals on one account inside a window.
03Fire the playThe matching play drafts outreach with the evidence.
04Rep reviewsTier 1 routes to a human. The rest sends itself.
05Friday rollupThe week's activity lands as one executive report.
The Outcome
4
Signal plays live, timed by evidence instead of calendars
1
Friday report, zero manual assembly

Leadership reads one document on Friday instead of chasing five threads, and re-engagement fires when something actually changed at the account.

Tool Chain
ClaudeSumbleG2ClaySEPSlack
Cyber Deception & Security Growth stage Outbound direction

An SDR director in software.

The Problem

Outbound had volume tools but no director. Which accounts, which personas, which message, in what order: that judgment lived in nobody's job description.

The Solution

The direction got encoded into an agent that selects targets, matches personas, sequences touches, and preps the humans. Closed-lost re-engagement and micro-campaign plays run alongside it.

01Select accountsFit and signal decide who gets attention this week.
02Match personasThe right buyer, the right angle, per account.
03Sequence touchesOrdered outreach built for the committee, not a contact.
04Prep the repA research brief and personalized deck land 30 minutes before every first call.
The Outcome
30 min
Before every first call: brief and deck, automatically
3
Plays live alongside the SDR director pattern

Reps walk into first calls prepared like the best rep on their best day, and the account selection logic survives every personnel change.

Tool Chain
ClaudeClayGoogle CalendarGoogle SlidesSEP
Energy Data & Analytics Enterprise Data quality

Greenfield accounts, enriched and reconciled.

The Problem

Thousands of greenfield records where the corporate family was a guess. Parents, operating LLCs, aliases, and former names blurred together, and Salesforce disagreed with reality on all of it.

The Solution

A research and reconciliation pipeline where humans review verdicts instead of spreadsheets.

01ResearchCorporate family mapped from the public web, with citations.
02ValidateEvery row checked against a strict enrichment schema.
03CheckpointWaves committed in batches, so nothing runs away.
04ReconcileField-by-field comparison against Salesforce.
05VerdictsMatch, differ, or empty, feeding a reconcile policy.
The Outcome
1000s
Of records researched with a citation on every claim
3
Verdict types replace cell-by-cell spreadsheet review

Account executives prospect against corporate families that are actually correct, and RevOps reviews exceptions instead of everything.

Tool Chain
ClaudeSalesforceWeb ResearchExcel
Education Technology Enterprise Product marketing

A product marketing asset factory.

The Problem

Every release meant rebuilding the same launch assets by hand, at every tier, against deadlines that did not move. Brand voice drifted with whoever wrote fastest.

The Solution

Release tiering decides the deliverable pack, and the pack generates itself against the team's messaging framework and brand voice.

01Tier the releaseClassification decides how big the launch treatment is.
02Select the packTier 2 and tier 3 deliverable sets, predefined.
03Draft in voiceAssets generate against the messaging framework.
04ReviewPMM approves and adjusts, never starts from blank.
05File itAssets land in the project tool and doc stack in use.
The Outcome
Tiered
Asset packs matched to release size, every time
1
Brand voice across every asset, enforced by the skill

Launches stop being a scramble of rebuilt documents. The team reviews strong drafts instead of writing cold ones.

Tool Chain
ClaudeClickUpGoogle DocsGoogle Drive
FAQ

About these case studies.

Why are the case studies anonymous?

Client names and client data never appear in public work. That is a standing rule, not a styling choice. Industries, systems, and outcomes still tell the story, and anything that would identify a client is removed rather than changed.

Are these real engagements or composites?

Real. Each case study maps to one company and one engagement. Nothing is merged, inflated, or invented, which is exactly why the names have to stay off.

How long do these engagements run?

They start with a single system shipped in week one. Retainers run month to month and recalibrate against what the portfolio needs. The deepest engagement here grew from one workstream to 25 over a number of months. The shape of that arc is on the how we work page.

Can I talk to a reference?

Where a client has agreed, yes. Anonymity on the site does not mean anonymity in diligence. Ask on a call and I will connect you directly.

Ready to get started?