AYAdrian Young
Customer Success Leader · Dublin, Ireland

I build Customer Success organisations, and the AI systems that run them.

Over ten years leading post-sales, account management and revenue teams at Deliveroo, Snapfix and high-growth startups, across Ireland, India, EMEA and North America. At Snapfix I own post-sales across 950+ accounts and build the software the team runs on: an AI operating layer, a permission-first MCP server, and an onboarding platform I shipped hands-on with Claude Code.

Get in touch LinkedIn
104.8%
Net revenue retention
950+
Accounts owned, post-sales
47%
Of Q1 2026 closed-won, CS-sourced
34→44
NPS, and climbing
93.9%
Renewal rate
The operating model

Three motions, one lean team, an agent layer underneath.

Customer Success stopped being a cost centre the day it started sourcing revenue. I run coverage across the full customer range by matching the motion to the account, then using AI to create capacity instead of adding headcount.

01 / High-touch

Strategic enterprise

High-touch management for the strategic enterprise accounts, where customer value, complexity and risk are highest and the commercial outcome needs a person, not a workflow.

02 / Scaled

Coverage through AI

Scaled coverage across 950+ accounts through AI-enabled workflows. Account health is scored across the estate with tiered intervention sequences, so the team's time goes where judgement moves the number.

03 / Digital

Onboarding as a product

Digital success through a self-serve onboarding platform that moved sub-enterprise customers off CSM-led calls into a product flow, cutting onboarding from about 12 weeks to five.

The AI layer, running in production

Production workflows on Claude, agents and MCP integrations that automate CRM call summaries, account health scoring, intervention sequences, QBR preparation, renewal workbooks and daily account briefings across 950+ accounts. QBR preparation has gone from days to 15 minutes.

Claude API Agents MCP Risk scoring QBR gen Renewals
Selected work

Systems I've shipped, hands-on.

Most Customer Success leaders sponsor an AI initiative. I design and build the systems myself, then write the guardrails that make them safe to run in production. A sample of what that looks like.

Onboarding platform

Product · full-stack

An end-to-end onboarding platform, built hands-on with Claude Code and deployed on Vercel. A 19-screen CSM operations dashboard, with daily triage, an email approval queue, a pipeline board and live CRM and tenant sync, plus a magic-link customer portal for setup, document upload, training and live progress. Onboarding quality is the strongest churn predictor in the estate.

~41Klines TypeScript
165PRs merged
56API endpoints
12→5onboarding weeks
Next.jsPostgresVercelClaude Code

MCP server, permission-first

Platform · infrastructure

A production MCP server that lets AI agents read and safely write across the post-sales system, with the permission model designed first. Per-user bearer tokens, issued and revoked from settings and tied to a named identity. It fails closed: missing, unknown or revoked tokens get a 401 before any tool runs. Reads come from the same shared builders as the dashboard, so the agent and the UI cannot drift; writes replicate each route's full validation, logging and provenance.

No tool sends email, approves a draft, or moves a pipeline stage.
MCPOAuth + PKCEFail-closed authHuman-in-the-loop

Autonomous bug-fix agent

Autonomous agent

A scheduled agent that takes the newest reported bug each weekday, writes a fix on its own branch and opens a pull request for review. It can only propose: it never merges or pushes to main, and feature requests are never auto-built. They wait for a human by design.

GitHub ActionsClaudePropose-only

AI operating layer for post-sales

AI systems

The layer the CS team runs on. Call transcripts flow structured into the CRM, from Fathom to Claude to HubSpot. Churn risk is scored across the whole estate with tiered intervention sequences, and QBR decks, renewals workbooks, daily account briefings and health dashboards are generated rather than assembled.

950+accounts scored
days→15minto prep a QBR
Claude APIAgentsHubSpotRisk scoring

Multi-agent follow-up program

Multi-agent system

A designed multi-agent system with a separation of powers. Analysis observes the numbers, an orchestrator decides and sets priorities, workers act on accounts, a composer writes, and deterministic code enforces every guardrail. Compliance is encoded as data, not promised: kill switches that fail closed, recipient allowlists, frequency caps, and a written constitution of 20 testable rules. A composer validator checks each message before it sends: exactly one URL and it must be the account's own, and every figure must appear in the source facts.

Analysis observes, the orchestrator decides, workers act, the composer writes, code enforces.
Separation of powersGuardrails-as-codeDPO-gated

Agent governance & routines

Governance

Seven agent routines running in production, from daily summaries and meeting briefs to a Friday weekly wrap that reads across six systems, drafts replies and never sends them. Audits run report-only, with no automated writes. The standing rule is evidence or nothing: every action item must cite a source that actually appeared in a tool result, or it is dropped.

Report-onlyDrafts-onlyEvidence-cited
By the numbers

Retention and expansion, owned.

104.8%
Net revenue retention
91.2%
Gross revenue retention
<10%
Annualised churn, from 20-30%
47%
Of Q1 2026 closed-won, CS-sourced
88
Closed-won expansion opportunities
15min
To prep a QBR, from days
12→5
Onboarding, weeks
40+
Countries served
Experience

Ten years in post-sales and revenue leadership.

Snapfix
Oct 2024 — Present
Director of Customer Success
Operational management SaaS for hotel groups, care homes, facilities and property portfolios. Own post-sales across 950+ accounts in more than 40 countries, concentrated in the UK, Ireland and North America.
  • Retention, renewals and expansion. 104.8% net revenue retention, 91.2% gross revenue retention and a 93.9% renewal rate. Annualised churn from 20-30% to single digits. A six-figure sum in recurring CS-sourced expansion across 88 closed-won opportunities, and roughly 47% of Q1 2026 closed-won revenue.
  • AI-first CS at scale. Production workflows on Claude, agents and MCP integrations that automate CRM call summaries, account health scoring, intervention sequences, QBR preparation, renewal workbooks and daily briefings across 950+ accounts. QBR preparation from days to 15 minutes.
  • An international enterprise customer base. Customers from single-site properties at entry-tier ARR to enterprise accounts, spanning global hotel and hospitality groups, commercial property and real estate, professional services, and engineering and facilities management, including multi-billion-turnover organisations.
  • Team and solutions engineering. Lead a team of seven across hubs in Dublin and India, with Dublin team leads running a second operational hub built from scratch and under budget. Also own and personally run the solutions engineering function.
  • Onboarding, rebuilt as a product. Designed and built an end-to-end onboarding platform with Claude Code, Next.js and Postgres. Cut onboarding from about 12 weeks to five, and moved sub-enterprise onboarding from CSM-led calls to a self-serve product flow.
  • Product and customer feedback. Work with Product and Engineering to turn customer signals into roadmap priorities: an overhaul of the checklist experience, enhanced QR-code workflows, and a redesign of user onboarding and creation from a managed service into a self-serve flow.
FoodCloud
Jun 2023 — Oct 2024
Head of Partnerships
Led the account management team for Ireland's leading food redistribution platform. Exceeded income targets and deepened engagement across the retail partner network.
High-growth startup
Jan 2021 — Dec 2022
Head of Customer Success
Book and team, simultaneously. Owned a 30-account enterprise book of contracts ranging from mid-tier to seven-figure ARR, including global technology, marketplace and consumer brands, while building and leading a distributed CS and AM team of eight across EMEA and North America.
  • Carried P&L, forecasting and revenue targets.
  • Led GTM operations for the On-Demand product launch, including pricing and self-serve strategy.
Deliveroo
Mar 2017 — Aug 2020
Head of CS / Account Management, Ireland
Managed key national accounts across a 1,000-client portfolio, delivering average 110% year-on-year growth through acquisition, retention and data-led account management.
  • Hired and led a commercial team of four to six.
  • Ran new market and product launches with Operations, Marketing and Product.
Kerry Group
Sep 2014 — Dec 2016
Commercial Executive
Sales, category management and marketing.
Toolkit

What I build with.

I build the automations my teams run on, then coach the team to use and extend them.

AI & automation
Claude APIAgentsSkillsMCP integrationsAgentic workflow designPrompt-driven automation
Revenue stack
SalesforceHubSpotFathomLookerTableauEqualsSQLChurn & cohort modelling
Build
TypeScriptNext.jsPostgresVercelClaude CodeMCP servers
Education & languages
Diploma, AI for Business — UCD Professional AcademyMerit · 2024
MSc Business — University College Cork1.1 · 2014
BA Hons — Dublin Institute of Technology1.1 · 2012
English (native), French (conversational), Spanish (basic)Languages
Let's talk

Building a Customer Success engine that compounds?

I am open to leadership conversations where commercial ownership, international team building and AI-enabled scale all matter at once.

Email me Connect on LinkedIn