Under the Hood

The technical architecture behind the pitch

Headless system design, the AI operating layer, live automations built and shipped, the full tool stack, and certifications — the proof behind everything on the homepage.

Solutions · Headless Strategy

One system of record. Any experience on top.

Headless means the data core and the interfaces are decoupled. The record lives in one governed place — and every team, tool, and AI agent consumes it through the same APIs. Swap any surface without touching the core.

Live Demo — headless in motion
Claude — connected to your revenue stack
Which accounts need my attention this week?
Salesforce
HubSpot
Gong
NNooks
Clay
ZoomInfo
Pulled live signals from your 6 connected systems. Here's what needs attention this week:
Top Priority Accounts — This Week
Experience Layer — composable surfaces
Rep Workspace
The six fields that matter — not three hundred
Exec Dashboard
Forecast, pipeline, and NRR in real time
Customer Portal
Self-serve quotes, orders, and renewals
AI Agents
First-class consumers of the record — not bolt-ons
Slack & Email
Updates delivered where work already happens
one API contract, every surface
API Layer
Auth & permissions REST / GraphQL Webhooks MCP for agents
governed, versioned, audited
System of Record
Clean data model · Single source of truth · Stage logic & governance · Full audit trail
Swap tools without migration Ship a new view in days, not quarters No UI vendor lock-in AI-ready by design
Reference Architecture — Live Diagram
Open full size ↗
Explore a Headless Architecture →
Under the Hood · Applied AI & Automation

What I've built and shipped

Real automation workflows across GTM operations, Finance, Compliance, and CS. Tool-specific, outcome-driven, production-ready.

01
Lead Management
Inbound Lead Qualification & SDR Routing
n8nOpenAI APIHubSpotCRM
↑AI scores inbound by ICP fit, urgency, and source — routes to SDR with full context attached. Improved qualification accuracy and eliminated manual triage.
02
Finance Ops
Purchase Order Extraction & Auto-Sync
MuleSoftIDPSystem of RecordFinance Systems
↑IDP extracts PO data from inbound email — MuleSoft syncs to the system of record and Finance automatically. 70% faster processing, near-zero manual entry.
03
Compliance
Automated Sanctions Screening
RPAScreening APIsCRM
↑Customer records checked against sanction lists automatically. Matches trigger a case and alert; non-matches cleared instantly. Full compliance at scale.
04
Sales Ops
AI Agent-Driven Quote Generation
AgentforceCPQ
↑Agentforce gathers account context, product config, and pricing rules — auto-generates draft quote routed for approval. Faster turnaround, fewer errors.
05
Pipeline Ops
Email-to-Opportunity Auto-Creation
n8nOpenAICRMSlack
↑Inbound emails parsed by GPT for intent and value. When criteria met, an Opportunity is created automatically and the AE notified. Zero missed inbound pipeline.
06
Deal Governance
Multi-Step Deal Approval Automation
n8nSlackCRM
↑Opportunity criteria trigger a Slack-based approval chain. Every decision is logged back to the system of record with a full audit trail. No lost threads.
07
Deal Intelligence
Agent Spotlight
Multi-Source Deal Review Agent — Salesforce · Gong · ZoomInfo
Salesforce Gong ZoomInfo Claude GPT-4o
↑Cross-references CRM stage data, Gong conversation intelligence, and ZoomInfo buying group coverage — runs SPICED exit-criteria checks automatically and outputs a structured verdict with ranked recommended actions. No analyst required.
AI Operating Layer

Intelligence embedded inside the operating layer

AI is treated as infrastructure, not theatre: system-of-record context, workflow triggers, deal intelligence, and Claude Workspace playbooks routed through the tools teams already use.

☁
System of Record
CRM · Warehouse · Finance · CS
◆
Claude Workspace
Playbooks · Reasoning · Execution
Technical Agents 6 agents
🏗
Architect
Data model · System design
📋
Manager
Sprint · Coordination
💻
Dev
Workflow · APIs · Automation
✅
QA
Test plans · Coverage
⚙
Admin
Config · Access · Governance
🧪
UAT
User testing · Sign-off
Business Agents 4 agents
⚠
Deal at Risk
Stall detection · Alerts · Next step
📊
Deal Review
Pipeline review · Coaching
👥
Buying Group
Stakeholder mapping · Gaps
📝
Deal Administrator
System hygiene · Docs · Close plan
Claude Workspace CRM / System Triggers Multi-Agent Orchestration Real-time System Context MCP Protocol Tool Use + Function Calling
Operating Architecture

The Bowtie model

Revenue doesn't stop at closed-won. The Bowtie maps the full customer lifecycle — acquisition on the left, retention and expansion on the right.

Bowtie Operating Model — Full Customer Lifecycle
Winning by Design Model
Awareness
10K
Prospects reached
Clay ICP signals
Intent data
Education
2.4K
CR1 · 24%
Outreach sequences
Content nurture
Selection
480
CR2 · 20%
MEDDPIC scoring
Gong call intel
Prioritise
144
CR3 · 30%
Deal risk agent
Champion tracking
Closed Won
32
22% CR4
ARR
£1.2M
Onboard
28
CR5 · 88%
CS handoff
TTFV tracking
Adopt
24
CR6 · 86%
Gainsight health
Adoption scores
Expand
18
CR7 · 75%
Upsell triggers
Churn risk agent
Advocate
11
CR8 · 61%
NPS tracking
Referral engine
Acquisition (Marketing + Sales)
CR1–CR4 · Lead to Closed
Retention & Expansion (CS)
CR5–CR8 · Customer Lifecycle
AI agents at every stage
GTM Systems Troubleshooting — Where the Bowtie Breaks
Diagnostic Map

Each broken Bowtie stage has a pattern. This maps the failure mode, symptoms, the right tool, and the AI fix.

01
ICP & Awareness Misalignment
Wrong accounts entering the Bowtie. High volume, low CR1. Clay signals ignored. ICP criteria undefined in CRM.
ClayICP scoringSegment design
TOP
↓
02
Education → Selection Drop (CR2)
MQLs failing to convert to SQLs. No MEDDPIC framework. Marketing-to-Sales handoff undefined.
Lead scoringMEDDPICHandoff SLA
L1
↓
03
Pipeline Hygiene & Forecast Failure (CR3–4)
Stale deals inflating pipeline. Core fields empty. Forecast built on dirty data. Board numbers miss every quarter.
System of recordHygiene agentForecast model
L2
↓
04
Knot Failure — Onboarding Drop (CR5)
CS handoff from sales undefined. No TTFV tracking. Right side of the Bowtie starts broken before it begins.
CS handoffTTFVGainsight
L3
↓
05
Adoption Stall & Churn Risk (CR6–7)
Gainsight health scores not connected to CRM. Churn invisible until renewal. Expansion plays undefined.
GainsightChurn agentNRR model
L4
↓
AI
AI Agent Fix — Deployed at Every Bowtie Stage
Clay enrichment fixes CR1. MEDDPIC scoring agent fixes CR2–3. Forecast agent fixes CR4. CS handoff agent fixes CR5. Churn risk agent fixes CR6–7.
Claude GPT-4o Agentforce
AI
Deal Review Agent — ZoomInfo · Gong · Salesforce Triangulation
Agent Spotlight

Three data sources. One AI agent. A structured deal brief in seconds — no analyst required. Cross-references stage fields, conversation intelligence, and buying group coverage against SPICED exit criteria to output a verdict and ranked next actions.

Salesforce
Stage · SPICED fields
Contacts · Close date
Activity log
ZoomInfo
Buying group
C-suite org map
Coverage gaps
Gong
Calls · Emails
Sentiment · Velocity
Engagement recency
›
›
›
Deal Review
Agent
SPICED scoring
Buying group check
Risk assessment
Verdict engine
›
Output Brief
✓Stage timeline & stall flags
●SPICED exit criteria gaps
◈Buying group coverage map
◆Advance / Hold verdict
→Ranked next actions
Stack Salesforce Gong ZoomInfo Claude GPT-4o Production · Enterprise SaaS
Integration Architecture

The integration map

Every data source, tool, and system — connected through an agentic layer that acts, not just observes.

Full-Stack Integration Map — 5-Tier Agentic Architecture
Agentic
Tier 1 — Data Sources
🌐
Website
Analytics
◎
Clay
ICP enrichment
◈
LinkedIn
Intent signals
⬟
Outreach
Sequences
⬡
Gong
Call recordings
◉
Gainsight
Health scores
◌
Billing
MRR / ARR
API / ETL
▼
Tier 2 — Integration & ETL
⬡
n8n
Orchestration
◈
MuleSoft
Enterprise API
◎
REST APIs
Native sync
⬟
Webhooks
Real-time
◉
dbt
Transforms
Normalised to CRM
▼
Tier 3 — Systems of Record + Data Warehouse
⬡
CRM
Salesforce / HubSpot
◈
Finance
Billing / ARR
◎
BigQuery
Warehouse
⬟
Snowflake
Data lake
◉
Tableau
BI
Feeds AI agents
▼
Tier 4 — Agentic AI Layer
Prospecting
Clay Enrichment Agent
In: Domain, ICP criteria, role
Out: Enriched contact + brief
+38% reply rate
Qualification
MEDDPIC Scoring Agent
In: Call transcript, CRM stage
Out: Score, gaps, actions
+12pts CR2 conversion
Deal Intelligence
Deal Risk Agent
In: Gong signals, CRM, activity
Out: Risk score 0–100
-34% surprise losses
Forecasting
Forecast Agent
In: Pipeline state, rep history
Out: Predicted close, flags
+40% forecast accuracy
Pipeline Ops
Hygiene Agent
In: Stage velocity, stale flags
Out: Alerts, auto-updates
62% → 94% hygiene
Customer Success
Churn Risk Agent
In: Gainsight health, NPS
Out: Risk tier, playbook
-28% gross churn
Actioned outputs
▼
Tier 5 — Business Outputs
◎
Board Forecast
Weekly commit
◈
Pipeline Review
Coverage pack
⬡
Rep Alerts
Slack / Email
⬟
CS Playbooks
Auto-triggered
◉
Exec Dashboard
Real-time
◌
NRR Reports
ARR + churn
Core Capabilities

Where I operate

End-to-end systems thinking across the full GTM stack — from architecture decisions to automation execution.

⬡

GTM Architecture

  • Bowtie model design & implementation
  • GTM motion design (PLG, SLG, hybrid)
  • System-of-record design, data modelling & governance
  • Opportunity, lifecycle, and commercial workflow design
  • Cross-functional alignment frameworks
◈

AI & Automation

  • LLM-powered deal intelligence agents
  • Clay-powered prospecting & enrichment
  • Quote & approval workflow automation
  • AI-assisted call analysis via Gong
  • Agent orchestration (Claude Workspace, n8n, Agentforce, MuleSoft)
◎

Forecasting & Pipeline

  • Bowtie-aligned forecast modelling
  • Pipeline coverage & velocity analysis
  • Risk-weighted opportunity scoring
  • Commit / best-case / upside tiering
  • Real-time executive dashboards
⬟

Integrations & Data

  • System of record ↔ Gong ↔ Clay ↔ warehouse
  • BigQuery / Snowflake data pipelines
  • MuleSoft & Zuora integration architecture
  • Marketing attribution modelling
  • API-first integration design
Tools I Use

Hands-on with the full stack

Not slideware — tools I run daily across CRM, data, GTM process, and AI.

Tool Stack

How I build

Opinionated on tooling. I pick systems operators can maintain — not just engineers.

Systems of Record
Salesforce HubSpot CPQ Workflow Automation
GTM Process
Gong Outreach Salesloft Clari
Martech
Clay Apollo ZoomInfo LinkedIn Sales Nav
Data & BI
BigQuery Snowflake Tableau Looker dbt
Automation
n8n MuleSoft Make.com IDP / RPA REST APIs
AI
Claude Workspace GPT-4o Agentforce Cursor
CS Tech
Gainsight ChurnZero Intercom Totango
Credentials

Certified where it counts

Formal certification across the platforms I build on — not just hands-on familiarity.

Salesforce
Platform App Builder
Salesforce · Issued Mar 2024
Salesforce
Agentforce Sales Consultant
Salesforce · Issued May 2023
Salesforce
Certified Business Analyst
Salesforce · Issued May 2025
Salesforce
Platform Foundations
Salesforce · Issued May 2023
Salesforce
Marketing Cloud Engagement Specialist
Salesforce · Issued Aug 2023
Salesforce
Platform Administrator
Salesforce · Issued Aug 2020
Gong
Launch Certified
Gong · Issued Jul 2026
Zuora
Billing Certified Solution Architect
Zuora
DealHub
Certified Admin — Essential
DealHub Academy · Issued Dec 2025
Start Here

Ready to see this applied to your stack?

Free 1-week systems audit — no pitch, no obligation.

Book Free Audit →
Guides Fractional RevOps UK AI Transformation for Revenue Teams Headless CRM Architecture Case Study: Process Discovery