Applied AI Build Studio / Greg Staunton

Applied AI systems for enterprise marketing workflows.

pl8ypus is Greg Staunton's applied AI build studio - focused on production-shaped systems for marketing automation, translation, customer operations, and workflow control.

View build evidence

Greg's take

Most AI projects do not fail at the model call. They fail in the workflow around it - bad handoffs, unclear review points, dirty data, weak evidence, and nobody owning what happens when the output is wrong.

pl8ypus is where I build the control layer around that problem. Translation, form protection, client operations, audience selection, and build orchestration are different surfaces of the same idea.

The model matters. The system around it matters more.

Background and capability

Enterprise marketing automation

20 years building and operating enterprise marketing stacks in B2B and global enterprise environments.

Oracle Eloqua / Salesforce workflows

Deep platform experience in Eloqua campaign and data flows, SFDC integration, consent management, and reporting.

Cloudflare Workers, D1, R2 and Access

Production builds on Cloudflare infrastructure: Workers, D1 databases, R2 storage, and Access/JWT identity layers.

AI-assisted architecture and implementation

End-to-end delivery with AI assistance at architecture, implementation, QA, and documentation layers.

Human-in-the-loop review systems

Review queues, accept/reject controls, audit trails, and manual override paths built into every AI output layer.

Production-minded QA and deployment

QA gates, acceptance reports, controlled slice-by-slice delivery, and hardening passes before deployment.

Applied AI Systems

Production-shaped builds.

View all systems

Build 01 Active

Translation AI

Problem: Marketing translation done manually, inconsistently, with no protected-term enforcement or review trail.

AI-assisted translation workflow for enterprise marketing assets. Provider-agnostic orchestration layer with protected terminology, structured payloads, QA gates, review status, and human approval before output is treated as final.

Anthropic API Protected terms QA gate Human review

Proves repeatable AI workflow design with control layer and evidence capture - not just a prompt.

View system →

Build 02 - Featured Migration in progress

Client Portal

Problem: Client support workflows live across email, files, invoices, status updates, and memory.

A portal sounds simple until it has to behave like a real support system. Requests, updates, documents, billing visibility, attachments, admin controls, audit trails, identity, and multiple users attached to the same client account.

That is where the toy version breaks. This build is about earning production one layer at a time: prototype the workflow, prove the client model, move the writes to D1, add identity, harden the access path, and keep every slice testable.

Cloudflare D1 Workers API Access/JWT Multi-client Request lifecycle Audit logging

Proves Greg can build the operational layer around client work, not just AI demos. This is the kind of system that forward deployed engineers design inside real customer environments.

View full case study →

Build 03 Enterprise used

Marketing Form Protection

Problem: Enterprise form spam pollutes Eloqua data, damages campaign reporting, and wastes sales time across marketing automation workflows.

Cloudflare Turnstile and Eloqua protection layer with exception queue, direct Eloqua forwarding, accept/reject handling, and multi-form architecture. Applied to a real enterprise Eloqua marketing environment, with client details anonymised.

Cloudflare Turnstile Workers KV storage Eloqua API

Enterprise-used workflow. Proves Greg can build applied AI-adjacent protection around real marketing operations, not just isolated demos.

View system →

Build 04 Active

Audience Finder AI

Problem: Audience targeting decisions made without structured evidence, scoring criteria, or review controls.

AI-supported audience and targeting workflow with scoring, evidence trails, and manual accept/reject controls. Includes queue management, export readiness, Cloudflare Worker backend, and KV memory layer.

Cloudflare Worker KV memory Accept/reject Scoring

Demonstrates human-in-the-loop design with backend infrastructure, not just a frontend demo.

View system →

Build 05 Operational

Tanya Build Cockpit

Problem: AI-assisted build sessions drift without structure, memory, or consistent QA checkpoints.

AI-assisted build orchestration layer using structured prompts, project memory, QA gates, implementation packets, and agent role separation. The operational scaffolding behind every pl8ypus build.

Structured prompts Project memory QA gates Agent roles

Shows the build discipline and AI collaboration layer that makes the other systems production-shaped rather than experimental.

View system →

Featured case study

Client Portal: from lightweight support idea to production-shaped Cloudflare system.

This is not a UI prototype dressed up as infrastructure. It is a business system being progressively migrated from localStorage to a Cloudflare-backed architecture - with documented slice boundaries, QA acceptance at each layer, and clear distinction between what is prototype, migration-ready, and production-hardened.

Identity

Cloudflare Access / JWT

Secure client login without managing auth infrastructure from scratch.

Data layer

Cloudflare D1 + Workers API

Structured D1 database with Workers API routes added slice by slice.

Client model

Multi-client shared instance

Everyone in the same client org sees shared requests, billing, docs, and attachments.

Build method

Controlled slice-by-slice

Each migration slice has acceptance criteria, QA gate, and deployment verification.

Request creation, comments, status changes toward D1/API

Documentation metadata and body endpoints

Hours logging and billing visibility

Admin controls and audit/activity logging

Production hardening pass on portal.greg-staunton.com

R2 storage planned for file attachments

Read full case study

Operating model

How Greg builds.

A repeatable pattern for moving from workflow problem to deployed, operator-ready system - without skipping QA, documentation, or production-readiness checks.

01

Discovery and workflow mapping

Understand the existing workflow, the people in it, and where AI changes the decision boundary.

02

System architecture

Design the smallest complete system: inputs, outputs, review layer, data model, and deployment shape.

03

AI-assisted implementation

Build with structured prompts, project memory, implementation packets, and clear agent role separation.

04

Human review and QA

QA gates, acceptance criteria, test auth flows, and edge case verification before each slice is considered done.

05

Controlled deployment

Deploy slice by slice with verification at each step. No big-bang releases. Rollback path is always clear.

06

Evidence capture

Document what was built, what decisions were made, and what the system does at a level operators can trust.

07

Repeatable patterns

Every build produces reusable scaffolding: prompt patterns, QA checklists, architecture templates, and deployment scripts. The next build starts from a higher floor.

See the full build method

Marketing systems expertise

The domain knowledge underneath the AI layer.

Applied AI systems are only as good as the underlying domain knowledge. The marketing operations background shapes every system architecture decision.

Oracle Eloqua

Campaign setup, data processing steps, program canvas, form processing, API integration, and platform governance.

Salesforce campaign and data flows

Lead routing, campaign member tracking, CRM sync, data quality rules, and bidirectional Eloqua-SFDC integrations.

Consent and GDPR workflows

Subscription management, consent capture, preference centers, audit-ready data handling, and compliance workflows.

Email and landing page operations

Template production, asset governance, dynamic content, A/B testing, and campaign QA before deployment.

Form handling and spam protection

Form validation, spam review, Turnstile integration, exception routing, and CRM quality protection.

Translation workflows

Localisation planning, asset versioning, protected-term management, multi-language review, and platform-ready output.

Campaign execution support

Campaign brief to live management, stakeholder coordination, asset sign-off, scheduling, and post-campaign analysis.

Reporting and operational dashboards

Performance tracking, attribution models, stakeholder reporting, and dashboard design for non-technical operators.

Value positioning

Built for the gap between AI models and real business systems.

Most AI value is created after the model call: workflow design, data shaping, review controls, user adoption, deployment safety, and operational reliability. That is the layer pl8ypus is built around.

The systems in this portfolio exist because understanding how a marketing team actually works - the data flows, the approval steps, the form constraints, the platform limitations - is just as important as knowing how to call an API.

Customer workflow integration

Built to fit inside existing marketing operations - not to replace them.

Production applications

Not demos. Systems with data models, review layers, error paths, and deployment evidence.

Repeatable patterns

Build scaffolding, QA templates, and architecture patterns that transfer to new customer contexts.

Systems non-technical operators can use

Marketing teams drive the review queue. Operators control the output. AI does the heavy lift in the middle.

Contact

Discuss a build, project, or collaboration.

For applied AI project enquiries, marketing systems support, speaking requests, and collaboration around practical AI systems for enterprise environments.

About Greg

Applied AI project

Marketing systems support

Speaking or demo session

Collaboration or general enquiry