Portfolio
2026 projects.
Each project solves a different problem for a different industry. The architecture, the lessons, and the thinking behind them inform everything we build for clients.
ConfideAI
LiveSecure AI clinical documentation for mental health professionals. 20+ evidence-based templates across SOAP, BIRP, and DAP formats — treatment plans, intake assessments, progress notes, correspondence, and psychoeducation materials. All AI processing runs inside Intel TDX hardware-secured enclaves. ConfideAI cannot access user data, even if compelled — cryptographic privacy, not policy-based. Templates interconnect so output from each feeds into the next, saving hours across an episode of care.
MycenAI
LiveAI-powered clinical documentation for psychedelic-assisted therapy practitioners. Structured templates for preparation, ceremony support, and integration sessions, adapted to the practitioner's therapeutic orientation. Built on ConfideAI's platform with hardware-secured confidential computing.
DreamCopy
LiveAI-powered property marketing platform. Upload photos and an address, receive a complete 7-channel campaign pack — SMS alerts, portal listing, Instagram, TikTok/Reels script, Facebook post, email campaign, and a print-ready brochure with specifications. AI analyses every photo for materials, light, spatial flow, and architectural features, then combines that with location intelligence to produce copy agents publish without editing.
CoachIQ
LiveAI-powered coaching intelligence platform for executive coaches. Captures one-to-one sessions and transforms them into developmental intelligence — pre-session briefs, commitment tracking, score trends, goal progress, team insights, and quarterly reviews. Longitudinal pattern recognition identifies developmental arcs and decision-avoidance across months of coaching data. On-device de-identification ensures real names never reach the AI; client-side encryption ensures the platform never sees plaintext data.
AML/CTF Compliance Program
DeliveredComplete AUSTRAC-compliant AML/CTF framework for a Perth accounting practice ahead of the July 2026 Tranche 2 deadline. Full traceability matrix mapping every legislative obligation to specific policy sections, risk assessment separating inherent from residual risk across ML/TF/PF dimensions, customer due diligence forms, and a 13-week implementation checklist. Designed so an independent evaluator in 2029 can trace every obligation back to the policy that satisfies it.
Thinking
What we have learned building these
Every production system teaches you something a demo never will. These are the lessons — written for business owners, not engineers.
Why AI Safety Features Are Load-Bearing Architecture
9 minWe Built an AI Invoice Verifier. Here's Where It Hits a Wall.
8 minHow to Build an AI Chatbot That Doesn't Lie to Your Customers
5 minWhat Production AI Teaches You That Demos Never Will
6 minHow We Turned Generic AI Into a Specialist
5 minWhy I Don't Use n8n (And What I Do Instead)
7 minView all articles →
Background
Paul Morgan
Principal, Perth AI Consulting
BA, BPsych, PGDipPsych
Background in organisational psychology, business intelligence, modelling, evaluation, and market research. 20 years designing reporting and evaluation frameworks across WA government and national banking — including work tabled in Parliament and evaluation systems built for the Kennedy Royal Commission, WA Police.
Now building confidential AI systems for industries where privacy isn't optional — healthcare, counselling, coaching, legal, and compliance.