Portfolio Eight systems and programsAnonymizedAuthorship recorded

AI Systems I Designed, Built, and Shipped

A six-system production AI stack for a multi-trade, multi-market field-service business, plus the reliability layer underneath it and the search and lifecycle programs beside it. Sole architect and builder on every one. Client and vendors anonymized throughout.

8systems and programs6in production1builder, AI augmented0agencies or contracted teams

Built by Patrick Joseph Panuncillon, AI Director and Applied AI Systems Architect. Every page states its own status honestly. In-build work is labeled as such.

The builds

Each page covers the problem, the design decisions, and the boundaries.

Described at the capability level. No code, prompts, credentials, schemas, customer data, or client identity appears anywhere in this repository.

Live, in daily production

AI Reply Assistant

A human-in-the-loop reply assistant inside the team's existing messaging tool. Drafts are grounded in the team's own conversation history, a routing and caching layer keeps running cost low, and a nightly learning loop keeps the voice current. A person sends every reply.

Read the case study ›Live

Reply Training Console

A safe training environment that captures a distinct voice for each department without touching production. It ingests conversations, anchors per-department voice, and drafts reply options with full-conversation context.

Read the case study ›Live, in tuning

Call-Time Dispatch Assistant

A booking panel that fills itself during the call. Identifies the caller, matches a qualified technician against the roster, and offers an arrival window from live road distances, so the job is booked while the customer is on the line.

Read the case study ›Live, capturing

Dispatch Understudy

A narrow, opt-in capture that turns an expert dispatcher's judgment into reviewable rules. Scheduling actions only, customer data masked, stays on company systems, proposes rather than acts.

Read the case study ›Live, under everything

Production Reliability Layer

Twenty-eight unattended workflows where health is proven by fresh dated output, an independent watchdog runs twice daily, silence raises an alarm, and recovery backfills the gap rather than skipping it.

Read the case study ›Live, multi-location

Search, Review, and Lifecycle

Answer-engine visibility on a topical authority method, AI-drafted review responses with a required human approval gate across all listing locations, and CRM lifecycle sequences personalized on real service history.

Read the case study ›In build

Campaign Automation

A short form and a brand-locked template let non-technical staff launch campaigns, behind a required human approval gate and a single scheduling path. There is no field that can break the brand.

Read the case study ›In build, in diagnosis

Invoice Automation

Deterministic invoice drafting from a versioned rules registry behind a fail-closed validation gate. The language model names the invoice type and words the customer lines, and never touches a total, a tax decision, or an account.

Read the case study ›

How to read this portfolio

Three things worth knowing before you start.

The interesting part is the boundaries

Not that AI systems were built, but where the limits were drawn: a model that never touches a total, a capture tool that cannot become surveillance, a form field removed rather than validated, a watchdog that does not trust a job to report on itself.

Status labels are accurate

Live, in tuning, in build, and scoped mean different things and are used consistently. Nothing in build is dressed up as shipped, because the labels are the easiest thing in the world to check in a walkthrough. Every status here is stated as of September 2026.

Authorship is claimed. Ownership is not.

Systems built under an employment or client engagement belong to that employer or client. This repository records who built them and expressly disclaims any claim of rights in them. The full disclosure standard, the build ledger, and four tiers of independent verification are set out in Authorship, provenance, and verification.

About

Patrick Joseph Panuncillon, Davao City, Philippines.

AI Director and applied AI systems architect. Nineteen years in operations before AI, which is why this work starts with the process logic rather than with the model. Available for AI Director and Head of AI roles, and for fractional Chief AI Officer and CTO engagements.

Contact consultwithpatrick@gmail.com or LinkedIn.

Confidentiality and intellectual property

This is an anonymized portfolio write-up. It names no client or employer and discloses no source code, prompts, credentials, endpoints, schemas, customer data, or exact commercial terms. Third-party vendors in the client's operational stack are described by category rather than by name. Figures are rounded or banded.

Intellectual property in every system described here belongs to the organization that commissioned it. Nothing on this page is a claim of ownership, a transfer of rights, or an offer to license, sell, or reproduce any system. This page records authorship and build credit only, and it is published in a personal capacity, not on behalf of any employer or client. See Authorship and verification for the full disclosure standard.

If any part of this page is inaccurate, or a rights holder considers it to exceed that standard, contact consultwithpatrick@gmail.com and it will be corrected or removed promptly.