About — Alan Hsieh

Areas of work

Where he focuses

Scheduled reporting

A report that runs on a set schedule, in a fixed format, checked against a ruleset before it goes anywhere. Alan designs the AI workflow that produces it, flags anything that needs a human decision, and keeps the team out of the loop until there's something that requires them.

Exception investigation

When a record doesn't match, the question is always: why? Alan builds the AI layer that does the investigation first — pulls the records, surfaces the likely cause, and hands the analyst a starting point instead of a blank screen.

Multi-jurisdiction output

The same underlying data, reformatted to meet different requirements, delivered on different schedules. Alan builds the piece that handles the repetitive variation so the team spends time on the decisions, not the formatting.

Approach

Works inside what already exists

Alan doesn't build new platforms. He finds the parts of a workflow that follow a predictable pattern and figures out exactly where AI can take over — inside whatever the team is already running. The people who need to approve what goes out still approve it. The audit trail still exists. Nothing changes except the amount of time the team spends doing the same thing twice.

Work

A sample of what's been built

These are real tools, built inside existing workflows. Each one started as a manual process.

Performance Report System

Automated client reporting for a multi-DSO healthcare group

A structured report produced on a fixed schedule, formatted to spec, with real data pulled from multiple sources. The AI layer drafts the report, checks it against the rules, and flags anything that needs a human review before it goes to the client.

CS12 — Practice Management Platform

Multi-org client tracking platform with exception flagging

A full platform that tracks performance across multiple client organizations, surfaces which accounts need attention, and logs all activity. Built to give the consulting team a single view across clients without manual aggregation.

RIH — Content Workflow Tool

AI-assisted content production with human review and approval

A tool that generates content drafts in a specific voice, organized by topic category, with a review step before anything gets used. The human decides what goes out — the AI handles the drafting.

Draft reply — Sarah K.

Re: Q2 pipeline review — I've reviewed the notes and flagged two gaps...

Flagged — Michael T.

Meeting brief ready for 2pm. Needs your sign-off before sending.

Summarised — Board update

4 action items extracted. No response needed, logged to Drive.

Draft ready — Vendor follow-up

Contract renewal. Awaiting your review before sending.

PMC — Internal AI Agent

AI agent embedded inside an existing team communication tool

An AI assistant that lives inside Google Chat, handles inbox triage and email drafts, and sends nothing without explicit human approval. Built inside the tool the team was already using — no new platform adopted.

Background

Alan Hsieh

Alan designs AI systems that do real operational work. He has built these for financial professionals, consulting firms, and multi-jurisdiction reporting workflows. His focus is on the decision layer — which tasks AI should handle, which it shouldn't, and how to make sure nothing moves without the right person in the loop.

Focus areas

Making AI work inside existing tools — not replacing them
Automating repetitive reporting and data workflows
Keeping humans in control of what goes out
Building systems that follow rules — compliance, approvals, audit trails

Alan Hsieh works with operations teams on the AI problems that look like people problems.