Founders
Solo founders
Real product ideas, early customers, rough prototypes, or a workflow that needs to become a product surface.
I’m Cong Fan. I build practical AI products and workflows from Brisbane.
I help founders and small teams turn messy AI product ideas, internal workflows, repo problems, and manual business processes into working prototypes, automation, and shipped product improvements.
Australia-wide and remote global.
中文也可以。
AI prototypes, internal tools, billing flows, n8n automation, and repo debugging.
Who this is for
People With a Real Workflow, Prototype, or Product Problem.
I am not trying to teach generic prompting. I work best when there is a concrete product surface, internal workflow, codebase, customer problem, or launch pressure that needs practical implementation.
Operators
Small businesses
Internal tools, AI workflows, admin automation, content systems, and practical integrations that reduce repeated work.
Product teams
Startup product teams
Fast prototypes, unstable AI features, billing validation, model workflow fixes, or production-ready improvements.
Australia-wide
Brisbane and remote
Brisbane-based, Australia-wide, and remote global for focused product implementation work.
Selected work
Projects and Work Surfaces I Can Point To.
I avoid invented growth metrics. The proof here is practical: a real problem, what shipped, and the public or inspectable surface you can use to judge the work.
AI video / image product
ReelRush
My product build surface for AI image/video generation. Work includes generation flows, pricing and credits, public pages, deployment checks, and ongoing product operations.
Problem: AI media products need more than a model call: inputs, generation flow, billing, retries, and public surfaces all have to line up.
Shipped: Generation UI, pricing and credits work, public pages, deployment checks, and operating fixes.
Evidence: Public product surface you can open.
reelrush.com.au
Small business AI workflow
Ad Maker
A small-business ad generation workflow inside ReelRush. The goal is practical output creation, not a generic chatbot or prompt demo.
Problem: Small businesses need usable ad output, not another prompt playground.
Shipped: A focused workflow that turns business context into reviewable ad creative.
Evidence: Built inside the ReelRush product surface.
reelrush.com.au
Automation / metadata
n8n + Supabase
Internal automation around company work emails, project management, metadata, and structured operational data. Built to reduce repeated manual coordination.
Problem: Repeated emails, project updates, and metadata work create hidden admin drag.
Shipped: Structured automation that connects AI extraction, reviewable metadata, and operational data.
Evidence: Internal workflow pattern, described publicly without private company data.
B2B website / SEO
AS Component
B2B industrial website work across product pages, SEO structure, and sales-support pages for a more discoverable product catalog.
Problem: Industrial buyers need clearer product pages and search-friendly category structure.
Shipped: Product-page structure, SEO-aware implementation, and sales-support surfaces.
Evidence: Public B2B website surface you can inspect.
ascomponent.com.au
How I use Codex and Claude Code to ship working prototypes
A practical loop for founder builds: brief, inspect, constrain, delegate, verify, and leave behind something usable.
What I look for before taking on an AI workflow
The owner, user, repeated action, existing toolchain, failure mode, and definition of done matter more than the model name.
When a repo problem is really a product problem
Runtime bugs often expose unclear scope, pricing truth, missing evidence, or a workflow that was never made inspectable.
Why I keep AI builds small
A narrow surface is easier to inspect, ship, and learn from. Most useful AI systems start as one reliable loop.
Product before tools
Name the user, workflow, business outcome, and definition of done before choosing the AI stack.
Evidence-first debugging
Start from exact errors, logs, screenshots, rows, browser state, and deployed behavior before proposing fixes.
Repo and runtime verification
Inspect the actual codebase, live route, function, database, or local process instead of trusting assumptions.
Narrow scoped delivery
Keep the build small enough to review and ship: one surface, workflow, integration, or agent lane at a time.
Documentation and handoff
Leave behind working files, verification notes, known limits, setup details, and next actions.
Limited capacity
Currently taking 1 build sprint and 2 audits per month so each engagement has enough attention.
Capability stack
How I Usually Help.
The useful part is usually not the tool choice. It is turning a vague or unstable workflow into something a customer, operator, or developer can actually use and verify.
Agentic coding
Using Codex and Claude Code against the real repo: inspect, change, test, verify, and leave a clear handoff.
CodexClaude Code
AI prototyping
Turning rough product ideas into working screens, model calls, data flow, and testable user paths.
AI productMVP
Supabase architecture
Data models, auth-aware workflows, metadata pipelines, access rules, and production checks.
SupabasePostgres
Stripe billing
Checkout, credits, webhooks, billing validation, test-mode paths, and user-visible pricing truth.
StripeCredits
AI image/video workflows
Product flows around image/video generation: inputs, model routing, output review, credits, and retry paths.
GPT ImageSeedance
n8n automation
Automation workflows for company emails, project management, metadata, and internal operations.
n8nOps
Repo debugging
Tracing failures through code, logs, browser state, data rows, deployments, and exact runtime evidence.
DebugVerify
Product shipping
Deployed systems, billing validation, working workflows, docs, and review-ready improvements.
ShipHandoff
Best Fit
A founder with a real product idea and enough context to make decisions.
A small business with a repeated workflow that can become an internal tool.
A startup team needing a fast prototype, repo rescue, or AI implementation lane.
An agency that needs technical AI implementation support for client delivery.
Not a Fit
Pure prompt-writing lessons.
Low-budget outsourcing without product ownership.
Generic chatbot projects with no product strategy.
Ideas with no owner, customer, workflow, or business outcome.
Projects needing a large agency or 24/7 support.
中文也可以。你可以直接用中文说明你想做什么、现在卡在哪里、已有代码/流程是什么、预算和时间线大概是多少。
Services
Focused Help When the Work Needs to Move.
If the notes and work above match what you are trying to do, these are the shapes I usually use. The outcome should be a shipped artifact, a verified workflow, or a clear handoff, not a strategy deck.
Build sprint
10-Day AI Build Sprint
Turn one AI product idea, automation workflow, or agentic coding task into a working prototype or production-ready improvement.
AUD $3,500-$8,000Limited to 1 build sprint per month
Audit
AI Product / Workflow Audit
Clarify the user, workflow, build path, tools, risks, and next practical implementation step.
AUD $690-$1,5002 audit slots per month
Repo debugging
Agentic Coding Sprint
Use coding agents with real repo inspection, debugging, tests, browser checks, and review-ready handoff.
AUD $2,500-$6,000Scoped around one repo or workflow
Ongoing support
Monthly AI Ops Retainer
Maintain automations, prompts, billing flows, internal tools, model workflows, and operating docs as the product changes.
AUD $2,000-$5,000/moFor teams already shipping
| Engagement | Best fit | AUD |
|---|---|---|
| AI Product / Workflow Audit | When the build path, users, workflow, tools, or risk profile needs to be clarified first. | AUD $690-$1,500 |
| 10-Day AI Product / Agentic Coding Sprint | When one AI product surface, workflow, or prototype needs to move fast. | AUD $3,500-$8,000 |
| Agentic Coding / Repo Debug Sprint | When the repo, billing flow, model workflow, route, or runtime behavior needs focused rescue. | AUD $2,500-$6,000 |
| Monthly Retainer | When an active product needs ongoing AI ops, automation, prompts, billing, and workflow maintenance. | AUD $2,000-$5,000/mo |
Do you work only with Brisbane businesses?
No. I am Brisbane-based and work Australia-wide. Remote global work is possible for focused scopes, and English or Chinese is fine.
Should this start as an audit or a build sprint?
Use an audit when the workflow, data, repo state, or commercial path is unclear. Use a sprint when there is already a clear owner, user, workflow, budget, and definition of done.
What AI workflow is a good fit?
The strongest projects have a repeated action, current materials, a human review point, and a first useful output that can be tested against real inputs.
Is Chat With Me the same as sending a brief?
No. Chat With Me helps shape the brief and can copy a draft into the form. The project brief form is the formal submission path.