A repeatable foundation.
Project structure, clear agent instructions, dependencies, setup and a reliable route from source files to deployment.
AI Build Review & Repair
You’ve built something.
Make it dependable.
The screen looks right. Now understand what’s underneath: access, data, deployment, recovery and the cost of keeping it running.
Starting fresh instead? Explore Business OS Core with onboarding.
A clearer path from prototype to dependable tool
Keep what works.
Repair what’s underneath.
The Gallery AIIllustrative report
What needs attention?
ObservedNo lockfile found
Review neededConfirm who can read each record
Not assessedLive sign-in and recovery
Fictional example. No project has been assessed by this illustration.
The Gallery AIIllustrative report
Your approved repair plan
SelectedAdd repeatable dependency setup
SelectedTest two users with separate records
Held for scopeReplace the complete database
Fictional example. No project has been assessed by this illustration.
The Gallery AIIllustrative report
A useful handover
RecordChanges and checks completed
RecheckWhat passed and what failed
Carry forwardOpen risks, costs and ownership
Fictional example. No project has been assessed by this illustration.
See the gaps, with evidence.
Start with a read-only local scan and a guided review. Separate observed findings from questions that need live testing.
Fix only what you approve.
Choose numbered findings. Protect the starting point, review the proposed changes and keep scope under control.
Prove what changed.
Repeat the relevant checks and retain evidence. Record what remains unverified and who owns the next step.
Illustrative walkthrough. The toolkit runs locally; this page does not inspect your project.
Looks good is a start.
Works properly is the goal.
A useful review looks beyond the interface. It shows what was checked, what needs a decision and what still needs real-world testing.
The right access.
Review credentials, sign-in, roles and database permissions. Test that one person cannot reach another person’s private records.
A system you can keep.
Ownership, running costs, recovery, known limits and operating instructions. Know what to do when something fails.
Inside the DIY kit
More than a prompt.
A process you can follow.
A local scanner gives you a starting inventory. Guided prompts turn evidence into decisions, selected repairs and checks. Your files stay in your environment; sharing them with an AI service follows your own account settings.
Read-only project scan
A Python tool looks for common foundation signals and gaps. It doesn’t install packages, run your app, change the project or send files to us.
Audit, repair and retest
Separate review, repair and retest workflows for Claude Code or Codex. Numbered findings, explicit repair choices, verification evidence and a record of what’s still unknown.
Costs and handover
A software-cost worksheet, access inventory, recovery checklist and operating guide help you keep working after the first repair.
Useful preparation for help
A short guided-session intake and repair brief let us review the situation before your booked time starts.
What the scanner can and cannot tell you
It checks source-file indicators and configuration hygiene. A detected authentication library is not proof that permissions work. Missing conventional files can be intentional. The report distinguishes findings that need review from observed facts.
It does not execute project scripts, scan a live site, test your database or certify security. Recognised credential-like filenames and all symlinks are excluded; suspected secret values are withheld. Authorised runtime tests and human review are still needed before relying on a production app.
Python 3.9 or newer is required for the local scanner. Claude Code or Codex with local project access is required for the guided repair workflows. Your coding plan and any API usage are separate.
The help you need.
The scope you choose.
DIY for capable builders. Guided support when you want a second pair of eyes. Scoped implementation when you want us to carry out the repairs.
Self-led
DIY Build Foundations Kit
$273.90
AUD including GST · $249 excluding GST
Scanner, guides, repair/retest prompts, evidence templates and running-cost worksheets.
- One-off kit licence
- Works with your own coding assistant
- Personal review and repairs not included
Early release. Tested on invented fixtures; independent customer outcomes still need a pilot. Register interest for access before you buy.
Register interestDone with you
Guided Build Review & Setup
$2,100
AUD excluding GST · one scoped project
Pre-check review, a 90-minute working session and one follow-up check within the agreed scope.
- Complete the intake first
- Prioritised findings and selected small fixes
- Anything larger gets a separate quote
Done for you
Scoped Repairs
From $5,000
AUD excluding GST · quoted after review
We implement the approved repair plan, test the relevant behaviour and hand over the changes.
- Named findings and acceptance checks
- Changes kept reviewable
- Open issues and recovery documented
No unlimited repairs or guaranteed security clearance. We confirm project suitability before booking paid work. Larger rebuilds, unsupported technologies and specialist security assessments are separately scoped. Software and AI costs are yours unless included in writing.
DIY Build Foundations Kit · early release
Register interest.
Leave your details and Thomas will be in touch about access before you buy.
Your software bill, separately
Know what it costs
to keep it running.
Your implementation fee pays for our work. Hosting, data, email and AI usage belong in your own operating budget. Start with this example, then adjust it.
Example: Netlify Pro + one Supabase Pro project + Resend Free starts at US$45/month before extras. The default A$1.50 exchange rate is illustrative, not a live quote.
Plan limits, repair costs and what can change the bill
Paid plan prices are starting costs; usage can increase the bill.
Hosting: Netlify Free has 300 monthly credits; Personal has 1,000 and Pro 3,000. Production deploys, traffic and compute consume credits. Exhausting a team’s credits can pause its projects; paid top-ups are optional. Vercel Hobby is for non-commercial personal use, so it isn’t the budget for a business deployment. Vercel Pro includes one deploying seat and usage credit; extra seats and usage cost more.
Database and email: Supabase Pro starts at US$25 for an organisation with a first Micro project; additional projects, compute and add-ons change the total. Its spend cap does not cap every charge. Free projects can pause after inactivity. Resend Free allows 3,000 emails/month with a 100/day limit; Pro starts at US$20 for 50,000, with overages extra.
Running the repair: You use your own Claude Code or Codex account. Your existing allowance may cover it, but a large project or repeated testing may use more. Set a budget before running; don’t buy a higher plan just because a kit suggests it. AI used inside a deployed app is billed separately from a coding-assistant subscription. Zero in the planner means you have entered no extra allowance, not unlimited free AI.
Optional tools: Keep subscriptions that still do a useful job. Social publishing and booking may retain per-channel or per-host charges. Storage, monitoring, backups, domain renewal and any specialist help also need a budget. Use the additional-tools and usage fields for these.
Checked 23 September 2026. Verify vendor terms before purchase: Netlify, Vercel, Supabase, Resend, Buffer, Calendly.
A small commitment.
A useful next decision.
Run the pre-check.
Identify the project, owner, hosting and data. Run the local scan and redact the report before sharing.
Choose the repairs.
Agree what matters first. Protect the starting point and approve a bounded set of changes.
Verify and take over.
Check what changed, practise a recovery route and retain your own instructions and cost record.
For the first service release, guided work focuses on React/Next.js projects using conventional Git-based deployment and supported hosting/database services. Other stacks are assessed before a quote.
The right help,
without another dependency.
Use your own accounts. Keep your source and records. Our job is to help you understand and improve the system, so the next step doesn’t always require another call.
How is this different from the free Workspace Health Check?
The free Health Check checks whether an AI workspace can find its information. Build Review & Repair focuses on the application you have built, its technical foundations and the evidence needed for a safe next step.
Do I need to move everything to Netlify?
No. A sensible existing host may already be right. We inspect the actual framework, data and deployment setup before recommending a change. Replacing working infrastructure adds work and risk without automatically improving the result.
Bring what you’ve built.
We’ll work out what needs attention and the smallest useful next step.