FAQ

Plain answers to the questions that come up most.

If your question isn't here, write to dev@golproductions.com. We respond within 24 hours.

General

About GOL.

What is GOL Productions?

GOL Productions builds deterministic infrastructure for AI-assisted engineering. Our products sit at the points where AI workflows fail — hallucinated code, broken commits, production regressions — and verify against reality before letting failure ship.

What languages does the engine support?

JavaScript, TypeScript, Python, Go, Rust, Java, C/C++, PHP, Ruby, C#, Shell, and most others with logical structure. The engine analyzes code structure, not just syntax.

Who is this for?

Engineering teams shipping AI-assisted code into production. The IDE Package suits anyone whose AI sometimes produces hallucinated functions. The Debugger suits anyone responsible for a production system that can't go down quietly.

How is this different from ChatGPT or Copilot?

ChatGPT and Copilot are general-purpose assistants. GOL is purpose-built infrastructure. Our products run as deterministic gates with hardcoded analysis rules, deterministic preprocessing, and strict output schemas. We don't suggest. We state.

When production breaks

Speed and reliability.

How long does a rescue actually take?

The Debugger averages 30 seconds from input to verified patch. Typical range is 4 to 18 minutes for full webhook-triggered rescues including PR creation and test verification.

What happens if the engine is overloaded?

The engine retries with exponential backoff (2s → 4s → 8s). If compute is still unavailable, the request enters a queued state and you're notified when capacity frees up. Requests are never silently dropped.

What if the trace is inconclusive?

The engine produces a gap analysis instead of a guess. You receive a precise report of what's missing — incomplete telemetry, mismatched symbol maps, insufficient context — pointing directly at the severed link in the chain of evidence.

Pricing & billing

What you actually pay.

How much does it cost?

The Debugger is $1.20 USD per successful rescue. The IDE Package is $0.01 AUD per commit. No tiers, no daily limits, no recurring subscriptions. You top up your wallet with any amount, and credits are deducted as you use the engine.

What counts as one call?

For the IDE Package, one call equals one commit that triggers any layer. Whether one layer fires or all four, the cost is $0.01 AUD. For the Debugger, one rescue equals one triage request submitted to the engine.

Do credits expire?

No. Credits sit in your wallet until you spend them. There's no monthly reset, no expiration window, no dormant-account policy.

Is there a free trial?

You get 100 free calls on signup for the IDE Package. The Debugger doesn't have a free trial, but the demo on the landing page walks through a real triage so you can see the engine work before topping up.

Can I cancel anytime?

There's nothing to cancel. No subscriptions exist. Stop using the products and you stop paying. Existing credits stay in your wallet.

Security & privacy

What we do with your code.

Is my code stored or logged?

No. We operate a zero-retention policy. Code is processed in ephemeral memory on Cloudflare edge nodes and is never written to disk, database, or any logging system. Once the response returns, the input is gone.

Where does processing happen?

Everything runs on Cloudflare's global edge network. Your request hits the nearest point of presence, is processed in an isolated V8 worker, and the response returns at edge speed. There are no centralized servers.

What is an API key?

It's how you authenticate programmatic access. Email login is for the dashboard. API keys are for webhook integrations and the local IDE Package. Both methods are server-side verified against your wallet balance.

Integration

Wiring it into your stack.

Can I integrate this with my monitoring stack?

Yes. The /api/ingest/error webhook accepts payloads from Sentry, Datadog, PagerDuty, custom scripts, or curl. Authenticate with your API key via the Authorization header. The engine returns a full diagnostic in JSON.

What format should I send my code in?

Paste the full source of the failing file plus relevant stack traces or error messages. The engine requires actual source — if you reference a filename without including the code, the engine returns INSUFFICIENT_CONTEXT rather than hallucinating a patch.

The IDE Package

The local commit-time gate.

What is the IDE Package?

A deterministic, local pre-commit hook that uses our AST parsing engine to block AI-hallucinated code from entering your repo. It installs into your repository and runs entirely on your local machine.

How is it billed?

Pay-per-call at $0.01 AUD per commit. When you sign up, you receive a unique key and 100 free calls. The local hook pings our edge network at commit time to verify your balance. If your balance hits zero, the hook blocks commits until you top up.

Does the IDE Package send my source code to your servers?

No. AST parsing, logic mapping, and hallucination detection happen strictly on your local machine. The only network request is a lightweight balance verification ping. Your source code never leaves your device.

What's the refund policy?

Top-ups themselves are non-refundable, but unused credits never expire. You may stop using the product at any time. Except as required by law (including the Australian Consumer Law for major technical failures), credits remain in your account indefinitely.

Support

Getting help.

How do I get help?

Email dev@golproductions.com. We respond within 24 hours, including for enterprise enquiries.

What's the uptime guarantee?

The platform runs on Cloudflare Workers, which has a 99.99% uptime baseline. The AI compute layer has its own SLA. Our system handles transient failures via automatic retry and queueing.

Question that's not here?

Email dev@golproductions.com. We answer within 24 hours.