The BYOK ASO tool: bring your own LLM key, control your AI cost

What BYOK means here, in one paragraph

With a bundled AI subscription, the vendor picks the model and the budget. RespectASO Pro works the other way: you buy a Pro license and bring your own LLM API key. The Pro workflows run on your machine with OpenAI, Anthropic, Google, OpenRouter, or the OpenAI-compatible service you connected, using your key. Your text goes straight to that provider, never through us, and we never see the prompt or the answer.

The three Pro AI workflows that use your key

Pro workflow	What it does	When it uses your key
AI Niche Researcher	Expands a seed keyword into a researched niche with related terms, intents, and country signals.	When you research a niche.
AI Competitor Analyzer	Breaks down a top-ranking competitor's positioning, metadata patterns, and likely keyword strategy.	When you analyze a competitor.
ASO Score Simulator	Evaluates a draft title/subtitle/keyword field bundle against scoring and competitor context.	When you run a simulation.

You only pay your LLM provider for runs you actually trigger. There's no idle subscription cost flowing to a cloud AI service.

And if you'd rather not pay per query at all, there's another tier: Local AI. Run a model on your own Mac with Ollama or LM Studio and the same workflows run on-device, with no per-query cost, no API key, and nothing leaving your machine. BYOK keeps cloud cost in your control; Local AI removes it entirely.

Supported providers

Provider	Where to get a key	What you control
OpenAI	OpenAI platform dashboard	Model choice, spend cap, usage limits
Anthropic	Anthropic Console	Model choice, spend cap, usage limits
Google (Gemini)	Google AI Studio	Model choice, spend cap, usage limits
OpenRouter	openrouter.ai/keys	Any model ID from hundreds of models, spend cap, usage limits
Custom endpoint	The service's own dashboard (OpenCode Zen, Groq, and similar)	The service, its API address, the model ID, and whatever limits the service offers

OpenRouter works a little differently: you enter any model ID from its catalog (budget options like DeepSeek included), and RespectASO runs an in-app compatibility test before the model can be used. OpenRouter routes each request onward to the host of the model you select.

The custom endpoint covers everything else that speaks the OpenAI API: enter the service's API address (its Base URL), your key, and a model ID, and the same compatibility test gates it. Requests go straight from your Mac to that service, so pick one you trust.

You can configure one or several keys. RespectASO Pro uses whichever provider and model you select in settings.

Where your text goes

1. You run a Pro workflow (for example the AI Niche Researcher) in the Mac app.
2. The Mac app prepares your text on your Mac, with the API key kept there.
3. It goes from your machine straight to the LLM provider.
4. The provider's answer comes back to your machine.
5. The result shows in the Pro workflow and is saved on your Mac.

RespectASO is not in the network path between your machine and the LLM provider. We never see your prompt, never see the completion, and never carry your AI traffic.

Why BYOK matters

• You control AI cost. You pay your provider for what you run. There's no per-seat AI bundle silently scaling with your usage.
• You control the model. Pick a budget model for casual research, a flagship model for hard-niche analysis. Switch models any time without touching your RespectASO licence.
• You control the data path. Your prompts and completions go between your machine and a provider you trust. They don't pass through RespectASO infrastructure.
• You can use existing AI budget. Many indie developers and teams already have OpenAI, Anthropic, Google, or OpenRouter API keys with monthly spend caps, or a key for another OpenAI-compatible service. RespectASO Pro slots into that existing budget instead of creating a parallel one.
• You can stay free. If you never enter a key, the entire free edition still works: keyword research, scoring, Country Opportunity Finder, multi-keyword search, app rank tracking, and competitor data. The Pro license unlocks AI workflows; BYOK runs them.

Where your keys live

Your API keys are stored in a local settings file on your Mac with restrictive (600) file permissions, readable and writable only by your user account. They are never sent to RespectASO. You can rotate keys, change providers, or remove keys entirely from inside the Pro settings screen.

The full data and network behaviour for both free and Pro editions is documented on the privacy-first ASO tool page, which lists everything that leaves your machine and where it goes.

Honest limitations

• BYOK means you have to set up an API key with one of the supported providers. If you don't already have one, the first run requires the small step of creating a key.
• You're responsible for the provider's spend caps and usage limits. RespectASO Pro doesn't proxy or impose its own AI cost ceiling on top of your provider's controls.
• Pro AI workflows ship only with the native macOS app, not with the Docker free edition. The self-hosted positioning page covers that distinction.
• Provider terms of service apply to your use of the LLM. RespectASO Pro doesn't override them.

FAQ

Is BYOK harder than a bundled AI subscription?

Slightly more setup once. After that, it's strictly better: you see exactly what each workflow costs, you can swap models without touching RespectASO, and you cap your AI spend in your provider dashboard the way you would for any other AI integration.

Can I use a free-tier model?

Yes, where the provider exposes one. RespectASO Pro doesn't restrict your model choice; the provider does. Provider free tiers and rate limits apply.

Does my AI text ever go through RespectASO?

No. The free edition has no AI workflows, and in Pro your text goes from your machine straight to the provider you chose.

How does this fit with the other positioning pages?

The privacy-first page covers data and network behaviour, the self-hosted page covers running the free edition on your own hardware, and the open-source page covers the AGPL-3.0 free edition. This page focuses on the AI cost and provider control story for Pro.