Replokit
for CTOs, security, and compliance teams

Which AI is your code actually calling?

Every team has added an LLM somewhere. Replokit reads your repositories and lists every AI provider, model, and AI tool in use — by repository — so you can answer the board, the security questionnaire, and the finance team in one place.

Your AI footprint, from the code

No surveys, no spreadsheets that go stale. What the code declares is what's running.

Every provider

OpenAI, Anthropic, Google, Meta, Mistral, Cohere, xAI, plus vector databases and AI frameworks.

Every model

The exact model identifiers each repository calls — gpt-4o, claude-*, gemini-* and more — with a repo count for each.

Change alerts

Paid plans rescan daily and record when a new model or AI service first shows up.

Spend estimate

A directional monthly range for each paid AI provider, alongside the rest of your infrastructure.

Use it for

  1. 01

    Security questionnaires

    Customers ask which AI vendors process their data. Answer with an inventory read from the code, not from memory.

  2. 02

    Vendor & compliance reviews

    Keep a current list of AI sub-processors for SOC 2, ISO 27001, and AI policy reviews.

  3. 03

    Cost and consolidation

    Spot teams paying for three providers and five models doing the same job, then standardise.

Questions

Which AI providers and models can it find?

Models from OpenAI (GPT, o-series, embeddings), Anthropic (Claude), Google (Gemini), Meta (Llama), Mistral, Cohere, and xAI, plus AI services and tooling such as Hugging Face, Pinecone, and LangChain declared in dependencies and configuration.

How does it detect models without reading our prompts or data?

Deterministic parsing of dependency manifests, environment-variable names, and configuration files for provider SDKs and model identifiers. It never reads production data, and code is never sent to an AI model.

Will we know when someone starts using a new model?

Yes, on paid plans. Daily rescans record a change event when a new AI model or service first appears in your codebase.

Does it show what we spend on AI?

It gives a directional monthly estimate for each paid provider it finds, so you can see where spend likely sits. It doesn't read your invoices — for exact figures, check each provider's billing.

Know your AI footprint today.

Connect GitHub read-only and see every AI vendor and model in minutes. Free for up to 10 repos.

Scan your stack — free