Yes. You have full control over model selection — you can replace the default LLM and configure custom models for embeddings, analytics, or response generation.

It depends on the model you use. CodeQA itself does not limit the context window — it’s fully determined by the connected LLM.

Currently, results are based only on the latest version of the repository. Future updates will add support for multiple branches, commit history, and selective inclusion/exclusion of changes.

Yes. CodeQA analyzes the entire codebase but references results at the file level, enriched with explanations that highlight the relevant lines.

CodeQA currently supports incremental updates, meaning only modified files are re-analyzed to keep your index up to date efficiently. You can choose to trigger indexing manually or set it to run automatically on a scheduled cycle, depending on your workflow preferences. Additionally, webhook-based syncing—which will enable automatic updates after each commit—is on our product roadmap.

CodeQA is designed to scale with your projects, from small repositories to enterprise-level monorepos containing millions of lines of code. Our incremental indexing and optimized storage architecture ensure that only changed files are reprocessed, keeping performance high even in very large codebases. If your organization works with particularly complex or distributed repositories, CodeQA can be custom-configured to support advanced scaling and syncing needs.

By indexing your private code repositories, CodeQA builds an internal knowledge base that’s grounded in your actual codebase — not on public or external data. This means every answer and suggestion the AI provides comes directly from your own, securely stored code files, reducing the risk of “hallucinations” or irrelevant responses.

On average, around 10,000 files per day with a single workstation GPU. Performance scales linearly with more GPUs or stronger hardware.

Yes. You can filter results using natural language commands — e.g., limit searches to specific projects, filenames, or extensions, or exclude certain files. Advanced prioritization (like custom weights) is not yet available.

CodeQA delivers answers within a few seconds. By first analyzing your question and conversation history, it ensures responses are both quick and accurate. Files are included at the end of the reply for easier context.

No — the size of your repository does not impact query response time. Once indexed, CodeQA retrieves and analyzes information from its optimized internal knowledge base, ensuring fast and consistent answers regardless of repository scale. Repository size only affects the initial or incremental indexing time, not how quickly queries are processed afterward. CodeQA is fully optimized for performance and scalability, delivering reliable results even for very large codebases.

Repository indexes are stored in isolated containers with no internet access. All contents, analysis results, and responses are encrypted with AES-256 and stored only in encrypted form.

We use a multitenancy system: you can assign users and repositories to specific groups. Future updates will add syncing with GitHub, Bitbucket, and GitLab permissions, plus more granular access levels.