One conversation · supported tools when needed

Xroga AI Chat

Research, analyze and act from one conversation.

Ask naturally. Xroga decides whether the answer needs live public evidence, an uploaded document, visual analysis, or a handoff to the repository-aware coding workflow—and reports when a tool is unavailable.

Research thread
Search the latest official Next.js documentation for this issue.
Current webRetrieving public sourcesRunning
Evidence first.

Xroga keeps source titles and URLs attached to the answer, and treats retrieved pages as untrusted content rather than hidden instructions.

1Official documentationSource retained

Automatic routing

You describe the outcome. Xroga selects the smallest capable path.

Understand intentSelect supported toolRun and inspectAnswer with evidence

Current product capability

Useful tools, with honest boundaries.

Capability depends on configured providers, your plan, and any account or project authorization required for the request.

Current web

Search when freshness matters.

Xroga can route current-information questions to Parallel, retain useful source URLs, and keep retrieved text separate from trusted instructions.

Working

Research

Compare evidence, then answer.

Research requests can inspect several public sources and synthesize the useful parts. Deeper autonomous, multi-hop research remains a developing capability.

Working · deeper mode partial

Documents

Read the material you provide.

Upload supported PDF, DOCX, text, Markdown, code, or image files for extraction and analysis. Large inputs are bounded before model use.

Working

Vision

Understand screenshots and images.

Ask what is visible, compare a reference with a product screen, or identify a UI problem. A connected project is required before Xroga can implement a code fix.

Working · code handoff contextual

X / Twitter evidence

X research stays X-only.

When the question specifically depends on public X posts, Xroga uses xAI's X Search. Non-X citations are rejected, and an uncited synthesis is not accepted as evidence.

Project collaboration

From analysis to a focused code task.

When a request requires implementation, Xroga transfers it to the repository-aware coding workflow. Project permissions, diffs, checks, and provider-backed publishing remain separate from ordinary Chat.

Explore the AI Coding Agent

Real requests

Ask in plain language.

These examples map to current conversation, research, document, vision, and repository-handoff paths.

  • Search the latest official documentation for this deployment error.
  • Research current authentication options for this stack and compare the trade-offs.
  • Read this specification and identify missing implementation requirements.
  • Analyze this mobile screenshot and explain what is broken.
  • Find the current official API for this service and map it to my project.
  • Search recent public X discussion about this developer tool.

Choose the right surface

AI Chat and AI Coding Agent solve different jobs.

Xroga AI Chat

Questions, current public research, documents, screenshots, API discovery, and cross-tool analysis.

Xroga AI Coding Agent

Repository understanding, implementation, GitHub changes, checks, repair, and deployment workflows.

Security and truth

Tools stay server-side. Failures stay visible.

Provider keys are not sent to the browser. Xroga does not claim it searched, changed a repository, tested a site, generated an asset, or deployed a product without corresponding runtime evidence. General-purpose Chat browser automation and image generation are not currently marketed as working capabilities.

Read the security boundaries

Start with the question

One conversation. The right supported path.

Get started — it's free