Software engineer working at a standing desk

Our product

RootCauseAI: AI investigation inside your network

We built RootCauseAI so support teams can investigate incidents with AI, without sending code, logs or data outside their own boundary.

Brisbane office towers at blue hour

The problem

When an incident hits, support engineers jump between tickets, commits, databases and documentation. Cloud AI could help, but many organisations, especially in regulated and resources sectors, cannot send that data to external services.

  • Hours lost switching tools
  • Sensitive code and logs
  • Security rules that block cloud AI

What we built

  • Runs locally

    Self-hosted with local models, so data stays inside your infrastructure.

  • Connects to your tools

    Jira, Redmine, Git, SQL databases and MongoDB.

  • Learns from history

    Indexes documentation and past incidents for context.

  • Reads the code

    Reviews relevant code paths in PHP, Java, C#, ASP.NET and Angular.

  • Suggests root causes

    Shows its evidence so engineers can check the reasoning.

What it taught us

  • Ordinary hardware works

    Useful private AI runs on a standard server without a GPU cluster.

  • Boundaries come first

    Read-only access and logging are designed in from the start.

  • Same pattern, any process

    We apply this design to private AI work for clients.

Questions

What is RootCauseAI?

RootCauseAI is a self-hosted AI investigation product we built and run ourselves. It searches code, logs, tickets and databases inside an organisation's own network to suggest the likely cause of an incident, showing its evidence so engineers can check the reasoning.

Can private AI run without a GPU cluster?

Yes, for many workloads. RootCauseAI runs useful local models on a standard server, which showed us that private AI for search, investigation and document work does not always need specialised hardware. Larger models and heavy volumes still benefit from GPUs.

Can the same self-hosted design work for other processes?

Yes. The design behind RootCauseAI, read-only access, logging from day one and local models inside your boundary, is the same one our private AI infrastructure work applies to document search, assistants and reporting for clients.

Need AI that cannot leave your network?

We will walk you through the RootCauseAI architecture and how it maps to your rules.

Fixed-fee phases, agreed before we start. Stop after any phase.