Skip to content
FAQ

Straight answers.

Including the ones about what it cannot do yet.

What is Intelli-Dev?

Intelli-Dev is an AI software delivery pipeline. It takes a work item from your tracker, or a plain-language request, and drives it through specification, human approval, build, and a set of quality gates, then opens a merge request in your own repository for your reviewers and your CI/CD.

It is a delivery pipeline rather than a coding assistant. The output is a reviewed change in your repository, not a suggestion in an editor.

How is this different from a coding assistant or an agent in my editor?

A coding assistant helps a person write code faster. Intelli-Dev carries a unit of work from request to merge request, including the parts that are not typing: writing the specification, splitting the work into stories, running checks, opening the merge request and following the pipeline result.

The other difference is durability. Assistants keep their context in a conversation. Intelli-Dev keeps it in an order that survives restarts, can be resumed after a failure, and leaves a record of what was asked, what was checked and who approved it.

Does it commit straight to my main branch?

No. It works on branches and opens a merge request. Your branch protection rules, your required approvals and your reviewers are unchanged, and nothing merges without them.

Which trackers and code hosts does it work with?

Azure DevOps and GitLab are the reference integrations, because they are what we build and test against ourselves. GitHub and Jira are next.

It is designed around the assumption that you already have a tracker and a code host and are not going to change them.

Can it run on our own infrastructure?

That is the model it is built for. We call it Whiteboxing: the whole product deployed inside your network, single tenant, working against your repositories, with model routing you can point at models you host yourself.

It is not generally available yet. The release engineering that a supportable on-premise product needs, including versioned releases and a tested upgrade path, is work in progress. If this is the deployment you need, say so when you join the waitlist.

Does our source code leave our network?

In a Whiteboxed install it does not need to. The pipeline runs in your environment and inference can be pointed at models you host, which is the point of building it that way.

Any other arrangement depends on how the install is configured, and we will be specific with you about it rather than vague.

What do the quality gates actually check?

Four gates run before a merge request is opened: an automated code review pass, a QA pass that generates and exercises tests against the acceptance criteria, a security review of the diff, and a spec completeness check.

The spec completeness gate is the unusual one. It compares the endpoints and behaviour the approved specification promised against what the built application actually serves, and blocks when something promised is missing. It exists because conventional gates judge the code in front of them and cannot see the code that was never written.

Where does a human stay in control?

At two points, both deliberate. A person approves the specification before any code is written, and your reviewers approve the merge request before anything lands.

The approval checkpoints are not an optional safety mode. They are the shape of the product.

What happens when it gets something wrong?

The gates are there to catch it before a person is asked to look, and the merge request is there to catch what the gates miss. When a stage fails, the order records the failure and the work can be resumed rather than restarted.

It will get things wrong. Everything that writes code does. The design assumption is that the checkpoints and the record matter more than pretending otherwise.

Does it deploy to production?

No, and that is on purpose. The pipeline ends at a merge request and a run of your existing CI/CD. It does not hold production deployment credentials.

A tool that has to own your build and hosting to be useful is a tool most organisations cannot adopt without a migration first.

What does it cost?

There is no pricing yet, because there is nothing yet to buy. Intelli-Dev is in early access and we are not taking orders.

When it is commercially available the on-premise model will be licensed rather than metered per request. Joining the waitlist puts you in the conversation before that is fixed.

How far along is it really?

It is real software that we use ourselves on real work, and it is not generally available. Two items gate our own path to selling it: hardening how integration credentials are stored and rotated, and an independent penetration test with findings addressed.

We would rather tell you that now than have you discover it during a procurement process.

Who is behind it?

IntelliInfra AI Pty Ltd, an Australian company based in Caboolture, Queensland. We also build Intelli-Assist, a business AI platform for Australian small and medium businesses.

How do I get early access?

Join the waitlist and tell us where your work lives and what is stuck. We are prioritising teams already using Azure DevOps or GitLab, and teams who need it to run inside their own network.

Question not answered?

Ask it on the waitlist form. A person reads every one.

Join the waitlist