Give your people AI.Keep control of your data.

Signals gives your teams a sanctioned AI workspace for sensitive work. Routine answers flow straight away, higher-risk ones are approved by a person, and every step can be reconstructed.

Start with one team. Show the value, then scale it as a supported service.

What your teams use it for
on the work that matters most

Each job runs on your own documents, with the safeguards that work needs switched on.

Research contracts and agreements

Drafted against your templates, with confidential terms kept in Australia.

Grant, CRC and ARC bids

Built from your past bids and capability statements, cited back to each source.

Partner and industry briefings

Answers on who does what, drawn only from documents each person may read.

Policy and procedure questions

Staff get the answer with the clause it came from.

Tender responses

Reuse past submissions, with commercial-in-confidence material approved before it leaves.

Board and executive reporting

Summaries with their sources, signed off before they are shared.

Research security reviews

Sensitive partner information stays inside your rules, and every check is recorded.

Compliance and audit evidence

A record of who asked what and who approved it, ready when the auditor calls.

See it working
on your own documents

Ask in plain language. Get answers you can trust, with every source shown and a person in the loop where it matters.

Ask your own documents and get cited answers

People ask in plain language. Every answer cites the passages it drew on, from documents the person is cleared to read.

Want the full picture?

Request a guided demo

Every question passes your checks
before anyone sees an answer

Pick a question and watch what happens. People only see what they are allowed to, and sensitive answers wait for a person before they go anywhere.

Try it

Pick a question. Watch your policy decide.

Public

How do I apply for a planning permit?

Internal

How long do I have to lodge a travel claim?

Confidential

Summarise the board paper for the executive briefing pack.

Confidential · sent outside

Draft a note to the supplier about the conflict of interest.

Restricted

Summarise the open security investigation.

Check the classPrinciple 2

Internal, from the question and its sources.

Search within clearancePrinciple 1

Only what you are cleared to read.

Approved model?Principle 3

Yes. Sent only to a model approved for this class.

Answer with sourcesPrinciple 4

Each passage is cited and the answer is checked.

Needs sign-off?Principle 5

No. Below the level your policy sets.

Signals in the LoopPrinciple 5

Not needed.

Released

Free to export, with its sources cited.

Every step is recorded, and the record cannot be changed.

Run Signalswhere your data has to stay

Run it with us, in your cloud or on your own servers. The safeguards are the same everywhere.

Hosted by us

We run it for you. Your documents and records are stored in Sydney.

Running now

Your cloud

Runs in the cloud account you already use.

Available on request

Your data platform

Runs inside the data platform your teams already work in.

Available on request

On premises

Runs on your own servers. Nothing has to leave your network.

Available on request

Architecture and security detail is in the security pack.

Request the pack

Choose where your questions
are processed

Your documents and records are stored in Australia. You also choose where questions are answered, from the newest global models to your own network. Confidential work in Australia is our recommended default.

The newest models for everyday work. Confidential questions and documents go only to models in Australia, and restricted questions never leave your own network.

Best for

  • The newest models for public and internal questions
  • Fits most universities and companies

Trade-offs

  • Confidential answers come from the models offered in Australia, which can trail the newest by months
  • Public and internal text may be processed overseas

Only administrators can change the setting. Signals shows what a change would stop before it is made, and records every change.

For IT and digital leaders

An AI service
your IT team can approve

Signals turns scattered AI pilots into a sanctioned service, with the controls and evidence your governance committee asks for.

Sanctioned AI tooling

One approved workspace for sensitive work, alongside the tools your staff already use.

Proportionate governance

Routine answers go out straight away. Higher-risk ones go to a person, and a team can be held to a stricter rule as its use changes.

Evidence that higher-risk uses were authorised

Every question, approval and export is recorded, so you can show who approved what and reconstruct it later.

Access decisions as system controls

People only see what their clearance allows, and every answer shows its sources.

Value you can show

Use, approvals and answer quality are recorded from the first day, so a pilot has evidence before it becomes a service.

A cost you can sustain

A team-sized price, with no lock-in to one model or one cloud.

Accountable when others hold the data

You choose where it runs and where questions are processed.

See whether every answer
is getting better

Every answer is scored the same way every time, so you can see your AI getting better.

Every answer is checked for

  • Supported by its sources
  • Answers the question
  • Cites its sources
  • Follows its instructions
  • Clear

Every document is checked to see if it is

  • A policy people must follow
  • Current
  • Complete

Built for organisations
that answer for their data

Small businesses

Useful AI on a small business budget

Give your team useful answers from its own documents on an affordable small model, without sending your data to public tools.

Enterprise

Stop data leaking out of the business

Customer and commercial information stays inside the business, and every decision is on the record.

Government

Meet the mandate for responsible AI

Your data rules are enforced automatically, with the evidence ready for your auditors.

Universities

Responsible AI across teaching and research

Give staff and researchers AI for teaching, research and contract work, with research security built in.

How organisations are using Signals
to keep sensitive work safe

Customer story

Swinburne University of Technology

AI for partner contracts and research proposals, without the risk.

The challenge

Swinburne's industry engagement team works with partner contracts, research proposals and commercial-in-confidence material every day. The team wanted AI to help with that work while keeping each partner's information inside the University's rules.

What it needed

It had to know which documents were sensitive, keep them away from unapproved AI, and bring a person in before anything sensitive went out.

How Signals did it

Every document and question is labelled by how sensitive it is. Sensitive answers go to a person for sign-off. The most sensitive questions are refused, and everything is recorded.

The Responsible AI Foundation

Your AI policy,
enforced inside the product

Developed with Swinburne's AI researchers, the Foundation turns your AI policy into rules the product follows on every question. Try each principle below.

A Swinburne spinout

Signals is being spun out of Swinburne University of Technology.

In use at Swinburne

The Signals suite already supports industry engagement and research teams at the University.

Principle 1Enforced

Your database is the boundary

People only see what their clearance allows, however they ask.

Switch the clearance and watch what this person can open.

  • PublicVisitor FAQ
  • InternalTravel policy
  • ConfidentialTeam budget
  • RestrictedBoard paper

See each principle with evidence from your own records.

Request a guided demo

Book a pilot today
and start with one team

Start with one team on its own documents. You see the value on real work before you decide how far to take it.

Step one

Scope it

We agree the team, the documents and the questions that matter most to them.

Step two

Bring in the documents

Upload files or connect a source. Each document is labelled by how sensitive it is.

Step three

Put it to work

The team uses it on real work, with every safeguard switched on.

Step four

Measure and decide

We measure the time saved with you, and you choose where it goes next.

Every pilot includes

  • Cited answers from your own documents
  • Every Responsible AI control, switched on
  • Signals in the Loop for your approvers
  • Support from our team throughout

Give one team a safe place to use AI this quarter.

Considering a pilot? Ask for the security pack and a guided demo.

Questions leaders
ask us

Will this become another tool IT has to chase?

No. Signals is built to be approved and run by IT. It uses your single sign-on, keeps a full record of use, and lets administrators decide which models each team may use.

How do we show the value of a pilot?

Use, approvals and answer quality are recorded from the first day. At the end of the pilot you have evidence of who benefited and what it costs to run.

Where is our data stored and processed?

Your documents and records are stored in Australia. You choose where questions are processed: with the newest models wherever they run, with confidential work kept in Australia, in Australia only, by Australian-owned providers, or on your own servers. Restricted questions never leave your control.

Does it work with the tools we already use?

Yes. People sign in with your identity provider, and documents come in from SharePoint, OneDrive, cloud storage or a server folder.

Will approvals slow our people down?

No. Most answers are released straight away. A person only steps in when an answer is sensitive or unclear.

How do we know the answers are any good?

Every answer is scored for accuracy, sources and clarity, so you can see quality improve over time.

Which models does it use?

The ones you approve, from the newest AI models to small models running on your own hardware.

Can we see the architecture and security detail?

Yes. Organisations considering a pilot receive the security and architecture pack and a guided demo of the full workbench. Ask us for it.

What happens in a pilot?

One team uses the workbench on its own documents. We measure the time saved together, and you decide how far to take it.

Give your people AI, keep control of your data

jglesta@swin.edu.au · Joseph Glesta, Senior Manager Industry Engagement, BGS, Swinburne University of Technology

Controls built in.

Access limited by clearance

Every action recorded

Single sign-on with a second factor