The “Now What?” of AI: your next step.
You saw the talk. Here is the one-page AI data guardrail template, free. Fill in the brackets, share it with your team, ship it Monday.
Missed the talk? The whole deck is at thehourglass.ai/ndis-talk
Half your team already uses AI in secret.
Rules beat bans.
What goes in
De-identified notes, templates, rosters and drafts: the green list your team can use today.
What never does
Participant-identifiable data, medication charts, incident reports: the red list, spelled out.
Which tools are approved
A one-glance table of approved tools and accounts, plus the rule that answers most questions: AI drafts, your team decides.
Every organisation climbs the same four steps.
Most of the room was on step one at the summit. That is not behind: that is step one of four, and nobody skips steps.
The 3 things, in order.
One page of guardrails
Give permission plus a guardrail. The AI policy is the foundation: it empowers your team instead of leaving them blocked on you. That page is the template above.
Write the list
Thirty minutes with your team. The ten most repetitive tasks this week, then pick one: high volume, low risk. The annoying one, not the critical one.
One pilot, two weeks
One task, one tool, two weeks. A finish line, not a science project. Your team with a template, a sector tool, or someone like us building it properly.
$586k
a year, sitting in the boring work
When we audited a 35-person business, nine interviews across eight teams identified $586k a year of repetitive work AI can draft: notes, reports, reconciliation, onboarding. Your version of this number is sitting in the list your team writes on Monday.
Want the list done for you?
A discovery call takes 20 minutes. We will find where the time is going in your organisation and tell you honestly whether AI is the fix: build, buy, or your team with a template.
Asked at every table.
It is a design choice, not a gamble. The guardrail template draws the line for you: de-identified drafting is fine in approved tools with data controls on, and participant-identifiable records never go into unapproved tools. Regulated industries did not skip AI; they scoped it. You already manage risk for a living, and this is one more risk register entry.
No. AI drafts the admin shell around the job: the shift note, the progress report, the roster shuffle. Your team reviews and decides, and the hours that come back go to face time with participants. The work that matters was never the paperwork.
The first two steps cost nothing: a one-page AI policy and a 30-minute team conversation. A two-week pilot with an existing tool runs to a few dollars per person per day. Meaningful budget only enters the picture once a pilot has shown exactly where the time savings are.
The one your pilot needs, not the shiniest one. For the boring list, ChatGPT or Claude on a work account plus a good template gets you shockingly far. Sector-specific tools exist for care management, compliance reports and coordination. And when workflows get complex, that is what we build.
You do not have to build the whole thing. You just have to do the next thing.