Skip to main content

Platform · AI

AI that does the record-keeping.
People who make the calls.

Calls, emails and submissions arrive faster than anyone can file them. VivoAI’s onshore AI transcribes, summarises, links and drafts, and your team confirms what goes on the record.
  • AI inference in AWS Sydney
  • Confidence on every output
  • Audit trail on AI-assisted decisions

The problem

The record is only as complete as the last person who filed it.

Engagement teams don’t lose evidence because they don’t care. They lose it because filing, linking and cross-checking compete with the actual work of talking to people.
  1. 01

    The call nobody wrote up

    A resident rings at 7pm. The officer means to log it in the morning, and by then three more calls have come in. The conversation happened; the record says it didn’t.
  2. 02

    The note that isn’t linked

    The complaint is logged, but not against the property, the issue or the approval condition it relates to. When the auditor asks about condition 12, nobody finds it.
  3. 03

    The promise that changed

    One person commits to something on a site visit. Months later, someone else changes it in an email. Nobody notices until the landholder’s solicitor does.

What the AI does

Five jobs the AI takes on, and where a person signs off.

  1. 01 · Capture

    Every call transcribed and summarised

    Calls are transcribed automatically and summarised in plain English, with themes and sentiment attached. The same happens for emails, survey responses, web forms and social media, so every channel lands in one record.

    Language analysis handles Australian English, place names and regional terminology, and every output carries a confidence level rather than pretending to certainty.

    Where a person decidesA person reads the summary against the transcript before it is relied on.

  2. 02 · Link

    Suggested links, confirmed by a person

    A record is only useful if it is attached to the right things. VivoAI suggests the stakeholder, property, issue and approval condition each interaction belongs to.

    An exact phone or email match links a known stakeholder automatically. Anything ambiguous becomes a suggestion that someone on your team accepts or rejects, and the record shows who did.

    Where a person decidesAmbiguous links wait for a person to accept or reject them.

  3. 03 · Check

    Contradictions flagged before someone else finds them

    Every commitment is tied to who made it, when, and to whom. When a later commitment appears to change an earlier one, VivoAI flags the pair for review.

    It is a flag, not a finding. Your team decides whether the commitments really conflict and what to tell the people affected.

    Where a person decidesThe engagement lead decides whether it is a real conflict.

  4. 04 · Warn

    Early warning when concern starts building

    Sentiment velocity measures how quickly concern is building, not just whether it is positive or negative. It tracks the rate of change across email, social media, survey responses and transcribed calls.

    You set the alert thresholds. When one is crossed, project leads and communications teams are notified, so you can act before a concern becomes a formal objection.

    Where a person decidesYour team sets the thresholds and decides how to respond.

  5. 05 · Draft

    Replies drafted from the record, reviewed before sending

    Onshore AI drafts responses from the call and the commitments on file, and prepares summaries, briefing notes and engagement plans. Each draft shows the records it was drawn from.

    Specialist time goes on judgement and verification, not on a blank page.

    Where a person decidesDrafts are marked for review; a person edits and approves them before they go out.

Consultation at volume

Every submission sorted by theme, with the unique ones kept.

An exhibition period can bring in more submissions than a team can read in weeks. VivoAI clusters them into themes, groups form-letter campaigns and surfaces the concerns that would otherwise be lost in the volume.

Themes sit in a tree you can drill into, from the high-level issue down to individual submissions, and cluster summaries export in a format ready for ministerial briefings, board reports and EIS responses.

Where
Sentiment and issues overlaid on project maps, down to postcode and, with cadastral data, to the property. Play back how concern moved across project phases, and see hotspots that need targeted engagement.
Who is connected
Who is connected to whom on an issue: shared organisations and affiliations, and who appears in the same records, filterable by issue, location and engagement history, so a new officer can see who has already been involved.

Guardrails

AI you can explain to an auditor.

The AI assists. People decide what goes on the record and what goes out in reply. Every AI capability on this page runs in the AWS Sydney region.

Human review
AI suggests; people confirm. Ambiguous links, flagged contradictions and drafted replies all wait for a person, with confidence scores and an audit trail on every AI-assisted action.
Confidence scores
Summaries, sentiment, themes, risk scores and classifications each carry a confidence level, and risk scores come with a stated rationale.
Audit trail
AI-assisted decisions keep a full audit trail: what was suggested, who accepted or rejected it, and when.
Onshore inference
AI inference runs in the AWS Sydney region, enforced in code rather than offered as a setting. The platform refuses to start if pointed anywhere else, and refuses to call any model outside its Australian allowlist.
Offshore paths
Offshore AI and geocoding are disabled by default and fail closed: an API key alone will not enable them.
Security baseline
Built against the Australian Government Information Security Manual (ISM), ISO/IEC 27001 controls and the ASD Essential Eight. Not IRAP-assessed. The AWS Sydney infrastructure we run on is IRAP-assessed; that covers our infrastructure.

AI questions

What teams ask before they trust the AI.

  1. What does VivoAI’s AI actually do?

    It transcribes and summarises calls, suggests the stakeholder, property, issue and approval condition each interaction belongs to, flags commitments that appear to contradict each other, warns when concern is building, and drafts replies, summaries, briefing notes and engagement plans. Your team confirms what goes on the record and what goes out in reply.

  2. Does a person review AI output before it is used?

    Yes. Ambiguous links, flagged contradictions and drafted replies all wait for a person, and drafts are edited and approved before they go out. Every AI-assisted action carries a confidence score and an audit trail of what was suggested, who accepted or rejected it, and when.

  3. Where does the AI processing happen?

    AI inference runs in the AWS Sydney region, enforced in code rather than offered as a setting. The platform refuses to start if pointed anywhere else and refuses to call any model outside its Australian allowlist, and offshore AI is disabled by default and fails closed.

  4. What is sentiment velocity?

    Sentiment velocity measures how quickly community concern is building, not just whether sentiment is positive or negative. It tracks the rate of change across email, social media, survey responses and transcribed calls, and notifies project leads and communications teams when a threshold you set is crossed.

  5. Can AI help us analyse a large volume of consultation submissions?

    Yes. VivoAI clusters submissions into themes, groups form-letter campaigns and surfaces the unique concerns that would otherwise be lost in the volume. Themes sit in a tree you can drill into, and cluster summaries export in a format ready for ministerial briefings, board reports and EIS responses.

  6. Does the AI understand Australian place names and terminology?

    Language analysis handles Australian English, place names and regional terminology. Every output carries a confidence level rather than pretending to certainty, and a person reads each summary against the transcript before it is relied on.

Next step

See the AI work on a record like yours.

Bring a recent call, complaint or consultation. We’ll show you what VivoAI would suggest, and where your team would confirm it.