Subsense

Read every submission. Not just a sample.

Subsense analyses, classifies, and synthesises high-volume public consultation submissions — so your team can tell ministers what the public actually said, backed by every voice, not just a convenient sample.

Built in partnership with Martin Jenkins.

See Subsense in action
Consultation Analysis
1,247 submissions
Identified Themes
Funding concerns
142
Implementation timeline
98
Regional impact
76
Workforce capacity
61
Sentiment
Support
34%
Concern
41%
Oppose
25%
Recent
Individual
Organisation
Individual

The consultation challenge nobody talks about.

Your team is running a public consultation. Five thousand submissions come in — maybe fifteen thousand. Your analysts are capable, experienced, and committed. But there aren't enough of them to read every submission in the time the Minister's office expects the analysis. So they sample. They categorise broadly. Important voices get lost in the volume. The synthesis is defensible, but everyone knows it could be deeper. The last consultation took fourteen weeks. Ministers wanted it in six.

How it works

Every submission read. Every voice counted.

1

Ingest

Upload your submissions in bulk — any format, any volume. Subsense processes text from PDFs, Word documents, emails, and online form responses.

2

Analyse

Every submission is read, classified by theme and sentiment, and cross-referenced against the full dataset. Key arguments and patterns are identified.

3

Synthesise

Subsense produces a structured analysis report with full traceability. Every finding links back to source submissions. Your analysts review, refine, and add their professional judgement.

Capabilities

What Subsense delivers.

Comprehensive reading

Every submission read in full — not sampled, not summarised from headers. Complete analysis across the entire dataset.

Theme and sentiment classification

Submissions classified by theme, sentiment, and stakeholder type. Patterns identified across thousands of responses.

Outlier detection

Novel positions and outlier arguments surfaced — not buried in the majority. The voices that matter most are the ones you might miss.

Te reo Māori support

Processes submissions in te reo Māori alongside English. Built for New Zealand's bilingual consultation landscape.

Full traceability

Every finding traces back to source submissions. Auditable, defensible, ready for Cabinet papers.

Structured synthesis

Produces analysis reports your team can review, refine, and present — working from a comprehensive foundation.

Built for

Built for teams running public consultations.

Policy Director

Running a public consultation generating 12,000+ submissions with a 6-week ministerial deadline for synthesis.

Complete analysis in weeks instead of months, with every submission read and every voice counted.

Chief Digital Officer

Agency-wide mandate to demonstrate AI capability with measurable impact on a real workload.

A production-deployed AI tool processing real government data with clear ROI.

Consultation Team Lead

Analysts overwhelmed by volume, resorting to sampling that risks missing important voices.

Analysts freed to interpret and advise, with comprehensive data as their foundation.

What changes when every voice is heard.

Your synthesis is built on every submission, not a sample. Analysis that took months takes weeks. Your analysts spend their time interpreting and advising — not reading and classifying. Your Cabinet paper cites a comprehensive, defensible methodology. And the people who took the time to submit know their voice was counted.

Your data stays in New Zealand

Built on Azure AI Foundry
Hosted within your tenancy
Enterprise-grade security
Privacy Act 2020 aligned
No data used for model training

See Subsense in action.

We'll show you how Subsense works on real consultation data — the kind of submissions your team actually deals with.