All resources
Case study

An AI assistant for practitioners working from statutory guidance

Structural analysis of a statutory practice framework, then a grounded assistant over the same corpus

DeliveredAugust 20263 min read
On this page
Client
A government agency responsible for a statutory practice framework
Domain
Practice guidance
Engagement
Proof of value, structural analysis then a practitioner trial
Period
Over several months
Applied
PolicyDNA, AccuFind Virtual Assistant
Status
Delivered

The challenge

An Australian government agency maintains a practice framework running to over a thousand pages, with more than five hundred cross-references between its documents. Practitioners in a number of separate organisations are expected to apply it consistently.

They reach for it in the field, between appointments, and on the phone. Until this work the options were a paper copy, keyword search across a folder of PDFs, the website, or asking a colleague.

The solution

  • The guide set analysed as a whole first, every guide read against the others and against the legislation behind them.
  • An assistant built over the same corpus. Tenant isolated, hosted in Australia, access limited to the trial cohort.
  • Every answer cites the paragraphs it was built from, and the system withholds an answer it cannot cite.
  • Four rounds of expert scoring against a fixed question set before any practitioner saw it, with retrieval and generation scored separately.
  • A two-week trial across several organisations, with a cohort trained on the assistant and use left optional.

The impact

177issues found before anything was built
100pieces of guidance held in more than one document
25places where two copies say different things
85+different names used for the same concept
What the analysis found in the guidance itself

In twenty five of those places the two copies say different things, so two practitioners could each follow the guidance correctly and arrive at a different answer. Finding them previously meant reading the whole set.

The agency holds all 177 as a consolidation roadmap, each one showing the passage it was found in and the passage it conflicts with.

9 in 10used it without being required to
83%of rated answers were judged useful
Every answercited the paragraphs it came from
Two weeks with a cohort trained on the assistant, use left optional

Practitioners rated about half the answers they received, so the 83 per cent covers the answers they chose to judge rather than all of them.

The agency set its stopping condition before the four rounds of expert scoring began: an answer that was wrong or misleading. None occurred in the final three rounds.

Delivered as a consolidation roadmap, a working assistant, the round-by-round evaluation, a practitioner trial findings pack and a forward plan.

Considering a grounded assistant over your own guidance?

AccuFind runs structural analysis and evaluation like this inside its engagements, and publishes the method it uses to score them. The corpus analysis is worth having whether or not an assistant follows.