1
2
3
4
5
6
7
8
9
10
11
12

Building and evaluating AI-assisted content workflows

By Tori Sanderson

View profile

Building and evaluating AI-assisted content workflows

By Tori Sanderson

View profile

Tori Sanderson
Founder & Managing Director

With over 15 years in digital strategy, product delivery and consulting, Tori leads teams that turn complex ideas into products that work. She brings creative energy, focussed clarity, and a knack for building positive, high-performing cultures.

Tori has delivered impactful digital products across banking, construction, healthcare, government and the not-for-profit sector. She’s driven by curiosity, connection and continuous learning – and by helping teams and clients achieve outcomes that matter.

Connect with Tori

Call 02 6189 4943[email protected]

How Avian helped a federal government agency evaluate whether AI could safely reduce the effort involved in producing complex regulatory content and build the evidence needed for an informed investment decision.

Designing AI PoCs to enable investment decisions

Government agencies are increasingly exploring how generative AI can improve productivity. The challenge isn’t simply whether AI works, but whether it delivers measurable value while meeting the standards expected of government.

One federal government agency engaged Avian to answer a practical question: could AI reduce the effort involved in producing complex regulatory content without compromising quality, accuracy or governance?

To test this, Avian built and evaluated a proof of concept using the agency’s own content, people and standards. The project measured quality, effort and return on investment, giving the agency a clear evidence base for future decisions.

Evaluating AI using business outcomes

Avian built the proof of concept using real production content rather than simplified examples.

Working alongside the agency’s content specialists, we translated existing content standards, writing guidance and regulatory requirements into a reusable AI workflow.

Independent content designers then evaluated the outputs against agreed quality measures before bringing them to publication standard. Alongside quality scores, we measured the time required to reach an acceptable first draft, creating a reliable comparison with the existing manual process.

Using the agency’s own content, people and quality expectations meant the results reflected real operating conditions rather than a controlled demonstration.

Project outcomes:

  • Independent evidence to support AI investment decisions
  • Around 92% less effort to produce a first draft, reducing drafting time from around six hours to approximately thirty minutes
  • Clear understanding of where AI adds value and where human expertise remains essential
  • A working proof of concept built within the agency’s existing Microsoft environment
  • A practical roadmap for responsible production use

Understanding where AI adds value

The evaluation showed that AI could produce accurate first drafts from structured source material while substantially reducing the effort required to begin the content design process.

It also highlighted where human expertise remains essential.

Content designers consistently improved clarity, readability, narrative flow and contextual judgement before publication. Rather than replacing specialist capability, AI proved most valuable as an accelerator that removed repetitive drafting work while leaving editorial ownership firmly with expert teams.

Supporting confident AI adoption

By the end of the engagement, the agency had far more than a proof of concept.

It had clear evidence of where AI delivers value, where governance controls are needed, and what would be required to use the technology confidently in production.

Because the solution was built within the agency’s existing Microsoft environment, the project also demonstrated how AI capability could be explored using existing government technology, security and governance arrangements.

Why it matters

Across government, organisations are looking for practical ways to improve productivity while maintaining high standards for quality, accountability and public trust.

The challenge is not adopting AI for its own sake. It is identifying where AI creates genuine efficiencies, understanding where people continue to add the most value, and making investment decisions based on evidence rather than expectation.

By designing and evaluating a proof of concept under real operating conditions, Avian helped a federal government agency quantify the potential return on AI, understand its practical limits, and build confidence in where the technology can deliver meaningful improvements.

We’re here to help—and always happy to chat.

Email us