A note on these notes

These guidance notes are intentionally a starting point. As RiskBot evolves and we learn more about how advisers use it day to day, we’ll add worked examples, deeper topic guides, and tips on making the most of specific features. For now, this covers the basics to get you productive quickly.

About RiskBot

RiskBot is an AI research assistant focused on Australian retail life insurance. It’s built on a curated library of insurer PDS extractions, adviser guides, technical reference material, and regulatory source documents. You ask questions in plain language. RiskBot responds with referenced, structured answers you can use directly in your research process.

RiskBot covers the full panel of 11 retail life insurers and is designed to go deeper than a generic search, with verbatim definitions, page references, and the ability to compare multiple insurers side by side.

Plans and why they work this way

RiskBot plans are built around query numbers rather than time. Every question you ask runs through a large language model (Claude, by Anthropic), and each query has a real underlying cost. Grouping plans by query count keeps pricing transparent and lets you match a plan to your research rhythm.

A few practical notes:

• Each message you send counts as one query. Viewing past conversations, browsing history, favouriting, and navigating don’t use queries
• Choosing an output format (Detailed, Summary, Client-friendly) doesn’t use a query on its own, but re-running the same question in a different format does
• You’ll see your remaining queries in the top nav bar at all times
• If you run out mid-month, you can upgrade to a higher plan or top up

Two modes of research

RiskBot has two modes, selected at the top of the chat area.

PDS research draws on the insurer PDS and adviser guide library. Use this mode when you want product-specific answers: definitions, exclusions, benefit structures, waiting periods, occupation classes, cross-insurer comparisons.

General knowledge draws on the technical reference and regulatory library. Use this mode for broader topics: tax, business structures, regulatory guidance, FASEA obligations, compliance considerations, market context.

Switching modes starts a fresh conversation. Your typed text is preserved so you don’t lose your question.

A note on query scope

RiskBot works by loading the relevant source material for each question you ask. For deep, detailed comparisons, this means there’s a practical limit on how many insurers can be included in a single query.

Detailed comparisons: four to five insurers per query

When you’re asking about definitions, benefit structures, exclusions, waiting periods, occupation classes, or other detailed PDS content, RiskBot typically works best with a selection of around four or five insurers per query. This is because each full PDS extraction is substantial, and loading the full detail for all eleven insurers at once would exceed what the AI model can process reliably in a single pass.

The Refine Query panel shows a context indicator as you select insurers, so you can see when you’re approaching the limit. If your selection is too large, RiskBot will let you know and you can narrow your focus.

Lighter queries: often fine across the full panel

Simpler queries that don’t need the full detail of each PDS (for example, eligibility ranges, high-level benefit period options, underwriting requirements, or which insurers offer a particular product type) will often work across all eleven insurers in a single query.

Practical approach

If you need a detailed comparison across more than five insurers, the most effective approach is to run a couple of targeted queries, each focused on a subset of insurers, and compare the results. This actually tends to produce clearer answers than trying to stretch a single query too wide.

We’re also actively working on ways to support broader comparisons within a single query over time.

Types of queries you can ask

Single-insurer definition: “What is AIA’s definition of total disability for IP?”

Exclusions: “What are the exclusions for Zurich trauma cover?”

Cross-insurer comparison: “Compare TPD own occupation definitions across AIA, TAL, and Zurich”

Operational detail: “How do IP offsets work for MetLife?”

Regulatory or technical: “What are the FASEA Code of Ethics obligations for life insurance advice?” (General knowledge mode)

Scenario-based: “Client has a pre-2021 agreed value IP policy and wants to increase cover. What should I consider?” (General knowledge mode)

Making the most of RiskBot

Name the insurers you care about. RiskBot will auto-detect insurer names in your query, but selecting specific chips in the Refine Query panel focuses the research and reduces cost per query. You’ll see a context indicator showing how much of the available budget your selection uses.

Pick the right output format. Detailed gives you full citations and verbatim blocks. Summary condenses into key points and tables. Client-friendly rewrites in plain English suitable for a client letter or Statement of Advice.

Use the cover type filter. In PDS mode you can narrow to Life, TPD, Trauma, or IP to keep responses tightly scoped.

Ask follow-up questions. Each conversation supports follow-up queries that build on the previous context. If you want a fresh slate, click “New conversation” in the sidebar.

Start broad, then drill down. Begin with a comparison question across a few insurers, then ask follow-up questions to dig into the ones that stand out for your client’s situation.

History and favourites

Every conversation is saved. The left sidebar shows your recent queries. The History page (top nav) groups your conversations by date, with search and mode filters to help you find older work.

Favourites let you star useful conversations so they’re easy to find again. On any response or history item, click the heart icon to favourite. Favourited conversations are protected from accidental deletion, and you can filter the History page to show only favourites.

Re-running with current data. When you open a past conversation, you’ll see the date it was originally run. If you want the latest answer (in case the PDS library has been updated), click “Re-run with current data” and RiskBot will run the same query against the current source documents in a fresh conversation. Your original response stays intact.

Copying outputs into your work

Every response has two copy options in the top-right corner:

• Copy (clipboard icon): copies the response as rich text with formatting preserved, so tables, bold text, and headings paste correctly into Word, Google Docs, or an email
• Copy as Markdown: copies the raw markdown, useful for technical systems or where you want a plain text version

The copied content includes the query, date, and mode at the top, so you have a clear audit trail in your research notes.

Feedback, ratings, and issues

RiskBot improves based on real adviser use. Three ways to help:

Thumbs up / thumbs down on any response. If you thumbs down, you can optionally add a short reason. This helps us identify patterns, where the tool performs well and where it needs work.

Issue icon (flag) on any response. Use this when something is specifically wrong: a definition that doesn’t match the PDS, a missing insurer, an inaccurate comparison. The full context (your query, the response, the insurers selected, the mode) is captured so we can investigate and fix.

General feedback: drop us a note at info@riskhub.com.au with suggestions, queries that didn’t work well, or features you’d like to see.

We read everything. Many of the current features exist because advisers in the early user group asked for them.

What RiskBot is not

A few reminders to keep it useful and safe:

• RiskBot is a research tool, not financial advice. It surfaces information to support your professional judgement, not replace it
• Critical definitions should always be verified against the source PDS before being used in client advice
• Scenarios should be described generically (e.g. “a 45-year-old self-employed accountant”), not with client-identifying information
• RiskBot does not process client health or medical information

Contact and support

For feedback, questions, or support: info@riskhub.com.au

Last updated: April 2026