Sharp vs Bulma — AI pricing vs AI policy

Two of the most-named AI tools for Australian mortgage brokers in 2026. They solve different problems — one answers "will this lender accept this borrower", the other answers "what's the live price". Here's the honest comparison.

Published 16 May 2026 · By James Larkey, founder of SecondBrain · ~5 min read

What each tool actually does

Bulma

Bulma is an AI policy intelligence platform for Australian mortgage brokers. It replaces the cognitive load of remembering hundreds of lender guidelines with instant, cited answers. A broker can ask a natural-language question — "Does this lender accept casual income with less than 6 months tenure?" or "What's this lender's policy on guarantor loans?" — and Bulma returns an answer with citations back to the lender's published policy. It's built for the early-fact-find and pre-lodgement question: will this lender even consider this borrower.

Sharp

Sharp is an AI bank pricing tool. A broker submits a request inside a dedicated Microsoft Teams workspace; Sharp queries every lender on the broker's configured panel simultaneously and returns indicative rates, fees, and offer position back in the same Teams channel — typically in about 4 minutes. It's built for the later question, once policy compliance has been established: what's actually the best price across the panel today. Sharp also runs inside the broker's own Microsoft 365 tenant, which means client data never leaves the broker's environment.

Side-by-side

Why they're not competitors

Policy and pricing are sequential steps in the broker workflow, not alternatives. A pricing comparison across the entire panel is irrelevant if half of those lenders won't lend to this borrower under their policies. Conversely, a perfect policy match is useless if the surviving lenders are 60 basis points apart on rate and you don't know which is which. The brokers we see getting the most leverage run Bulma early in the file to narrow the panel by policy fit, then run Sharp on the surviving lenders to get the real pricing position.

Where Bulma wins

Bulma is the right tool when the question is does this lender's policy fit this borrower. A senior broker can carry hundreds of lender quirks in their head; a newer broker can't. Bulma flattens that experience gap by surfacing cited answers in seconds. For brokerages with junior staff, complex client profiles (self-employed, casual income, foreign income, guarantor structures), or panels with many non-bank lenders, the policy lookup time saved is the highest-leverage win.

Where Sharp wins

Sharp is the right tool when policy is settled and the question becomes which lender is actually cheapest right now. Bank pricing changes weekly — sometimes daily — and policy tools don't track live pricing. Sharp queries each lender directly so the comparison reflects the rate the broker can actually quote the client today. Because it runs inside the broker's own Microsoft 365 tenant, the security review for adopting Sharp is materially shorter than for vendor-hosted SaaS.

The combined workflow most brokerages converge on

1. Fact-find: capture client profile.
2. Capacity (Quickli or similar): filter lenders by serviceability.
3. Policy fit (Bulma): narrow further by lender-policy compatibility.
4. Pricing (Sharp): pull live pricing from the surviving lenders, present comparison to client.
5. Lodgement: submit through the aggregator (SecondBrain automates the data entry).

Three different AI tools across the workflow, each doing exactly the job it's purpose-built for, no overlap. This is what the AI broker stack actually looks like in 2026.

When neither is right

If you write one loan a month, both tools are overkill. If your panel is tiny and you know every lender's quirks by heart, the policy tool is overkill. These tools earn their place when volume is high enough that small per-file time savings compound into real recovered capacity.

Frequently asked questions