ISCOVER
A new AI tool aimed at simplifying lender policy research and client recommendation has launched, co-founded by two brokers.
A new artificial intelligence (AI) platform designed to streamline lender policy research and support broker compliance has officially launched, co-founded by LMG broker couple Estelle Dejean and Devon Mohi from FINANS Brokers and software specialist Emilien Bourgeois.
The tool, called Thryvve, enables mortgage brokers to input client scenarios using natural language and receive lender-specific policy responses, based on verified policy documentation. The platform launched publicly on Monday (4 August).
Brokers using the platform can ask detailed policy questions and receive answers directly linked to lender documents. It also includes functionality to compare lenders and generate client-facing summaries.
The founders said the platform was built to reduce administrative work and support brokers in meeting best interest duty (BID) obligations.
Speaking to The Adviser, Dejean said the idea stemmed from challenges she faced in her own brokerage.
“We were constantly switching between lender portals, PDFs and chats with BDMs, trying to find the right policy fit. It was repetitive and inefficient,” she said.
“We didn’t want just another tool — we needed a solution that thinks the way brokers do, that understands complex scenarios and delivers accurate, verifiable results. That’s what we built.”
She added: “Especially under BID, brokers can’t afford to rely on guesswork or just stick with their usual lenders. We need tools that support smarter, more compliant recommendations. Thryvve helps us do that.”
Mohi added: “As our brokerage scaled, so did our need for speed, accuracy, and confidence, without sacrificing compliance. For us it was about being able to recommend to our clients faster, with certainty and provide them with a better experience. We couldn’t find anything that truly solved the problem, so we built it.”
How does it work?
Thryvve was tested by more than 60 brokers in a six-week beta phase prior to launch. Feedback cited reduced time spent on loan research and “portal hopping”, increased confidence in recommendations, and improved policy navigation.
Brokers who tested the platform used it to clarify complex loan scenarios, such as whether clients can access equity in a property with multiple owners on title, or how income is assessed for specific occupations (for example, a diesel mechanic employed by Brisbane City Council) by different lenders. These queries highlight the need for lender-specific guidance on ownership structures and income policy.
The platform was developed using large language models (LLMs), built on OpenAI’s GPT-4 framework and customised with a proprietary orchestration layer designed to ensure responses are grounded in lender documentation.
The system uses retrieval-augmented generation (RAG) architecture to minimise hallucinations by drawing only from approved source material.
Thryvve’s engineering lead Emilien Bourgeois said the platform was deliberately designed to avoid speculative answers. “If a policy doesn’t exist or can’t be verified, the tool will say so — that’s critical in a regulated industry like broking,” he said.
Lenders send policy changes to Thryvve ahead of go-live dates, allowing the team to release updated rules the moment they go live, or schedule them for future-dated updates.
Unlike keyword-based tools, Thryvve analyses lender documents at a granular level, enabling users to assess complex borrower scenarios across a range of variables, according to the co-founders.
It analyses multiple parts of a scenario, from employment type to loan purpose, product type,
and niche layers like exit strategies or ownership structure, then pulls lender policies from
different areas and compiles an intelligent summary.
“It doesn’t just spit out a yes/no answer. It shows you the ‘why’, the strategy, the considerations,
and the lender positioning, all backed by real policy,” said Bourgeois.
”For every key point in your scenario, the platform tells you exactly where in the lender’s policy
the information came from, citing the policy section so you can easily refer back to the source
yourself.”
The platform operates via a subscription model, starting at $45 per user per month. It is aggregator-agnostic and the team are currently in discussions with several aggregator groups, including LMG, for potential integrations.
The founders told The Adviser that they have plans to expand the platform’s functionality to include pricing and borrowing capacity tools within the next 12 to 18 months.
Plans are also afoot to see whether lenders would adopt the technology in-house. Some lenders are reportedly exploring direct integrations into their broker portals or using Thryvve
internally with their own assessors.
“We’ve sat down with dozens of lenders,” Dejean said. “They all agree, this isn’t just another tool
like some of the ones available on the market which have been presented to them. It’s deeper,
smarter, and it’s broker-first.”
Thryvve is one of several AI-powered tools launched recently by brokers seeking to improve speed and accuracy in loan recommendations, including include The Broker’s Bible (recently acquired by LMG) and Cynario.
Major brokerages such as Loan Market and Aussie have also recently announced major investments in AI strategies to help free up broker time and boost efficiency.
[Related: Aussie to roll out agentic AI]
JOIN THE DISCUSSION