Why We Paused Building Our AI Advice Review Tool (And Why We Restarted)
- 3Lines Platform
- 6 hours ago
- 2 min read
3Lines co-founders Julia Vojkovic and Eloise Somerford recently sat down with Pat from the Ensombl community for their third AdviceTech conversation.
These discussions typically start with a loose agenda and quickly evolve into a broader conversation about technology, compliance and what is happening inside AFSLs.
This time the focus was on something the 3Lines team has been quietly developing in the background: Regi.
During the discussion, Julia and Eloise spoke about why they paused development of an AI advice review tool around 18 months ago, what has changed in AI capability since then, and why RegiReviews and RegiVets are now being tested in beta with a cross section of AFSLs and consultants.
One of the key themes explored was the complexity of advice reviews.
While objective checks such as disclosures, warnings and fees are relatively black and white, the real challenge lies in subjective judgement. Assessing whether reasoning is sound, whether another reviewer would reach the same conclusion, and whether the recommendation is genuinely appropriate for the client.
When the team first explored this space, the technology simply was not capable of supporting that level of reasoning. Rather than releasing a product they did not trust, development was paused.
As discussed in the podcast, AI capability has changed dramatically over the past 12-18 months, which has allowed the concept to be revisited.
The conversation also covered:
The difference between RegiVets (pre-vetting) and RegiReviews (post-advice supervision)
Why pre-vetting alone is not sufficient to meet supervision expectations
Why Regi is being developed as a standalone platform before integration into future platform architecture
And some of the realities behind building technology in the advice industry
