Abacus Global (ABX) AI Policy Review: 9,314 Policies in Q2

Abacus Global Management, Inc. (ABX) said on its August 6, 2026 earnings call that it reviewed 9,314 qualified policies in the second quarter. Management attributed a year-to-date review count of more than 50,000 policies to using AI to add leads and shorten review time, but it did not break out AI's separate contribution[1]. That makes AI policy review at Abacus a working operation with a countable output, though not yet a measured financial result.

Where AI sits in the life settlements business

Abacus works in life settlements. It buys life insurance policies from policyholders who no longer want to keep them. It then either resells the policies to institutional investors for a spread, or holds them, keeps paying the premiums and waits for the payout. It also manages policy portfolios for third parties. Policies reach the company mainly through financial advisors and insurance agents, through brokers, and through a direct-to-consumer channel. The company's origination team first has to check each policy to confirm that it is in force and that it meets the purchase criteria.

The AI discussed on this call is used at that front end. The team uses AI to expand the flow of potential policy leads and to shorten the review time for each policy case, so the same team can finish reviewing more policies in a quarter[1]. The company did not give the approach a product name and did not explain how the technology works. Buying policies is the starting point for Abacus's later trading and holding revenue, so this application sits in the company's core operations and covers one business unit, policy origination.

Already running when first disclosed

When the company first mentioned this application publicly, on the August 6, 2026 call, it was already in use. Management talked about review volume that had already been completed, not about a plan or a pilot[1]. So far this call is the only time the company has discussed the application in public. The company has not said when it started using it or what stages it went through before.

The AI uses that Abacus described on earlier calls centered on pension participant verification, longevity financial planning and health data analysis. Those differ from this application in who uses them, where they sit in the workflow and what they process. Nothing indicates that this application is a renamed, upgraded or merged version of any of them.

Two figures measured on different bases

The two numbers management gave are not measured the same way. The 9,314 figure is a single-quarter reading for the second quarter and counts only qualified policies. The figure of more than 50,000 is a cumulative year-to-date reading that also counts non-qualified policies. The two cannot be added together, and one cannot be divided by the other to get a qualification rate[1].

Company executive William McCauley said in prepared remarks that the milestone was achieved by using AI both to add top-of-the-funnel leads and to improve the review time of each case[1]. He also said the company reviewed 8,786 qualified policies in the first quarter. But the company did not say when it began using AI, so the first quarter cannot be treated as a pre-adoption baseline, and the company did not attribute the difference between the two quarters to AI. It also did not split how much AI contributed to leads versus review time.

From review volume to revenue: the unquantified links

Review volume determines how large a pool of policies the company can choose from. A life settlements business first has to buy suitable policies, and how many it can buy depends first on how many policies the front end can see and how quickly it can get through them. With higher review throughput, the same team can reach more purchasable policies without loosening its standards. Purchase volume may rise as a result, and revenue may follow.

The financial metric at the end of this path is life settlements revenue, which is driven by the scale of policy purchases. The company has not yet tied review volume to that metric with numbers. Neither link, from review volume to purchase volume or from purchase volume to revenue, has a public figure. Review volume is an operating throughput measure, not a financial result. Reviewing more policies does not mean buying more policies: policy supply, deployable capital and pricing discipline all affect how many policies the company ends up buying.

What is confirmed and what is not

What can be confirmed today is that AI is running at the very front of Abacus's policy origination and has produced a review volume that can be counted. The constraint on how many policies the front end can see, and how fast it can review them, has loosened.

Two things have not been separately quantified. The first is AI's independent contribution to review volume. The second is how many of the additional reviewed policies became actual purchases. Only when the company discloses the number of purchases, or the capital deployed, that came from these reviewed policies in the same period will it be possible to judge whether higher review volume has carried through to revenue.

Application assessment

• AI Policy Origination Review | Business position: core operations | Application stage: limited production | Coverage: single business unit | Value type: revenue growth

Sources

[1] Drillr · Abacus Global Management, Inc. (ABX) · 2026-08-06 · Earnings call

Quote: In Q2, we have been able to review 9,314 qualified policies as compared to 8,786 qualified policies in Q1, with total policies reviewed year-to-date, including non-qualified, reaching over 50,000, a milestone we've been able to achieve by augmenting both top-of-the-funnel leads in our review time of each case with artificial intelligence.