ALL · The Allstate Corporation

Insurance - Property & Casualty · mkt cap $54.2B · calls: Q1 FY2026 vs Q4 FY2025

53.0 conviction · conf-adj 53

conf 2/10 partial

enthusiasm:24.0 · trend:8 · quantifies:0 · impact:0 · under_radar:5 · credibility:5 · business_impact:8 · disruption:0 · commitment:0 · confirmation:3

Enthusiasm latest 8 / prev 6 (rising)

Allstate's AI thesis shifted from claims, telematics, pricing sophistication, and predictive models in Q4 FY2025 to a broader Q1 FY2026 story around ALLIE, generative AI expense reduction, agentic AI decisioning, agent productivity, and AI-driven direct sales. The credibility is moderate: management gives concrete operational examples and some deployment/data scale, but does not provide revenue, margin, headcount, or dollar savings from AI.

GROUNDED NEXT-FY IMPACT vs CONSENSUS

Grounded on actual base — revenue $66.5B · net income $10.3B · net margin 15.5% · diluted EPS 38.19

These are next-fiscal-year annual uplift estimates, not next-quarter numbers.

Aggregate next-FY est. rev uplift: % · next-FY EPS uplift: % · vs analysts: unclear · priced in: medium · confidence: 2/10

Claim	Figure	Arithmetic
AI handles millions of emails productivity · soft	millions of e-mails	Volume metric only. No cost-per-email, FTE count, wage base, or opex base disclosed anywhere in the inputs, so no $ saving can be computed without inventing the base. after_tax_saving requires a saving_$ that does not exist here against the $10.282B net-income base.
AI selling directly, live in 3 states revenue · soft	3 states	Geographic deployment footprint, not a disclosed policy count or premium amount. Management explicitly frames it as 'more of a learning'. No revenue base per state disclosed -> cannot size incremental revenue versus $66.460B revenue or EPS flow-through.
Telematics asset >2 billion miles other · soft	over 2 billion miles	Cumulative data-asset stock, not a flow. No disclosed premium lift, loss-ratio improvement, retention lift, or cost saving attached. Underpins pricing/claims edge but unmonetized in the disclosure -> no arithmetic bridge to $66.460B revenue or $10.282B net income.
Tracks 50M cars every 15 seconds daily other · soft	50 million cars/day	Operational telemetry throughput, data-collection scale only. No disclosed monetization rate, underwriting margin improvement, or claims saving, so no financial uplift can be calculated.

Assumptions: Used current revenue $66.460B, current net income $10.282B, current net margin 15.47%, default 21% tax rate and incremental net margin = current 15.47% where applicable. None of the four quantified claims carries a dollar amount, a percentage, or a base obtainable from the inputs, so no after-tax saving, incremental revenue, or phasing could be derived. All claims are adopter-side (AI improving Allstate's own pricing, distribution, claims, and service costs); zero supplier-side AI revenue. EPS% would have been sized against consensus adjusted EPS (FY25 epsAvg 30.46) had any dollar figure existed.

Top line: No quantifiable next-FY topline uplift is supportable. The only top-line-tagged claim (AI selling in 3 states) is an early-stage pilot Allstate itself calls 'more of a learning', with no disclosed policy count or premium per policy. Telematics (>2B miles) and 50M-car tracking are data assets supporting pricing accuracy but carry no disclosed revenue figure. On a $66.46B revenue base, nothing here moves the line in a way the math can size.

Bottom line: No quantifiable bottom-line uplift. Email automation and forward-looking claims-reserving/agentic-AI commentary point at real productivity/loss-cost savings, but management gave only volume counts and qualitative 'middle innings' language — no opex base, FTE count, or savings $ to run after_tax_saving = saving_$ * 0.79 against the $10.28B net-income base. Genuinely promising operationally but unsized in this disclosure.

[impact n/m (all claims soft/unanchored)] Consensus FY25 sits at $58.98B revenue / $30.46 EPS / ~$7.71B NI, already well below the disclosed actual base ($66.46B / $38.19 / $10.28B). With zero dollar-anchored AI figures, there is no computable AI increment to place above or below the consensus trajectory — the program cannot be shown to be either priced in or mispriced from these numbers. Default to medium: the operational scale (2B miles, 50M cars) is real and partly reflected in current results, but no incremental AI dollars are quantified to argue an upside surprise.

MODEL CONSENSUS (impact)

partial

Near-total agreement: all four claims unanchored/soft, adopter-side, no computable uplift. Took conservative confidence and explicit 0 supplier revenue.

Conflicts reconciled

supplier_rev_uplift_pct: X=0 vs Y=null -> used 0 because both confirm pure adopter-side with explicitly zero supplier AI revenue
confidence: X=3 vs Y=2 -> used 2, the more conservative value given no calculable AI increment

Field	Opus 4.8	GPT-5.5
Rev uplift %	–	–
EPS uplift %	–	–
Priced in	medium	medium
vs analysts	unclear	unclear
Confidence	2	3
Top line	No quantifiable topline uplift. The only top-line-tagged claim (AI selling in 3 states) is an early-stage pilot Allstate itself calls 'more of a learning' with no disclosed premium contribution. Telematics (>2B miles) and 50M-car tracking are data assets that support pricing accuracy but carry no disclosed revenue figure. On a $66.46B revenue base, nothing here moves the line in a way the math can size.	No quantified next-FY revenue uplift is supportable. The direct-selling claim is live in 3 states and closing policies, but without policy count or premium per policy, incremental revenue cannot be sized versus $66.460B revenue.
Bottom line	No quantifiable bottom-line uplift. The email-automation and forward-looking claims-reserving/agentic-AI commentary point at real productivity/loss-cost savings, but management gave only volume counts and qualitative 'middle innings' language — no opex base, FTE count, or savings $ to run after_tax_saving = saving_$ * 0.79 against the $10.28B net-income base. Genuinely promising operationally but unsized in this disclosure.	No quantified EPS uplift is supportable. AI-handled emails imply service productivity, and telematics/tracking may improve underwriting or claims, but none disclose savings or profit improvement to compare with $10.282B net income.
Reasoning	Consensus FY25 sits at $58.98B revenue / $30.46 EPS / $7.71B NI, already well below the disclosed actual base ($66.46B / $38.19 / $10.28B). With zero dollar-anchored AI figures, there is no computable gap to place above or below the consensus trajectory — the AI program cannot be shown to be either priced in or mispriced from these numbers. Default to medium: the operational scale (2B miles, 50M cars) is real and partly reflected in current results, but no incremental AI dollars are quantified to argue an upside surprise.	Consensus estimates provided run from 2023 to 2025, ending at 2025 revenue of $58.980B and EPS of $30.46004 versus the supplied 2025 actual base of $66.460B revenue and $38.19 diluted EPS. Because the AI claims contain no calculable next-FY dollar uplift, there is no measurable AI increment to compare against consensus growth or the actual 2025 outperformance.

Rows highlighted where the two models disagreed.

QUANTIFICATIONS

AI-handled emails: millions of e-mails (not specified, bottomline)
“We do a bunch of it does millions of e-mails for us, people want to spend time doing it.”

AI direct-selling deployment: 3 states (current, topline)
“The AI can also just sell directly. And we're live in the market doing that right now on a particular product. It's more of a learning, but it's doing it in 3 states. It's closing policies.”

telematics data asset: over 2 billion miles of data (current cumulative, both)
“So we've been in telematics. We now have over 2 billion miles of data.”

vehicle tracking data: 50 million cars every day, every 15 seconds (current, both)
“And then we track 50 million cars every day, every 15 seconds.”

PAST (realized)

So we've been in telematics. We now have over 2 billion miles of data.
Increased the sophistication and precision of pricing models, enabling more accurate pricing.
Marketing investment increased to $2.1 billion in 2025, up from $900 million in 2019, enabling us to effectively reach more consumers with a more competitive price and better customer experience.
We do a bunch of it does millions of e-mails for us, people want to spend time doing it.
And we have to get them the money. So if it goes down that river, lots of people dip into the river and, like, take money out.

CURRENT (now)

The system enables the use of artificial intelligence to improve customer experience and lower costs.
We're leveraging this technology platform in building Allstate's Large Language Intelligent Ecosystem, which we call ALLIE, to harness the power of a genic AI.
These marketing analytics that are terrific, they enable us to price lead purchases in real time, determine effectiveness of programs by media channel and message.
The claims team is using a massive amount of data to assess the effectiveness of controlling severity and executing the claims function.
And then we track 50 million cars every day, every 15 seconds.
And we're live in the market doing that right now on a particular product.
It's closing policies.

FORWARD (guidance)

I would say one of the things we're hopeful about is with advanced computing power that we can increasingly get more specific on what's in the reserves.
We think with advanced analytics, we may be able to get another angle just looking at all the individual cases, which is really complicated.
The opportunities we see continue to get bigger.
I think the real benefit from this will come from a agentic AI, where agents are talking to agents and making decisions in subsecond real-time response rate that people then can't compete with you.
We think it offers potential to really build off of what we did in transform to growth.
So I would put it middle innings with the later innings probably be where being where you really see the benefit of artificial intelligence and the insights and tools that we can use to improve in the claims organization.
We think it will remove a lot of service work out of agents offices.
Secondly, it will help them be smarter and on behalf of agents who provide more advice and do less individual work.

TRACK RECORD — PROMISE vs DELIVERY

63/100 track record delivers 6 calls reviewed

Allstate rarely sets hard number-plus-date AI targets; most AI/automation talk (ALLIE, generative AI in claims/coding, predictive injury models) is reported as current-state progress without an attached quantified goal. The one clearly quantified analytics-driven goal it did set, SAVE's 25M-interactions-in-2025 target, was comfortably beaten (46M by Q3), so on thin evidence management has delivered on what it quantified.

SAVE program goal of improving 25 million customer interactions in 2025 (up from 20M+ in 2024) via analytics-driven coverage/discount optimization and proactive outreach — promised Q1 FY2025
delivered By Q3 2025 reported 46 million interactions improved and 5M+ customers helped to cut premiums >5%; by Q4 SAVE had reduced 7.8M customers' premiums by 17% on average — target clearly beaten

Build out 'applied AI' / ALLIE agentic-AI ecosystem to lower costs and improve service (AI reviewing adjuster emails, ~15% of coding by AI) — promised Q3 FY2025
too-early Described as current-state usage with no attached numeric cost/revenue target or deadline; reaffirmed in Q4 2025 and Q1 2026 without quantification, so not a judgeable promise

Use predictive/precision claims models to identify injured parties earlier and control loss costs, supporting affordability — promised Q4 FY2025
too-early Q1 2026 showed favorable prior-year reserve releases and auto underlying combined ratio improving to 89.5, consistent with model benefits, but no specific numeric model-attributed target was set

PRICED-IN (REFINED)

MEDIUM

Est. revisions rising · Fwd P/E 19.0 · EV/Sales 0.9x

AI claim maps to Property Liability, Protection Services, Allstate Health And Benefits

Consensus EPS ramps sharply across forward fiscal years, and the last-quarter price target average of 264 is above the last-year average of 236.88, so estimates/targets imply rising expectations even though rating counts have not migrated upward recently. A rising estimate trend makes AI upside more priced-in, not less, but valuation is only moderately stretched at 19.0x forward earnings and 0.9x EV/sales with low TTM P/E context. AI efficiency would most plausibly flow through Property Liability first, with secondary relevance to Protection Services and Allstate Health And Benefits. The mixed setup supports a medium priced-in verdict rather than high.

COVERAGE — ENTHUSIASM TRAJECTORY + CATALYSTS

3Q4 FY20243Q1 FY20255Q2 FY20259Q3 FY20254Q4 FY20254Q1 FY2026

AI enthusiasm across 6 calls — trend ↗ rising

AI moved from vague analytics and digitization to explicit GenAI/agentic use cases, then receded into broader analytics-enabled growth.

RECENT AI CATALYSTS & NEWS

2026-04-01 · 5 Momentum Stocks to Buy for April After a Capricious March

BUSINESS IMPACT - QUALITATIVE MATERIALITY

7/10 qualitative impact material medium-term · mixed evidence

Where AI matters: pricing, claims, distribution, service cost

Allstate is applying AI to core P&C levers: pricing precision, telematics, claims severity control, customer service automation, agent productivity, and early direct policy sales. The scale of data and deployed workflows is real, but management has not tied AI to dollars of premium growth, loss-ratio improvement, or expense savings.

Caveats: No disclosed Rev/EPS contribution from AI initiatives; Agentic AI execution and integration complexity; Regulatory scrutiny of AI pricing, underwriting, and claims decisions; Autonomous driving could reduce long-term personal auto premium volume

AI DISRUPTION / CANNIBALIZATION RISK tailwind · 2/10

AI does not automate away insurance risk-taking, regulatory capital, underwriting, or claims obligations; it mostly improves how those functions are priced and serviced. The main AI-adjacent threat is autonomous driving reducing personal auto loss/premium pools over time, but Allstate can reprice, shift coverage, and use its own telematics data to adapt.

OPTIONS / MARKET STRUCTURE

option liquidity: good

ATM IV

—

TYPICAL BID-ASK

—

OPEN INTEREST

—

proxy inputs — dollar-ADV $314M · beta 0.207 · px $210.46

source: proxy (no options chain on FMP)
FMP /stable/ exposes no options-chain endpoint on this key, so ATM IV, bid-ask spread and open interest are unavailable. Liquidity below is a PROXY from dollar-ADV, beta and price level (a stand-in for option depth), not measured option-market data.

CONFIRMATION — INSIDERS · 13F · LANGUAGE

Mixed — insiders selling, institutions adding, management language 5/10 measured.

INSIDERS selling 13 open-market sell(s) vs 0 buy(s) — net distribution

INSTITUTIONS (13F) adding as of 2026-03-31: 155 new / 168 closed positions; 795 increased / 547 reduced; institutional ownership +0.65pp; -4 net 13F holders

MGMT LANGUAGE 5/10 measured AI discussion is brief: real ownership of an internal platform, but no figures, timelines, rollout detail, or demonstrated impact.

commit “The system enables the use of artificial intelligence to improve customer experience and lower costs.”

commit “We're leveraging this technology platform in building Allstate's Large Language Intelligent Ecosystem, which we call ALLIE”

hedge “to harness the power of a genic AI.”

VERBATIM AI QUOTES

“The system enables the use of artificial intelligence to improve customer experience and lower costs.”
— Jesse Merten, Q1 FY2026

“We're leveraging this technology platform in building Allstate's Large Language Intelligent Ecosystem, which we call ALLIE, to harness the power of a genic AI.”
— Jesse Merten, Q1 FY2026

“The -- but -- so I think it's hard. So what I can say is that the -- from our standpoint, our capabilities continue to grow exponentially.”
— Thomas Wilson, Q1 FY2026

“The opportunities we see continue to get bigger.”
— Thomas Wilson, Q1 FY2026

“The easiest way thing to do is generative AI, which is, I think last time I called it the you might remember [ Ken ] sneakers.”
— Thomas Wilson, Q1 FY2026

“It's good. It cuts out expenses, you can cut out call center people.”
— Thomas Wilson, Q1 FY2026

“We do a bunch of it does millions of e-mails for us, people want to spend time doing it.”
— Thomas Wilson, Q1 FY2026

“I think the real benefit from this will come from a agentic AI, where agents are talking to agents and making decisions in subsecond real-time response rate that people then can't compete with you.”
— Thomas Wilson, Q1 FY2026

“We're building that.”
— Thomas Wilson, Q1 FY2026

“We think it offers potential to really build off of what we did in transform to growth.”
— Thomas Wilson, Q1 FY2026

“The AI can also just sell directly.”
— Thomas Wilson, Q1 FY2026

“And we're live in the market doing that right now on a particular product.”
— Thomas Wilson, Q1 FY2026

“It's closing policies.”
— Thomas Wilson, Q1 FY2026

“These processes are being enhanced by optimizing the method of inspection, focusing on adjuster training, and using advanced computing capabilities.”
— Jesse Edward Merten, Q4 FY2025

“Utilizing new tools and quality assurance processes to enhance claim handling, predictive models are also being used to identify potentially injured parties earlier in the process to resolve claims promptly and control liability.”
— Jesse Edward Merten, Q4 FY2025

“That doesn't mean we fully implemented all of the AI-enabled technologies.”
— Jesse Edward Merten, Q4 FY2025

“This is proprietary to The Allstate Corporation.”
— Jesse Edward Merten, Q4 FY2025

“So I would put it middle innings with the later innings probably be where being where you really see the benefit of artificial intelligence and the insights and tools that we can use to improve in the claims organization.”
— Jesse Edward Merten, Q4 FY2025

“So on autonomous driving, I would say it's a curveball we've been watching for about fifteen years.”
— Thomas Joseph Wilson, Q4 FY2025

“So we've been in telematics. We now have over 2 billion miles of data.”
— Thomas Joseph Wilson, Q4 FY2025

ANALYST QUESTIONS ON AI

Q (Q1 FY2026, Taylor Scott): Second question, I actually want to circle back on artificial intelligence, specifically and I know you guys have had a strategy over time to improve the expense ratio, so you could get even more competitive in the market and spur some growth. Could you talk about how I expand on that, what you're planning to do? And sort of how you see yourself positioned relative to some of your peers, one of which I think has begun to roll it out more aggressively and reduce their workforce more.
A: from our standpoint, our capabilities continue to grow exponentially. The opportunities we see continue to get bigger. And we're figuring out how to address and deal with some of the implementation and deployment issues because it's not simple. I can't tell you that it's all in market today. It's -- stuff is complicated. But if you can pull it off, it works really well. The easiest way thing to do is generative AI... It cuts out expenses, you can cut out call center people... We do a bunch of it does millions of e-mails for us... I think the real benefit from this will come from a agentic AI, where agents are talking to agents and making decisions in subsecond real-time response rate that people then can't compete with you. We're building that.

Q (Q1 FY2026, Pablo Singzon): I wanted to shift to AI again, but this time as it relates to your distribution strategy and how you reach customers. I presume it helps direct distribution, but how do you think it affects your agents, whether captive or independent, there's an argument that it makes them productive, but do you think Issues away from them?
A: AI can help them in a whole bunch of ways. First, it can help us have a better product. and better pricing and deliver better service for people. That's in general, just -- it will help us be a better company. Secondly, as it relates to their specific work, we think it will remove a lot of service work out of agents offices... Secondly, it will help them be smarter and on behalf of agents who provide more advice and do less individual work... The AI can also just sell directly. And we're live in the market doing that right now on a particular product. It's more of a learning, but it's doing it in 3 states. It's closing policies.

Q (Q4 FY2025, Bob Huang): Autonomous driving, it's been an increasingly more topical discussion. Just curious on your thing like, your view on the pace of the technological development there and then how that could potentially impact personal auto just from a like, is it more of a threat? Is it more of an opportunity? How you're thinking about it? How you're positioning it?
A: So on autonomous driving, I would say it's a curveball we've been watching for about fifteen years. And that's a good thing because we've been at it for fifteen years. So we've been in telematics. We now have over 2 billion miles of data. You need that kind of data to be able to adjust to what autonomous driving can do what different cars can do... feel like there's as long as we're ahead in pricing, we're very sophisticated, we're involved in telematics, we're watching the data, and then we'll be fine.

Q (Q4 FY2025, Michael Zaremski): Maybe high level, you know, you've been working on auto claim process improvements for many years now. Trying to understand, you know, what base bonding are we in? You know, AI, I'm sure, is helping it. Is it will the benefits are they are is The Allstate Corporation more of a first mover, or this is proprietary to you all? Or, you know, are you using third parties and industry will eventually catch up over time?
A: Certainly not early innings, but not done... That doesn't mean we fully implemented all of the AI-enabled technologies... This is proprietary to The Allstate Corporation. We're not leveraging third-party insights or technology... So I would put it middle innings with the later innings probably be where being where you really see the benefit of artificial intelligence and the insights and tools that we can use to improve in the claims organization.