The Q2 2026 call is where CVS Health put AI into its operations by name, in answer to an analyst asking for the timing and magnitude of technology investment and how material the cost savings and revenue would be. Management listed uses rather than amounts: an AI claims advisor cutting processing time on complex claims by over , agentic AI on the Aetna and CVS Caremark call center platform, advocate case preparation down from minutes to , hundreds of millions of pharmacy calls moved to conversational AI and pharmacist hours refocused, over pages of clinical records analyzed, and AI in specialty adherence. The 10-Q again has no AI passage.
The one dollar figure is the CFO's: operating expense savings over the last few years of more than , from technology efficiencies and AI together. It was generated over the last few years and probably includes the 2024 restructuring plan, expected to save over in 2025, which the company does not attribute to AI. The ledger reads it as cumulative and converts it by its fixed cumulative-years band: the quarter's share is at the point, and with the restructuring savings taken out is left as a loose ceiling for AI across the operating expense savings channels, whose high ends are set to sum below it. The other figure, a commitment of more than to technology over a decade, is a technology total that may include capital spending.
Insurance segment operating expenses rose against segment revenues up , running against their prior-year share, a small move in the direction the claims predict that the 10-Q explains by increased business investments without naming AI. Pharmacy segment operating expenses, without the prior-year litigation charge, rose against revenues up , against the claims, with Rite Aid volume named. Every size is the ledger's own, each on a range that starts at or near zero: claims at , member and provider service at , prior authorization at . Internal AI spend is named as technology and AI together and is left unsized, with the commitment's pace as its ceiling. Health100 began its targeted launch after the quarter ended and is left unsized.
The claims processing, member and provider service, pharmacy call, clinical record review and adherence channels opened this quarter; the call carried more AI passages than Q1, short of the doubling the ledger records as a step. Prior authorization, clinical record review, navigation and adherence are read as relabelled: each capability predates LLMs at the company and no AI-attributed movement is measured. Adherence credits AI beside technology and automation with nothing to separate its part, so it is left unsized; the pharmacy call and clinical record channels give only cumulative counts of AI work with no period, which read directional and are left unsized.
Sized channels against the income statement, Q2 2026
5 of 10 channels sized
Each blue mark is one channel's dollars for the quarter; a bar is the range of an estimate. Grey marks are the company's reported lines. The distance between them is the point: how large the AI channel is next to the line it sits in.
Total revenuesreported line
$106.10bn
Total operating costsreported line
$101.39bn
Cost of products soldreported line
$58.86bn
Health care costsreported line
$31.48bn
Operating expensesreported line
$11.05bn
AI model, cloud AI and conversational AI vendor billspend · our inference
Member and provider service cost displaced by agentic AIcost displaced · our inference
Medical cost avoided through AI-guided member navigationcost displaced · our inference
Claims processing cost displaced by the AI claims advisorcost displaced · our inference
Prior authorization handling cost displaced by AI and automationcost displaced · our inference
$100k$1mn$10mn$100mn$1bn$10bn$100bn$1000bn
AI channel, dollars for the quarter low to high of an estimate reported lineLog scale: each gridline is ten times the one before.
New money and old money, Q2 2026
6 expanded4 relabelled
Each channel is tagged once for whether its money existed before language models, from the company's annual report and call at the start of the period. A bar splits one flow's sized dollars by that tag. The incremental total is the part that would not be there without the models: a new channel counts in full, an expanded one only for what AI added, a relabelled one at zero.
Flow, sized total
Split by novelty
Incremental total
Paid for AI$908k to $21mn sized
expanded $908k to $21mn
Incremental total $0 to $21mnpoint $0$6.0mn in 1 channel has no traced baseline
Cost displaced by AI$0 to $59mn sized
expanded $0 to $22mnrelabelled $0 to $37mn
Incremental total $0 to $22mnpoint $9.7mn
The sized total counts every channel the company credits to AI, including relabelled money that existed before language models and the ledger's own estimates for it. The incremental total counts relabelled channels at zero. A flow is split by layer where its dollars sit at more than one: end use, compute sold to builders, and hardware. The same dollar can be a buyer's spend, a cloud's revenue and a chipmaker's revenue, so the layers are never added together.
Paid for AI2 channels · $908k to $21mn sized · $0 to $21mn incremental · 1 not sized
vendor bill
AI model, cloud AI and conversational AI vendor bill
0% to 0.02% of the quarter’s revenue
Incremental total: counts at zero at the point and in full at the high end, with no traced baseline.
our inferencedisclosure: described· motive: exploratory· before LLMs: expanded
The quarter names AI products in use (the HIO assistant, agentic AI on the call center platform, conversational AI on pharmacy calls) without naming a vendor or a cost. The size shown is the ledger's own estimate, , an AI share of the technology part of the corporate line (). Outside AI vendor bills are counted here only; the internal build channel excludes them, so no dollar counts twice.
Evidence: 5 quotes, 3 figures, 3 confounds, 2 from before coverage
What the company pays outside vendors for AI: the Google Cloud AI technologies Health100 is built on, the model and platform bills under the HIO assistant, the agentic AI on the call center platform and the conversational AI that takes retail pharmacy calls. No vendor other than Google Cloud is named and nothing is split out of the operating expense lines. This channel holds every outside AI vendor bill; the internal build channel excludes them so that no dollar counts twice. The conversational AI part may run on speech technology that predates LLMs, and the anchor already shows investment in AI and cloud capabilities, so the channel is expanded and sized as a level; no quarter of the bill before AI can be traced, so its size lands in the undetermined amount.
Why this motive
Carried from Q1: the named vendor serves a platform still in launch (claim c1). The agentic call center AI and conversational pharmacy AI are in use (claims c24 and c15), but no vendor or cost is named for them, so the tells do not move the motive.
Before LLMs: expanded
At the anchor the 10-K already described concerted investments in emerging technology capabilities such as AI (anchor claim cvs-anchor-c1) and continued investment in cloud capabilities (anchor claim cvs-anchor-c2), with no amount and no vendor named; information technology sat inside operating expenses, consolidated for FY2024, with no line of its own. No quarter of the outside AI and cloud AI bill before AI can be traced. The size is the whole of an activity that existed before.
“The Company is making concerted investments in emerging technology capabilities such as artificial intelligence (“AI”) to further automate, augment and improve its operational capabilities, and to improve the experience for providers, patients, and consumers.”
“There is continued focus and investment in enterprise data platforms, cloud capabilities, digital products to offer innovative solutions and a seamless experience to the Company’s members through mobile and web channels.”
Corporate/Other segment operating expenses (holds information technology, digital, data and analytics) · 2026-CQ2
Corporate/Other segment operating expenses, prior year · 2025-CQ2
Technology investment committed over the next decade, a floor (includes AI; may include capital spending, an inference from the anchor capex mix; stated as a total) · as-of 2026-08-05
What else could explain it
line composition: The corporate line holds executive management, legal, compliance, human resources, finance and integration costs beside technology.
other: Technology spend inside the operating segments is not in the base, so the base may understate the technology budget; part of the AI bill may also be capitalized inside the technology commitment.
relabel: The conversational AI on pharmacy calls may run on speech technology that predates LLMs, and the anchor shows investment in AI and cloud capabilities; the channel is read as expanded, sized as a level with no traced baseline.
Quotes
“Last month, we began the targeted launch of our Health 100 platform, including HIO, our new AI-powered assistant. HIO is designed to simplify the consumer experience and help people engage more effectively in their care journey. Early feedback has been encouraging, and we are excited to expand access later this year.”
“That said, we also made a commitment a year ago that we were going to invest $20 billion+ in our technology efforts over the next decade. This is going to be a very deliberate, and this is the consistent word you're hearing, a very responsible and deliberate process in terms of where we expect to get the appropriate ROI and the use cases that will help improve the experience and our ability to serve our customers.”
“Over the course of the last couple of years, we've removed hundreds of millions of calls out of our retail pharmacy business and transferred them into conversational AI.
This has allowed us to refocus 1 million hours of pharmacist hours to allow them to deliver better experiences and clinical care at the pharmacy counters.”
“CVS Health is deploying agentic AI to simplify and streamline call center interactions for members and providers engaging with Aetna® and CVS Caremark® businesses on a secure call center platform.”
“Adjusted operating loss increased $64 million, or 15.5%, in the three months ended June 30, 2026 compared to the prior year primarily driven by increased business investments.”
Internal investment in AI and technology capability
Not sized
The investment is named as technology and AI together, a big push in that area, inside a commitment to technology efforts over a decade, and AI and emerging technologies (claims c5, c8 and c18); nothing separates AI's part (the methodology rule for AI named beside another cause). The commitment's quarterly pace, at least , is quoted as the ceiling.
Matched line moved : the whole line, not this channel.
described, no sizedisclosure: described· motive: exploratory· before LLMs: expanded
An analyst asked again for the timing and magnitude of technology investment. The CEO restated a commitment made a year earlier to invest more than in technology over a decade, a pace of at least a quarter, a technology total that includes AI and may include capital spending (an inference from the anchor capital spending mix), and so is a ceiling on the AI part, not its level. AI is named beside technology and nothing separates its part, so the channel is left unsized.
Evidence: 8 quotes, 5 figures, 3 confounds, 1 from before coverage
Money the company spends building and rolling out AI itself: the AI share of what management calls a commitment of more than a stated sum to technology over a decade, the AI academy for Aetna leaders and colleagues, the staff building Health100 and the AI tools named on the calls. Much of the technology program may be capitalized (an inference: technology took most of capital spending at the anchor, at about the yearly pace of the commitment); this channel carries only the expensed part and is not a capital channel. Outside AI vendor bills are excluded and sit in ai-vendor-bill. The filings explain rising segment and corporate operating expenses by increased business investments without naming AI.
Why this motive
Asked again for the timing and magnitude of technology investment, management restated a decade-long technology commitment and described AI as a big push and a big investment, with no AI amount (claims c5 and c8): investing for later, the exploratory tell, as in Q1.
Before LLMs: expanded
At the anchor the 10-K already described concerted investments in emerging technology capabilities such as AI (anchor claim cvs-anchor-c1); gross capital expenditures were for FY2024, of them for technology, digital and other strategic initiatives. No AI amount was given then or in the covered quarters, and no quarter of the expensed part before AI can be traced. The size is the whole of an activity that existed before.
“The Company is making concerted investments in emerging technology capabilities such as artificial intelligence (“AI”) to further automate, augment and improve its operational capabilities, and to improve the experience for providers, patients, and consumers.”
Technology investment committed over the next decade, a floor (includes AI; may include capital spending, an inference from the anchor capex mix; stated as a total) · as-of 2026-08-05
Quarterly pace of the technology commitment (one fortieth of the decade total, a floor) · 2026-CQ2
Corporate/Other segment operating expenses (holds information technology, digital, data and analytics) · 2026-CQ2
Corporate/Other segment operating expenses, prior year · 2025-CQ2
Change in Corporate/Other operating expenses, year over year · 2026-CQ2
Reported line it is matched to
Corporate/Other operating expenses were against , up . The 10-Q explains the larger corporate loss by increased business investments, and the CFO said investment in capabilities and technology showed up in the corporate segment this quarter (claim c19); neither names AI.
2026-CQ2: 2025-CQ2: 2026-CQ2:
What else could explain it
line composition: Business investments are not split by kind, and the commitment the CEO restated is a technology total that may include capital spending (technology capital spending at the anchor ran at about the commitment's yearly pace, anchor figures cvs-anchor-f4 and cvs-anchor-f5).
relabel: AI has run in operations for years by the CEO's Q1 account; part of what is called AI investment may be the existing technology budget under a new name.
transformation program: The investment is named as technology and AI together (claims c5 and c18); the technology program is the other cause and is not split.
Quotes
“The tailwinds and the trajectory of that business, what we haven't talked about is the investments in technology and AI specifically, which is allowing us to focus on the things that are creating friction in the marketplace, and we're trying to improve the healthcare experience. A big push and a big investment in that area.”
“We're going to be very clear, though, given some of the concerns around AI, we're going to be very focused about where we will and where we will not deploy AI.”
“That said, we also made a commitment a year ago that we were going to invest $20 billion+ in our technology efforts over the next decade. This is going to be a very deliberate, and this is the consistent word you're hearing, a very responsible and deliberate process in terms of where we expect to get the appropriate ROI and the use cases that will help improve the experience and our ability to serve our customers.”
“We're committed to continuing the investment across businesses as I think about AI and emerging technologies, ultimately to improve the health of consumers, and we'll keep you updated.”
“We had some investments in capabilities and technology, and that continues. We have it across our business segments. We've been investing in colleagues and capabilities at PCW for some time, and we think we're seeing the fruits of that labor. We continue to do it. You saw some of it pop up in the corporate segment this past quarter.”
“I think Steve and Prem did a nice job of how we're prioritizing and the results that we're seeing, but I think you're going to see us balance two parts of technology and AI. One is the efficiencies and the productivity that we spoke to, but equally as important is the growth aspect of the investments that we're making.”
“Adjusted operating loss increased $64 million, or 15.5%, in the three months ended June 30, 2026 compared to the prior year primarily driven by increased business investments.”
Q1 2026described · described, no size · exploratory
Q2 2026described · described, no size · exploratory
Cost displaced by AI6 channels · $0 to $59mn sized · $0 to $22mn incremental · 2 not sized
back office · expensive to verify
Prior authorization handling cost displaced by AI and automation
0% to 0.01% of the quarter’s revenue
Incremental total: counts at zero.
Matched line moved : the whole line, not this channel.
our inferencedisclosure: described· motive: product-defensive· before LLMs: relabelled
The Aetna president put prior authorization among the AI uses, with approved in real time and over within a day, the same shares as in Q1. Segment operating expenses ran against their prior-year share of revenues, which the 10-Q explains by increased business investments without naming AI. The size shown is the ledger's own estimate, , for clinical review and intake; provider calls are read under member and provider service.
Evidence: 3 quotes, 4 figures, 3 confounds, 4 from before coverage
The administrative cost of handling prior authorizations at Aetna (clinical review and intake; provider calls about prior authorizations are read under member-provider-service-automation so that no dollar counts twice) that is lower because requests are approved in real time by technology management now groups with AI. Medical necessity decisions are costly to get wrong and regulated, so the work is expensive to verify. The approvals themselves move medical cost, which is not part of this channel.
Why this motive
Carried from Q1: provider friction and the industry commitment frame the automation. This quarter the Aetna president lists prior authorization among AI uses to build trust with providers (claim c13), which is the same tell; no cost line is shown to fall.
Before LLMs: relabelled
At the anchor the company already ran an online preauthorization tool and automated review of eligibility, refills, dosage and utilization in its pharmacy systems (anchor claims cvs-anchor-c4 and cvs-anchor-c5); the cost sat inside Health Care Benefits operating expenses, for FY2024. The real-time approval share is the same in both covered quarters and no AI part of it is measured.
“The Company’s NovoLogix® online preauthorization tool helps identify and capture cost savings opportunities for specialty drugs billed under the medical benefit by identifying outliers to appropriate dosages and costs, and helps to ensure clinically appropriate use of specialty drugs.”
“These systems also streamline the process by which prescriptions are processed by staff and network pharmacists by enhancing review of various items through automation, including plan eligibility, early refills, duplicate dispensing, appropriateness of dosage, drug interactions or allergies, over-utilization and potential fraud.”
“As we seek to reduce general and administrative expenses, we must balance the potential impact of cost-saving measures on our customers and other services and performances.”
“The Company expects benefits from ongoing enterprise-wide cost savings initiatives and investments in efficiencies, which aim to reduce the Company’s operating cost structure in a way that improves the consumer experience and is sustainable.”
Share of prior authorizations approved in real time · 2026-CQ2
Share of prior authorizations approved within a day, a floor · 2026-CQ2
Health Care Benefits operating expenses growth, year over year · 2026-CQ2
Health Care Benefits total revenues growth, year over year · 2026-CQ2
Reported line it is matched to
Health Care Benefits operating expenses were against , up while segment revenues rose : of revenues against , against the prior-year rate. The 10-Q attributes the rise to increased business investments and does not name AI.
line composition: The segment line carries selling expense, depreciation and amortization and all administration; prior authorization handling is not split.
relabel: Real-time approval predates LLMs (anchor claims cvs-anchor-c4 and cvs-anchor-c5), and the real-time share is the same as in Q1.
transformation program: The CFO's savings figure joins technology efficiencies and AI without a split (claim c17), and the window it covers holds the 2024 enterprise-wide restructuring plan, expected to save over in 2025 and not attributed to AI (anchor claim cvs-anchor-c13); none of it is credited to AI here.
Quotes
“We're utilizing these tools to help reimagine the prior auth process. For example, 83% of our prior authorizations are approved in real time and over 95% within 24 hours.”
“It's also showing up as real improvements in our operations. It's helping ensure we're best in class across the business segments. Over the last few years, as David said, we've generated actually some significant savings, over $1 billion in OPEX savings reductions. That's through the focus on both technology efficiencies, and ultimately, AI.”
“Operating expenses increased $139 million, or 3.1%, in the three months ended June 30, 2026 compared to the prior year primarily driven by increased business investments.”
Claims processing cost displaced by the AI claims advisor
0% to 0.01% of the quarter’s revenue
Incremental total: counts in full.
Matched line moved : the whole line, not this channel.
our inferencedisclosure: direction only· motive: exploratory· before LLMs: expanded
The release says the second-generation Claims Assist Manager, an AI-powered agentic claims advisor, cuts processing time by over for complex claims that need manual review; the CEO says it will do so on a very large annual claims volume. Processing time is not cost; segment operating expenses rose , slightly less than segment revenues, which leaves them against the prior-year rate. The size shown is the ledger's own estimate, . The counterparty is mixed: the anchor 10-K says claims processing is done either directly or through vendors (anchor claim cvs-anchor-c6).
Evidence: 5 quotes, 4 figures, 3 confounds, 5 from before coverage
The cost of processing medical claims at Aetna, in particular complex claims that need manual review, that is lower because the Claims Assist Manager, which management calls an AI-powered agentic claims advisor in its second generation, cuts processing time. Payment errors are costly and regulated, so the work is expensive to verify. Comparable with the claims-handling door at Progressive.
Why this motive
A second-generation tool launched in the quarter, with a processing-time floor and the cost effect stated as a future reduction (claims c2 and c25). Management aims at a best-in-class cost structure (claim c9), but no displaced line shrinks; with tells in conflict the less durable motive is taken, as at Progressive.
Before LLMs: expanded
At the anchor claims processing was one of the customer service operations the company ran directly or through vendors, on technology platforms for benefit administration and service operations (anchor claims cvs-anchor-c3 and cvs-anchor-c6), inside Health Care Benefits operating expenses of for FY2024. The agentic claims advisor changes the processing time per complex claim. The size is the change AI made, not the whole line.
“The Company is making concerted investments in emerging technology capabilities such as artificial intelligence (“AI”) to further automate, augment and improve its operational capabilities, and to improve the experience for providers, patients, and consumers.”
“The Health Care Benefits segment currently operates and supports an end-to-end suite of information technology platforms to support member engagement, enrollment, health benefit administration, care management, service operations, financial reporting and analytics.”
“Our ability to attract and retain customers and members is dependent upon providing compliant, cost effective, quality customer service operations (such as call center operations, PBM functions, retail pharmacy and LTC services, retail, mail order and specialty pharmacy prescription delivery, claims processing, customer case installation and online access and tools) that meet or exceed our customers’ and members’ expectations, either directly or through vendors.”
“As we seek to reduce general and administrative expenses, we must balance the potential impact of cost-saving measures on our customers and other services and performances.”
“The Company expects benefits from ongoing enterprise-wide cost savings initiatives and investments in efficiencies, which aim to reduce the Company’s operating cost structure in a way that improves the consumer experience and is sustainable.”
Reduction in processing time for complex claims that need manual review, from the AI claims advisor (a floor) · 2026-CQ2
Health Care Benefits operating expenses growth, year over year · 2026-CQ2
Health Care Benefits total revenues growth, year over year · 2026-CQ2
Operating expense savings over the last few years from technology efficiencies and AI together, a floor, cumulative with an unstated start (probably including the 2024 enterprise-wide restructuring plan, cvs-anchor-f6) · as-of 2026-06-30
Reported line it is matched to
Health Care Benefits operating expenses were against , up while segment revenues rose : of revenues against , against the prior-year rate. The 10-Q attributes the rise to increased business investments and does not name AI.
line composition: The segment line carries selling expense, depreciation and amortization and all administration; claims operations are not split.
transformation program: The CFO's savings figure joins technology efficiencies and AI without a split (claim c17), and the window it covers holds the 2024 enterprise-wide restructuring plan, expected to save over in 2025 and not attributed to AI (anchor claim cvs-anchor-c13); none of it is credited to AI here.
other: A first-generation claims advisor existed before coverage; what the second generation adds over the first is not given.
Quotes
“We've recently launched a new AI-enabled Claims Assist Manager, which will reduce our processing time by over 20% and accelerate payment for providers on hundreds of millions of claims every year.”
“First, of course, we're deploying AI to create a more efficient, faster, more accurate environment, particularly related to our claims payment and some of the transactional work that we do. David mentioned the Claims Assist Manager, which is revolutionizing how we provide payments more accurate and more timely to providers, which is really important. We're also using it to work towards a best-in-class cost structure that's very important to this business as we move into more efficient environments.”
“It's also showing up as real improvements in our operations. It's helping ensure we're best in class across the business segments. Over the last few years, as David said, we've generated actually some significant savings, over $1 billion in OPEX savings reductions. That's through the focus on both technology efficiencies, and ultimately, AI.”
“Aetna launched its second generation Aetna Claims Assist Manager (“CAM”), an AI-powered agentic claims advisor platform designed to streamline claims processing and improve payment accuracy. CAM reduces processing time by over 20% for complex claims that require manual review, helping providers get paid faster and more consistently.”
“Operating expenses increased $139 million, or 3.1%, in the three months ended June 30, 2026 compared to the prior year primarily driven by increased business investments.”
Member and provider service cost displaced by agentic AI
0% to 0.01% of the quarter’s revenue
Incremental total: counts in full.
Matched line moved : the whole line, not this channel.
our inferencedisclosure: direction only· motive: exploratory· before LLMs: expanded
The release says agentic AI is being deployed on the call center platform for Aetna and CVS Caremark members and providers; the Aetna president says case preparation for an Aetna One Advocate fell from minutes to , a cut of , with the time going to members, and that AI agents schedule appointments. No volume or cost is given. The size shown is the ledger's own estimate, . The counterparty is mixed: the anchor 10-K says call center operations are run either directly or through vendors (anchor claim cvs-anchor-c6), so the displaced cost is part staff and part vendors.
Evidence: 7 quotes, 5 figures, 3 confounds, 6 from before coverage
The cost of serving members and providers who contact Aetna and CVS Caremark (call center agents, some through vendors, including provider calls about prior authorizations, and the Aetna One Advocate staff who prepare member cases) that is lower because agentic AI handles call center interactions and cuts case preparation time. Comparable with the support door at Airbnb. Retail pharmacy calls are a separate channel.
Why this motive
Agentic AI is being deployed on the call center platform (claim c24), and the one measured task saving is said to give advocates more time with members rather than reduce staff (claim c10): a deployment under way with no displaced line, the exploratory tell.
Before LLMs: expanded
At the anchor call center operations were customer service operations run directly or through vendors (anchor claim cvs-anchor-c6), and an anchor-call executive named a drug category as one of the highest drivers of call volume (anchor claim cvs-anchor-c15); the cost sat inside Health Care Benefits operating expenses of and Health Services operating expenses of for FY2024. Agentic AI changes how many interactions reach a person and how long a case takes. The size is the change AI made, not the whole line.
“The Company is making concerted investments in emerging technology capabilities such as artificial intelligence (“AI”) to further automate, augment and improve its operational capabilities, and to improve the experience for providers, patients, and consumers.”
“The Health Care Benefits segment currently operates and supports an end-to-end suite of information technology platforms to support member engagement, enrollment, health benefit administration, care management, service operations, financial reporting and analytics.”
“Our ability to attract and retain customers and members is dependent upon providing compliant, cost effective, quality customer service operations (such as call center operations, PBM functions, retail pharmacy and LTC services, retail, mail order and specialty pharmacy prescription delivery, claims processing, customer case installation and online access and tools) that meet or exceed our customers’ and members’ expectations, either directly or through vendors.”
“As we seek to reduce general and administrative expenses, we must balance the potential impact of cost-saving measures on our customers and other services and performances.”
“The Company expects benefits from ongoing enterprise-wide cost savings initiatives and investments in efficiencies, which aim to reduce the Company’s operating cost structure in a way that improves the consumer experience and is sustainable.”
“But this category alone is driving obviously significant costs for our clients, and it's also driving significant expense within our organization just to support what is now one of our highest drivers of call volume around people trying to find access to the product and making sure that we get consistent supply in the market.”
Minutes an Aetna One Advocate spent preparing a member case before AI · 2026-CQ2
Minutes an Aetna One Advocate spends preparing a member case with AI · 2026-CQ2
Reduction in case preparation time for an Aetna One Advocate · 2026-CQ2
Health Services segment operating expenses · 2026-CQ2
Health Services segment operating expenses, prior year · 2025-CQ2
Reported line it is matched to
Health Care Benefits operating expenses were against , up while segment revenues rose : of revenues against , against the prior-year rate. The 10-Q attributes the rise to increased business investments and does not name AI.
line composition: Neither segment line splits member or provider service; Health Services operating expenses fell on the absence of a prior-year litigation charge (claim c29).
transformation program: The CFO's savings figure joins technology efficiencies and AI without a split (claim c17), and the window it covers holds the 2024 enterprise-wide restructuring plan, expected to save over in 2025 and not attributed to AI (anchor claim cvs-anchor-c13); none of it is credited to AI here.
other: The advocate measure covers a single task for a single role, and the time is said to go back to members.
Quotes
“First, of course, we're deploying AI to create a more efficient, faster, more accurate environment, particularly related to our claims payment and some of the transactional work that we do. David mentioned the Claims Assist Manager, which is revolutionizing how we provide payments more accurate and more timely to providers, which is really important. We're also using it to work towards a best-in-class cost structure that's very important to this business as we move into more efficient environments.”
“Our Aetna One Advocate, for example, used to spend 90 minutes preparing and reviewing a case to help a member. Now it takes them only two minutes. They can spend more time now with the member.”
“Improving the member experience. We can now help them along their care pathway. We can actually use our AI agents to help schedule appointments, creating less administrative burden on everyone.”
“It's also showing up as real improvements in our operations. It's helping ensure we're best in class across the business segments. Over the last few years, as David said, we've generated actually some significant savings, over $1 billion in OPEX savings reductions. That's through the focus on both technology efficiencies, and ultimately, AI.”
“CVS Health is deploying agentic AI to simplify and streamline call center interactions for members and providers engaging with Aetna® and CVS Caremark® businesses on a secure call center platform.”
“Operating expenses increased $139 million, or 3.1%, in the three months ended June 30, 2026 compared to the prior year primarily driven by increased business investments.”
“Operating expenses decreased $236 million, or 20.2%, in the three months ended June 30, 2026 compared to the prior year primarily driven by the absence of a $291 million legacy litigation charge recorded in the prior year.”
Retail pharmacy call handling moved to conversational AI
Not sized
The only measures are counts of AI work over an unstated span: calls moved and pharmacist hours refocused, , over the course of the last couple of years (claim c15), with no period or start date and no cost (the methodology rule for counts and cumulative counts with no period).
Matched line moved : the whole line, not this channel.
described, no sizedisclosure: direction only· motive: product-defensive· before LLMs: expanded
The group president says hundreds of millions of retail pharmacy calls moved to conversational AI over the last couple of years, refocusing pharmacist hours on clinical care. Those are cumulative counts of the work with no start date, which read directional and size nothing, and none is a cost. The channel is left unsized.
Evidence: 5 quotes, 3 figures, 4 confounds, 4 from before coverage
Pharmacist and pharmacy staff time that used to go to phone calls into retail pharmacies and that conversational AI now takes. Management says the calls moved to conversational AI and the hours were refocused on clinical care at the counter, not that payroll fell, so the money is staff time that may be redeployed rather than cost removed.
Why this motive
Calls moved to conversational AI and the freed hours were refocused on clinical care and better service, with store service scores credited partly to AI (claims c6, c15 and c16). The efficiency tell (a displaced line shrinking) is not met: segment operating expenses excluding the prior-year litigation charge rose faster than revenues. Service customers now expect is the product-defensive tell; with tells in conflict the less durable motive is taken.
Before LLMs: expanded
At the anchor pharmacy teams worked through information systems and a workflow tool the company said served customers while lowering operating costs (anchor claim cvs-anchor-c8), and calls about drug access were among the highest drivers of call volume (anchor claim cvs-anchor-c15); the cost sat inside Pharmacy & Consumer Wellness operating expenses, for FY2024. Conversational AI changes how many calls reach a pharmacist. The size is the change AI made, not the whole line.
“Our ability to attract and retain customers and members is dependent upon providing compliant, cost effective, quality customer service operations (such as call center operations, PBM functions, retail pharmacy and LTC services, retail, mail order and specialty pharmacy prescription delivery, claims processing, customer case installation and online access and tools) that meet or exceed our customers’ and members’ expectations, either directly or through vendors.”
“The Company has continued to invest in information systems to enable it to deliver exceptional customer service, enhance safety and quality, and expand patient care services while lowering operating costs. The proprietary WeCARE Workflow tool supports pharmacy teams by prioritizing work to meet customer expectations, facilitating prescriber outreach, and seamlessly integrating clinical programs.”
“The Company expects benefits from ongoing enterprise-wide cost savings initiatives and investments in efficiencies, which aim to reduce the Company’s operating cost structure in a way that improves the consumer experience and is sustainable.”
“But this category alone is driving obviously significant costs for our clients, and it's also driving significant expense within our organization just to support what is now one of our highest drivers of call volume around people trying to find access to the product and making sure that we get consistent supply in the market.”
Pharmacy & Consumer Wellness total revenues growth, year over year · 2026-CQ2
Reported line it is matched to
Pharmacy & Consumer Wellness operating expenses were against , which held a litigation charge; without it they rose against segment revenues up , above the prior-year rate. The 10-Q attributes the rise to business investments and Rite Aid volume and does not name AI.
acquisition: The Rite Aid asset acquisitions add volume and cost to the segment line (claim c28).
one time item: The prior-year line held a legacy litigation charge, taken out of the bridge.
other: The hours are cumulative over a couple of years and were refocused, not removed; no quarter of them and no part that became lower cost is stated.
transformation program: The CFO's savings figure joins technology efficiencies and AI without a split (claim c17), and the window it covers holds the 2024 enterprise-wide restructuring plan, expected to save over in 2025 and not attributed to AI (anchor claim cvs-anchor-c13); none of it is credited to AI here.
Quotes
“It started with, one, we had to fix our service in our stores, and I'm proud to report we have the best NPS levels that we've seen in many, many years, powered by our technology, the solutions that we put in place, leveraging AI and leveraging what I'd say is better consumer experiences through our digital applications.”
“Over the course of the last couple of years, we've removed hundreds of millions of calls out of our retail pharmacy business and transferred them into conversational AI.
This has allowed us to refocus 1 million hours of pharmacist hours to allow them to deliver better experiences and clinical care at the pharmacy counters.”
“It's also showing up as real improvements in our operations. It's helping ensure we're best in class across the business segments. Over the last few years, as David said, we've generated actually some significant savings, over $1 billion in OPEX savings reductions. That's through the focus on both technology efficiencies, and ultimately, AI.”
“Operating expenses decreased $168 million, or 3.2%, in the three months ended June 30, 2026 compared to the prior year primarily due to the absence of a $542 million legacy litigation charge recorded in the prior year, partially offset by continued business investments and operating expenses to support increased volume from the Rite Aid asset acquisitions.”
Clinical record review in care delivery done with AI
Not sized
The only measure is a cumulative count of pages analyzed with AI, over , with no period or start date (claim c3); a cumulative count sizes nothing (the methodology rule for counts with no period).
described, no sizedisclosure: direction only· motive: exploratory· before LLMs: relabelled
The CEO says AI has been used to analyze over pages of clinical records in care delivery, over a period not given, a count of the work that reads directional and sizes nothing. The channel is left unsized.
Evidence: 3 quotes, 3 figures, 4 confounds, 1 from before coverage
Clinician and coder time spent reading clinical records in the care delivery business (Oak Street Health and Signify Health) that is lower because AI analyzes the records. Management frames it as reducing provider administrative burden and supporting coordinated care. Clinical documentation is expensive to verify, and in this business it also feeds risk adjustment, which management does not mention in the passage.
Why this motive
Management says it continues to expand and refine its use of AI on clinical records to reduce provider administrative burden (claim c3), with a page count and no cost: expansion still under way, the exploratory tell.
Before LLMs: relabelled
At the anchor the care delivery platform already used artificial intelligence and machine learning over claims, pharmacy data and medical records (anchor claim cvs-anchor-c11); the cost sat inside Health Services operating expenses of for FY2024. No covered source says the record analysis uses a different technology or moves a line.
“Canopy integrates an immense amount of data about patients from a broad set of sources, including payor claims data, pharmacy data and medical records from hospitals and specialists and provides actionable insights and workflows to accelerate effective clinical management and oversight. Canopy leverages artificial intelligence and machine learning capabilities to create and refine a clinical rules engine (predictive models and prescriptive algorithms) that informs care delivery and addresses hospital admissions and readmissions, medical costs and patient retention.”
Pages of clinical records analyzed with AI in care delivery, a floor (cumulative, period not given) · as-of 2026-06-30
Health Services segment operating expenses · 2026-CQ2
Health Services segment operating expenses, prior year · 2025-CQ2
What else could explain it
relabel: The care delivery platform already applied AI and machine learning to medical records at the anchor (anchor claim cvs-anchor-c11).
other: Clinical record review in value-based care also feeds risk adjustment documentation, which moves revenue; management does not mention it in the passage.
line composition: Health Services operating expenses also carry pharmacy benefit administration and fell on the absence of a prior-year litigation charge (claim c29).
transformation program: The CFO's savings figure joins technology efficiencies and AI without a split (claim c17), and the window it covers holds the 2024 enterprise-wide restructuring plan, expected to save over in 2025 and not attributed to AI (anchor claim cvs-anchor-c13); none of it is credited to AI here.
Quotes
“We continue to expand and refine our use of AI and analytics to reduce provider administration burden and support better, more connected care for patients. This includes appropriate use of AI to analyze over 1 billion pages of clinical records to enable more personalized and coordinated care.”
“It's also showing up as real improvements in our operations. It's helping ensure we're best in class across the business segments. Over the last few years, as David said, we've generated actually some significant savings, over $1 billion in OPEX savings reductions. That's through the focus on both technology efficiencies, and ultimately, AI.”
“Operating expenses decreased $236 million, or 20.2%, in the three months ended June 30, 2026 compared to the prior year primarily driven by the absence of a $291 million legacy litigation charge recorded in the prior year.”
Medical cost avoided through AI-guided member navigation
0% to 0.03% of the quarter’s revenue
Incremental total: counts at zero.
our inferencedisclosure: described· motive: exploratory· before LLMs: relabelled
The Aetna president says AI agents help members along care pathways and that the website and app have embedded AI. No measure of medical cost was given. The size shown is the ledger's own estimate, against health care costs of .
Evidence: 2 quotes, 2 figures, 2 confounds, 4 from before coverage
Health care costs paid to providers that are lower because Aetna members choose lower-cost providers and care pathways through navigation tools (Informed Choice, Smart Compare, care pathways, the Aetna website and app) that management says rest on technology it has developed in AI. Medical choices are expensive to verify. The line is health care costs, which move mostly for reasons unrelated to AI.
Why this motive
AI agents on care pathways and AI embedded in the website and app (claims c11 and c12) are named with no measure and no line moving: the exploratory tell for a passing mention. No narrative-defensive tell of its own (cuts, flat headcount, mandates) is quoted.
Before LLMs: relabelled
At the anchor Aetna already used advanced AI tools to pick and engage high-risk members for care management (anchor claim cvs-anchor-c14) and digital products to guide members (anchor claim cvs-anchor-c2); health care costs were for FY2024. No AI part of medical cost has been measured since.
“There is continued focus and investment in enterprise data platforms, cloud capabilities, digital products to offer innovative solutions and a seamless experience to the Company’s members through mobile and web channels.”
“The Health Care Benefits segment currently operates and supports an end-to-end suite of information technology platforms to support member engagement, enrollment, health benefit administration, care management, service operations, financial reporting and analytics.”
“Canopy integrates an immense amount of data about patients from a broad set of sources, including payor claims data, pharmacy data and medical records from hospitals and specialists and provides actionable insights and workflows to accelerate effective clinical management and oversight. Canopy leverages artificial intelligence and machine learning capabilities to create and refine a clinical rules engine (predictive models and prescriptive algorithms) that informs care delivery and addresses hospital admissions and readmissions, medical costs and patient retention.”
“We're actually using a lot of really advanced AI tools to identify those members. We really think we have excellent analytics to be able to pinpoint who those members are and how are we best able to engage with them, what are the types of things that will get them to engage with us”
other: Health care costs move with utilization, membership mix, the exit from the individual exchange business, risk adjustment and prior-year reserve development, all far larger than any navigation effect.
relabel: AI tools in care management and digital member guidance predate LLMs at the company (anchor claims cvs-anchor-c2 and cvs-anchor-c14).
Quotes
“Improving the member experience. We can now help them along their care pathway. We can actually use our AI agents to help schedule appointments, creating less administrative burden on everyone.”
“I could go on and on, a lot of deployment there in our website and our app have both embedded AI capabilities that make it an industry-leading capability.”
Revenue arriving through AI2 channels · 2 not sized
product revenue
Health100, the AI-native consumer engagement platform
Not sized
Not in market during the quarter: the targeted launch began the month after the quarter ended, by the CEO's account on the call of 2026-08-05 (claim c1), and no price, client or revenue has been given.
described, no sizedisclosure: described· motive: exploratory· before LLMs: expanded
The targeted launch of Health100, with the HIO AI-powered assistant, began the month after the quarter ended; management expects growth from it and gives no price, client or revenue. The channel is left unsized until it is in market.
Evidence: 4 quotes, 1 confound, 2 from before coverage
Money the company may earn from Health100, a health technology services subsidiary building what management calls an AI-native platform that any payer, pharmacy benefit manager, pharmacy or provider can connect to, with a consumer app and the HIO AI-powered assistant. Management frames it as a growth investment; no price, client count or revenue has been given, and the targeted launch began after the end of Q2 2026, so the covered quarters carry no revenue from it.
Why this motive
A targeted launch that began after the quarter ended, with early feedback and wider access to come (claim c1), framed as the growth side of the technology investment (claims c20 and c21): still pre-revenue in the quarter, the exploratory tell.
Before LLMs: expanded
At the anchor the 10-K already described digital products for members on mobile and web and a pharmacy digital strategy of online and mobile tools (anchor claims cvs-anchor-c2 and cvs-anchor-c12); none of it was a separately reported revenue line. Health100 would offer a successor of that engagement work to outside payers and providers. No amount was given then or in the covered quarters. The size is the whole of an activity that existed before.
“There is continued focus and investment in enterprise data platforms, cloud capabilities, digital products to offer innovative solutions and a seamless experience to the Company’s members through mobile and web channels.”
“The Company’s digital strategy is to empower the consumer to navigate their pharmacy experience and manage their condition through integrated online and mobile solutions that offer utility and convenience.”
other: The platform is a successor to the company's own member and pharmacy apps; revenue it earns from outside payers may displace engagement the company already did for its own members.
Quotes
“Last month, we began the targeted launch of our Health 100 platform, including HIO, our new AI-powered assistant. HIO is designed to simplify the consumer experience and help people engage more effectively in their care journey. Early feedback has been encouraging, and we are excited to expand access later this year.”
“I think Steve and Prem did a nice job of how we're prioritizing and the results that we're seeing, but I think you're going to see us balance two parts of technology and AI. One is the efficiencies and the productivity that we spoke to, but equally as important is the growth aspect of the investments that we're making.”
“That's why we're leaning in heavily to the consumer-based healthcare technology business, the innovations that we're driving, and the growth that we expect, in large part, seeing some of this come through even in the early signs of where we expect Health 100 to ultimately grow and impact healthcare in this country.”
“Lastly, I'll say, we think we have a unique opportunity, as we announced at Investor Day, with Health 100 to create a unique solution that integrates across any pharmacy, any provider, any PBM, et cetera, that we can bring into the marketplace, that really allows all consumers across the country to benefit from the integration of what we're bringing together and what we can bring to market.”
Q1 2026described · described, no size · exploratory
Q2 2026described · described, no size · exploratory
product revenue
Prescription revenue from AI-assisted adherence programs
Not sized
The specialty adherence level is credited to technology, automation and AI together (claims c4 and c14), and the retail STARS use of AI gives no measure (claim c30); nothing separates AI’s part (the methodology rule for AI named beside another cause).
described, no sizedisclosure: described· motive: exploratory· before LLMs: relabelled
The CEO and the group president call the specialty pharmacy AI-driven, crediting technology, automation and AI together for adherence above against an industry standard of , and say retail STARS programs use AI for adherence. The money effect is refills; none is given, and nothing separates AI’s part, so the channel is left unsized. It is tagged relabelled.
Evidence: 3 quotes, 4 figures, 2 confounds, 2 from before coverage
Pharmacy revenue from refills that would not happen without adherence programs that management says embed AI: the specialty pharmacy, which management calls AI-driven, and the retail STARS adherence programs. Payers and plan sponsors pay for most of these fills. Management frames the programs as clinical quality and lower client cost and gives no revenue effect.
Why this motive
The adherence level is credited to technology, automation and AI together (claims c4, c14 and c30), for programs the anchor shows running on clinical algorithms before LLMs, and no change in revenue or in the rate is attributed to AI: AI named as a cause with no measure, the exploratory tell. No narrative-defensive tell of its own is quoted.
Before LLMs: relabelled
At the anchor the Health Engagement Engine and data science clinical algorithms already flagged medication adherence issues for pharmacists and served cost management and health improvement (anchor claims cvs-anchor-c9 and cvs-anchor-c10). The covered quarters describe the same programs as AI-driven, with no revenue attributed to AI.
“The Company’s Health Engagement Engine technology and data science clinical algorithms enable the Company to help identify opportunities for pharmacists to deliver face-to-face counseling regarding patient health and safety matters, including medication adherence issues, gaps in care and management of certain chronic health conditions.”
“The Health Engagement Engine® technology and proprietary clinical algorithms help connect various parts of the enterprise and serve an essential role in cost management and health improvement, leveraging cloud-native technologies and practices.”
Specialty pharmacy adherence rate, a floor · 2026-CQ2
Industry adherence standard the CEO compares it with · 2026-CQ2
Health Services mail and specialty revenues (specialty mail claims, Specialty Connect, mail order) · 2026-CQ2
Health Services segment total revenues · 2026-CQ2
What else could explain it
relabel: The Health Engagement Engine and clinical algorithms already targeted adherence at the anchor (anchor claims cvs-anchor-c9 and cvs-anchor-c10).
mix shift: Specialty revenue grows with drug mix, brand inflation and generic launches, which the 10-Q names as the segment's revenue drivers.
Quotes
“Building on that foundation, our focus on technology, automation, and embedding AI into our processes have made CVS Specialty the most tech-enabled specialty pharmacy in the industry. Our tools and processes are designed to identify patients earlier, initiate therapy faster, and support them continuously to help improve adherence and outcomes. While others in the industry are working towards achieving the standard of 80% adherence, CVS Specialty consistently operates above 90%.”
“When you look across our pharmacy assets, in our specialty pharmacy, we have one of the most tech-enabled and AI-driven specialty pharmacies in the country. We're improving the adherence rates to over 90% of these drugs, which help lower the cost of these highly expensive drugs, as well as improve the quality of care.”
“Secondly, in our retail pharmacies, we leverage AI in our STARS programs to help improve adherence and other things as well there to have the operational and the clinical benefit as relates to that.”
Q2 2026. Growing slower than revenue: health care costs (+0.5%), operating expenses (−1.5%), total operating costs (+5.0%). A displaced cost shows up as a line that stays under the dashed revenue line. These are the audited lines, as first reported; nothing here is attributed to AI by the filing.
Cost of products soldHealth care costsOperating expensesTotal operating costsRevenue