AI ROI Brief · November 5, 2026

Name the line: two firms that tie AI to a dollar figure

Most enterprise AI pilots still show no measurable P&L impact. This brief tracks the minority, the five percent that do, and what they do differently. This week the pattern is naming the line. The two firms below do not just say AI helps; they say which ratio or income-statement line it moves and by how much.

The 30-second version

Intact Financial says AI earnings will double to CAD 500M in 24 months, and shows the pieces: 600 models, 94% of Canadian personal lines priced by ML, quote turnaround down 70%

Talking point

Most firms say AI is working. Intact puts a dollar figure on it, splits it between loss ratio and expense ratio, and names the workflows underneath: pricing, quote intake, claims documents. The catch for a mid-market firm is the backstory: management says the capability took more than a decade to build in-house. Copy the accounting, not the timeline. Name the dollar line each AI use case is meant to move before you build it.

Content angle

Short post or single graphic, 'Where the money shows up': Intact's 60/40 split between loss ratio and expense ratio, then one question for any regulated firm: for your last AI pilot, which ratio or P&L line was it supposed to move, and who reports on it?

Lens: Executive AI readiness orientation — methodology and technology, not law. Figures are management-reported from earnings coverage and have not been independently audited; both articles note AI-assisted production. Associations reported here are not proof of cause. Fractional C-Sweet provides executive AI readiness education, not investment, legal or compliance advice.

Cost

Dollar figures and the line they move.

InsuranceRegulated

Intact Financial says AI earnings will double to CAD 500M in 24 months, and shows the pieces: 600 models, 94% of Canadian personal lines priced by ML, quote turnaround down 70%

Intact's Q3 2026 results (reported Nov 4): recurring pre-tax AI earnings of about CAD 250M expected to reach CAD 500M over 24 months, roughly 60% from loss ratio and 40% from expense ratio; commercial quote automation cut turnaround 70%; legal-claims document summarization prep time down about 50%.

Talking point

Most firms say AI is working. Intact puts a dollar figure on it, splits it between loss ratio and expense ratio, and names the workflows underneath: pricing, quote intake, claims documents. The catch for a mid-market firm is the backstory: management says the capability took more than a decade to build in-house. Copy the accounting, not the timeline. Name the dollar line each AI use case is meant to move before you build it.

Content angle

Short post or single graphic, 'Where the money shows up': Intact's 60/40 split between loss ratio and expense ratio, then one question for any regulated firm: for your last AI pilot, which ratio or P&L line was it supposed to move, and who reports on it?

Source: Investing.com, 'Intact Financial Q3 2026 slides: AI push drives earnings beat,' Nov 4, 2026

BankingRegulated

Customers Bancorp reports 76,000 hours saved in Q3 and revenue per employee up 13%, with headcount down from 905 to 860

Q3 2026 results (reported Oct 23, per Reuters via Investing.com): AI and automation saved a cumulative 76,000 hours, loan underwriting cycle time fell 95%, commercial deposit onboarding fell more than 90%, and the efficiency ratio improved about 340 basis points quarter over quarter to 47.2%.

Talking point

This one connects hours to the income statement: revenue per employee up 13%, profit per employee up 34%, and 60% of new hires since Q1 in revenue-producing roles. Hours saved only become ROI when someone decides where the freed capacity goes. Here management says it went to revenue producers, not more support staff. The checking time on AI output is not mentioned, so the 76,000 is likely a gross figure.

Content angle

LinkedIn post or talk slide, 'Hours saved are not a return until you spend them': Customers Bancorp's 76,000 hours, then the 60% revenue-producer hiring mix as the proof that the capacity was redeployed. End on one question: where did the hours your last pilot saved actually go?

Source: Investing.com (Reuters), 'Customers Bancorp Q3 2026 slides: AI drives efficiency gains,' Oct 23, 2026

Productivity & Workflow

Proof, measurement and sequencing.

Partner channelOwn data

Proof assets get built when they have a budget line, an owner and a date

A planned video case study was replaced with a partner-funded video podcast that must be finished and invoiced inside a funding window of about three weeks.

Talking point

The proof that exists is the proof with a budget line, an owner and a deadline. A podcast or case study is a place to state results, not a result. Fund the measurement alongside the story.

Content angle

Workshop exercise, 'Fund the proof': for any AI initiative, write down the budget line, the owner and the date by which a before-and-after number will be published.

Source: Fractional C-Sweet own activity

Executive operating systemOwn data

Read-only first: the safest way to start an executive AI workflow

The pilot sequence we use: read and summarize first, add editing a week later, and extend to email and shared files only after the first measurements are in.

Talking point

Pilots that show returns tend to start read-only. It limits risk before any number is claimed, and it gives you a clean before-and-after on one task. Write the metric down before the first run.

Content angle

Post or talk segment, 'Read-only first': why the first version should read and summarize but not write or send, and the one number to capture in week one.

Source: Fractional C-Sweet own operating practice

Bottom line

Bottom line

Keep reading

Next briefs

Get the briefs in your inbox

AI in the News, Legal Signal, Compliance & Cyber, and ROI briefs — written for executives. No spam, unsubscribe anytime.