Practical AI for finance and ops leaders. Weekly, from Umar Din.

Leader, welcome back.

Saving time with AI is easy. It's also where most teams stop, and it's the half that doesn't move your numbers!

Read time: 5 minutes

Most teams are using AI to do one thing: go faster. Clear the admin, draft the email, speed up the close. Useful, and only half of what it is for.

I see it in almost every enquiry we get. Companies want two things, help their people understand AI, and save time. Both fair. Both the easy half. But it shouldn’t stop there.

15 years of tech change in finance and ops teaches you one thing fast.
You can see exactly where the hours leak. Reconciliations. Status updates. Copy-paste reporting. The 4th version of the same email.
AI clears that out, and it is usually where we start, because it is real, measurable, and it gets people on board.

But saving time is table stakes now. The question I keep pushing clients toward is the one they almost never ask first: how do we use this to make more money, not just spend less of it?

That is a different conversation entirely. It means pointing AI at the top line, not only the cost line. Proposals out same-day instead of next-week. Pipeline qualified properly instead of rotting in a list. Pricing and upsell signals pulled from data you already own and never look at. Outreach that is genuinely personal at a scale one person could never reach by hand.

The time savings get you in the room. The revenue is what makes the AI pay for itself ten times over.

If your team is only using AI to go faster, you are using half of it.

You have seen the headlines. AI is coming for the jobs, the layoffs are starting, your role is next. It is the loudest story in business right now, and it is not invented.
The cuts are real, and plenty of companies are happy to point at the robots.

But listen to the people actually building this stuff, and a different read emerges.

Salesforce has not grown its engineering team in two years. The panic version says AI froze hiring. What Marc Benioff actually said is that the 15,000 engineers he already has got around 30% more productive with AI, so he does not need more. Nobody was cut. Everyone got an amplifier.

Cognition, the maker of the Devin coding agent, just raised over $1bn at a $26bn valuation, with revenue up from $37m to $492m in a year. The scary read is "the agent replaced the coder." The truer one: someone now points a squad of those agents at work that used to take a whole team. The job moved up a level, to the person aiming the tool.

Even Jensen Huang, who sells the chips this all runs on, told CEOs to stop blaming AI for layoffs and called it "just too lazy." His point: the tech still needs a human to set the goal and catch the mistakes. That human is worth more now, not less.

The market clearly thinks there is something here. Anthropic just filed to go public at a $965bn valuation, on revenue that reportedly jumped from $10bn to $47bn in a year.

So what does it mean for you?
The dividing line is not how senior or technical you are.
It is whether you hand work to the machine or wait for work to be handed to you.

The finance person who pulls ahead this year treats Claude or Copilot like a new analyst on the team. Hands it the variance pack, the reconciliation, the first draft of the board narrative. Checks the output. Then spends the freed-up time on the judgment call only they can make.

Same point as the reflection up top. The people AI rewards are not the ones it makes spare. They are the ones it makes bigger.

One bit of context first, because it is having a moment. Microsoft just dropped a Dynamics 365 ERP plugin into Copilot Cowork in preview, and finance and ops people online have not stopped talking about it.
It lets Cowork work inside your ERP: open forms, run reports, pull a trial balance, build a dashboard from live data, all on your own login and security roles. If you run D365, worth a look.

But you do not need an ERP to get value from Cowork today.
Here is one to set up this week:

Install Cowork Agent in Microsoft 365 Copilot (or in Claude), point it at your sent mail, and schedule it for Monday morning.
It finds the asks you are still waiting on, drafts a nudge for each in your own writing style, and emails you the batch. All you do is review and hit send.

Paste this, swap in your email, schedule it for Monday 8am:

The trick that makes it land is point 2: tell it to learn your voice from your own sent mail, so the drafts sound like you, not a robot.
Set it once and the Monday chase-up becomes a five-minute review with your coffee!

Claude Opus 4.8 (worth watching). Anthropic's newest model landed last week and is already inside Microsoft 365 Copilot. The headline is not that it is smarter.
We have found it is better at admitting when it is unsure and catches more of its own mistakes. That matters: a tool that confidently makes things up is one people quietly stop trusting. One that says "check this, I'm not certain" is one they keep using.

ChatGPT in your spreadsheets (handy). ChatGPT now works directly inside Excel and Google Sheets through a sidebar. Cleanup, formulas, budgets and trackers, without bouncing between tabs. If your finance team lives in spreadsheets, worth 10 minutes of a play.

Microsoft shipped a pile (look when you can). Two to clock: Scout, an always-on personal agent meant to sit across your day and flag what needs you, and a 365 Copilot redesign that loads about twice as fast and only shows the tools relevant to your prompt. Unglamorous, but speed and less clutter are what keep a tool in daily use.