Article

17 Oct 2026

What your inbox knows that your CRM doesn't

Your inbox, CRM and whiteboard each tell part of the story about your pipeline. Connect them to one AI and the gaps between them start to show.

Cartoon: a manager presents a tidy deals board while a robot finds a cobwebbed card and emails burn under the table

Here's a moment most leadership teams will recognise. It's the weekly pipeline meeting, someone shares their screen and starts working down the deals board, and about a third of the way through, a look travels around the room. Everybody knows that deal went cold in September. Everybody knows that one closed last week and nobody moved the card. Nobody says so, partly because it would mean admitting their own column looks much the same, and partly because the board has become a polite fiction everyone has agreed to maintain until the end of the quarter. Underneath the updates, people are thinking something simpler and more uncomfortable: the real picture of where the business stands lives in a dozen separate inboxes, and no single system can show you all of it.

Those gaps between systems are exactly where AI connectors start paying for themselves.

What a connector does

A connector gives Claude access to another system, so it can pull information from it and, where you allow it, take action in it. Claude offers a directory of connectors for tools like Microsoft 365, Monday.com and Miro, and you can add custom ones for services the directory doesn't cover. You don't need to tell Claude which connector to use, either. Name the system in plain English, check my email or look at the deals board, and it works out the rest. On its own, each connector saves a bit of copying and pasting, and several in one request can save an afternoon.

Asking across systems

To test this properly, I wrote one prompt and let Claude work through it. I asked it to act as an expert account management director and build a mind map of the accounts I'm actively working on, based on the last 30 days of my email in Microsoft 365. Next, it had to cross-check that list against our deals board in Monday.com and tell me which deals looked out of date. After that came a new Miro board called Live Opportunities, built with our brand skill and showing where each opportunity stands. Finally, I asked for an infographic summarising the month, generated through Higgsfield, an image tool connected to Claude the same way.

Because the job was large and touched four systems, I ran it in Cowork on Opus, Claude's most capable model, which can split a job like this across several sub-agents and work through each part more thoroughly.

Before touching anything, Claude came back with questions. Where should the mind map live? What should it do with deals that looked out of date? What should the infographic show about deal values? And the one that stopped me in my tracks: what counts as an account?

The question nobody had agreed

Is an account a company you've invoiced, one you've had three calls with, or one that's received a proposal? Most sales teams have never agreed an answer, and their CRM reflects it, with each person applying their own definition and nobody noticing until the forecast stops adding up. Claude couldn't do the job without a definition, so it asked for one.

Claude also asked what I wanted done with the out-of-date deals before changing anything, which is exactly the instinct you want from a tool connected to the system your sales forecast depends on.

What came back was a Miro board that pulled the email activity and the deals data into one view of live opportunities, alongside a short update listing which deals on the Monday board needed changing and what the changes should be. Laid side by side, the inbox showed who I was really talking to and the deals board showed who I'd said I was working with, and the value lived in the difference: the kind of deal marked "proposal sent" that hasn't seen an email in two months, and the conversation running hot in your inbox that never made it onto the board at all.

It wasn't flawless. A couple of the connecting lines on the Miro diagram overflowed into each other, and the infographic was serviceable at best (fine for an internal update in a pinch, though I'd want more before a client saw it). Fixing the lines takes one more sentence to Claude, and the infographic makes a reasonable starting point for something better.

Knowing what to ask becomes the skill

What I've noticed, watching teams work like this for the first time, is that the people who get the most out of it are the ones who can describe clearly what they want, in what order, and what a good answer looks like. Most leaders have spent years honing exactly that skill every time they've briefed a capable colleague, and it transfers almost perfectly. You'll know the feeling already: the brief you wrote carefully before a week's holiday that came back as exactly what you'd pictured.

For years, joining up information across systems meant either a patient person with three browser tabs open or an integration project with a budget code. Here it took one carefully written prompt and a few minutes of answering questions, and the person who wrote the prompt got to spend their time deciding what to do about what it found.

Before you connect everything

When you give an AI access to your email or your CRM, it can see whatever those permissions allow, and in most businesses that's more than anyone realises. Check what's shared before you switch anything on, and decide which actions Claude can take on its own and which need your approval first. Be careful about what leaves the building, too. The version of this exercise I recorded for LinkedIn used anonymised, randomised data, because broadcasting a real client pipeline to the internet would be a memorable career move for all the wrong reasons.

The most useful output of the whole exercise may well have been that question about accounts. Ask it at your next leadership meeting, what actually counts as an account here, and count the answers. If you get more than one, your inbox and your CRM have been telling different stories for a while, and the obvious next question is which one your forecast has been listening to.

Where to go next

Before you connect AI to your email and files, our M365 AI Safety Audit shows exactly what it will be able to see. To build your first connected workflows with the team, look at the AI Hackathon. Related: How much should your AI do without asking?.