Applied AI

Agentic CRM sounds like an enterprise story. Here's the SME version.

Microsoft says AI agents will turn your CRM from a system of record into a system of action. The idea's right — and you need far less than their stack to get the part that pays.

FidwenJuly 20265 min read

Microsoft has published a piece arguing that your CRM is "a beautifully organised rear-view mirror" — a system built to record what already happened, not to do anything about it. It's a sales pitch for their own agentic tools, so read it with that in mind. But the core diagnosis is the truest thing a CRM vendor has said in years.

For thirty years, CRM has been the place people report to after the work is done. It stores the customer truth and asks your team to keep it fed. The result is what Microsoft nicely calls the "CRM tax": your most capable, most expensive people spend a chunk of every week typing up what they just did instead of doing more of it. The record goes stale, the forecast becomes a guess, and before long nobody quite trusts either.

"Agentic CRM" flips that round. Instead of asking people to describe reality, AI agents capture and update the record automatically from the emails, calls and messages already flowing through the business — and surface the next sensible action inside the tools people already use, rather than in a separate system they have to log into. On the principle, we agree completely. It's the same thing we tell clients about agents generally: the win isn't a cleverer database, it's giving people their week back.

Where a seller's week actually goes
On work that could be automated or delegated25 hrs
On high-impact selling, where people add unique value9 hrs
Source: Gartner, cited in Microsoft's Dynamics 365 blog (2026). Weekly averages, self-reported.

That's the prize, and it's a real one. But this is where we'd steer a smaller business away from swallowing the pitch whole.

Read the proof points with a level head

The article stacks up impressive numbers — sellers closing 20% more deals, higher conversion, more revenue per head. Read the small print and most of it is Microsoft's own internal telemetry, comparing people who use the tools heavily against people who don't. That's a correlation, not a result you can bank: your best salespeople tend to adopt new tools first anyway. The named customer stories are early and enthusiastic, not independently measured. None of it is wrong, exactly — it's marketing doing its job.

The demo always works. The real question is whether your data and your governance are ready for an agent to act on its own — and for most firms, not yet.

The bigger catch for an SME is that the whole vision assumes you're already standing on their stack — the enterprise CRM, the Copilot licences and the productivity suite, all wired together. That's a fine place to be if you're there. It is not the price of entry to the actual benefit.

The version that pays for a smaller business

You do not need a seven-figure platform to get the part that matters: agents quietly removing the admin so your people can sell, quote and look after customers. You can get most of the way there with tools you already own and a couple of well-scoped agents. What decides whether it works isn't the brand of software — it's three unglamorous things.

What actually decides it
One use case

Pick a single job — first response, quotes, chasing invoices — with a clear number that says it worked.

Clean-enough data

An agent is only as good as what it reads. It doesn't need perfect data — just data it can trust for this one job.

A human who signs off

Before anything touches money, a customer or risk, a person decides. The agent drafts; you approve.

Fidwen.

Get those right and the technology is almost a detail. Get them wrong and the most advanced agentic CRM on the market will still produce confident nonsense, only faster.

Which job would you hand an agent first?

Get an honest read on where AI would actually pay in your business — and which use case to start with. About three minutes, no sign-up.

Take the free AI analysis →

"In the flow of work" means meeting people where they already are

The one phrase from the Microsoft piece worth stealing is "in the flow of work." For an enterprise, that means Outlook and Teams. For your business it might mean the shared inbox, WhatsApp, the quoting spreadsheet, or whatever system your team actually lives in. The principle holds at any size: don't make people go to the AI. Bring the AI to where the work already happens, and it gets used. Put it behind another login and it won't.

There's a genuine point buried in the trust framing, too. When every enquiry gets a fast, accurate, personal response — because an agent caught it, pulled the context and drafted the reply for a person to send — customers feel it. Trust is built in the boring consistency of following through, and that's exactly the sort of thing a well-governed agent is good at.

A short test before you buy anything

If a vendor — Microsoft or anyone else — is selling you agentic CRM, three questions cut through the demo:

Answer those and you'll know whether you need a new platform or just a well-aimed agent and a tidy-up. Usually it's the second — and it costs a fraction of the first.

Share

Thinking about agents in your sales or service?

We help smaller firms pick the one job worth handing an agent, get the data and sign-off right, and put it live where the work already happens — without a platform migration.

Book a call Start a conversation

Sources: Ryan Braadstad, "Agentic CRM in the flow of work: How AI is transforming sales and rebuilding customer trust," Microsoft Dynamics 365 Blog (25 June 2026), and the figures it cites — Gartner on how sellers' time is spent, LinkedIn's B2B buyer research, and Microsoft's own internal sales telemetry. Those results are largely vendor-reported or self-selected and are directional rather than independently audited; the commentary and the SME framing here are our own.