Don't try to replace your CRM. Make it irrelevant instead. Those sound like the same goal, and they are the opposite strategy. One is a multi-year migration that invites every stakeholder to fight over the same territory. The other is a series of small moves nobody objects to, until one day people simply aren't living in Salesforce anymore.
I've watched enough enterprise transformation efforts to notice a pattern. The ones that fail start with the platform. The ones that work start with the work. This is the argument I've been making in pieces over the past few weeks, pulled into one place: how a Context-First Enterprise actually gets built, from the first workflow to the point where context becomes the operating layer.
You don't win the CRM war by winning the CRM war. You win by making the CRM stop mattering.
Most enterprise AI fails before anyone writes a prompt
The failures rarely come from the model or the data. They come from choosing the wrong thing to automate first. Companies reach for the high-visibility workflows: forecasting, CRM data entry, the executive dashboard. Those are exactly the workflows with the most scrutiny, the most politics, and the most people invested in the current version. You're picking a fight on your worst terrain.
The workflow people pick
Forecasting, pipeline hygiene, the board dashboard. Maximum visibility, maximum scrutiny, maximum number of people who will argue with the output.
The workflow that wins
Executive account prep, customer briefs, implementation summaries. Everyone agrees it's broken. Nobody defends the status quo. The win is undeniable.
Start with something unglamorous that every single person already agrees is painful. You don't need to convince anyone the task is worth fixing, so the only thing left to prove is that the fix works. That's a much easier bar, and clearing it buys you the one currency that matters in enterprise change: trust.
The first workflow I'd move: the executive account brief
Here's the one I'd start with, because I've lived it. Preparing for a customer meeting today looks like this: open Salesforce, then the support queue, then engineering issues, then the renewal dates, then product usage, then email, then Slack, then last quarter's QBR deck. Assemble it all by hand into a PowerPoint. Three hours gone, and not one minute of it created a single ounce of customer value. It's administrative overhead manufactured entirely by disconnected systems.
Now picture the version where you ask one question:
"Prepare me for tomorrow's meeting with Acme."
And you get back a brief: executive summary, open opportunities, support risks, adoption trend, key relationships, renewal exposure, engineering issues in flight, recommended talking points, and the next actions. In seconds, not hours.
Notice what didn't happen. Nobody ripped out the CRM. Nobody migrated a single record. We took one painful workflow and connected it across every system that held a piece of the answer. The CRM is still there. It just stopped being the place you go to do the work.
Why starting small is the whole strategy
The instinct in a big company is that big problems need big projects. With AI, the reverse is true. Small, undeniable wins are how you earn the right to attempt the big ones. The sequence looks like this:
- Start with one high-value workflow. Something broken enough that everyone agrees, unglamorous enough that nobody defends it.
- Build trust through quick wins. When people get their afternoon back, the conversation shifts from "will this work?" to "what do we do next?"
- Create shared memory, not more integrations. The goal isn't another point-to-point connector. It's a common context every workflow can draw from.
- Let context become the operating layer. Once the shared memory exists, each new workflow is faster to add than the last.
- Let the CRM become a system of record. Not the center of work. Just the place the record lives.
After a year or two of this, something quietly remarkable happens. Salesforce is still there. It still holds the data. But people aren't living in it anymore. The center of gravity moved, and no one had to fight a migration battle to make it happen.
Adoption you drive with efficiency lasts. Adoption you force with a rollout gets rolled back.
The reframe: context as the operating system
Step back and the pattern becomes a thesis. For thirty years, the enterprise stack has been organized around applications: the CRM, the CPQ, the ERP, the ticketing system. Work meant moving between them, and humans were the connective tissue, carrying context from one system to the next by hand.
The Context-First Enterprise inverts that. Context stops being something scattered across applications and becomes the layer the work runs on. The application is no longer where you go; it's just where a fact happens to be stored. The system of record recedes, and the shared context moves to the front.
The real question of this decade was never "will AI change work?" Of course it will. The question is what sits at the center of work when it does. My bet is that it won't be an application. It'll be context.
Name one workflow you could move out of your CRM this week.
Not replace the whole platform. Just one workflow, painful enough that everyone agrees, that you could connect across systems and hand back finished. Start there. That's the whole playbook.
That's the argument, start to finish. Don't storm the CRM. Move one workflow, earn trust, build shared memory, and let context become the operating layer. The platform stays. Its centrality doesn't.
Consolidated from a series of field notes on why enterprise AI transformation succeeds by starting with the work, not the platform.