On Wednesday, Marc Benioff and Dario Amodei announced Claudeforce during Salesforce’s quarterly earnings call. The stock jumped the same evening. A few days later, my phone rang. The director of a company we’re about to sign for a HubSpot project. Salesforce was the other name on his shortlist. “Igor, what is this? What do they have that I haven’t seen from you?”
Fair question. At our workshops he’d watched me run HubSpot through Claude. Reviewing my team’s sales activity and reporting on their work. Updating a contact, a company, a deal. Checking where a customer stands. Watching deals move through the pipeline. Even fixing website copy, straight in HubSpot’s CMS. Now he’s watching the same thing on CNBC, packaged as the event of the year.
My answer: you haven’t missed a thing. You saw the premiere before Wall Street did.
Let’s be honest: Claudeforce is investor theatre first. The announcement landed right in the middle of an earnings call. It came after a year in which SaaS investors lost serious money. Benioff told analysts it was time for the “SaaSpocalypse” nonsense to stop. As theatre, it worked. There’s even a signature move. For the first time ever, Salesforce attached its “force” suffix to another company’s product. That tells you who held the better cards. According to Menlo Ventures, Anthropic holds 40% of enterprise spend on large models.
But under the show sits real money. Salesforce will spend around $300 million on Anthropic’s models this year. Its stake in Anthropic is worth around $5 billion. Claude is becoming the default model behind Salesforce’s agents. And the partnership’s first product is a plugin with 37 prebuilt “skills” for salespeople: meeting prep, pipeline updates, email drafts, activity reports.
Read that list again. It’s exactly what my caller saw at our workshop.
The most interesting sentence didn’t come from the ad. Patrick Stokes, Salesforce’s president of applications, told CIO that the value of Salesforce rises dramatically once people stop using it through the classic interface. He also admitted where the idea came from. Anthropic’s own salespeople preferred running their Salesforce through Claude.
A company that spent a quarter of a century selling screens, fields and views is now advertising, as its main benefit, that you no longer have to open its interface.
That’s not innovation. That’s a confession. A CRM was never valuable because of its interface. The value is in the data and the processes above it. The interface was always a cost: clicking, data entry, training, salespeople’s resistance. And Claudeforce prepares the ground for the next step. Soon it won’t be only people using a CRM. Agents will work in it too, doing specific jobs instead of employees. Your CRM’s next user won’t be human.
Two days later, HubSpot co-founder Dharmesh Shah introduced YouSpot. An AI-native CRM for one-person companies. Intro price: one dollar a month. His framing: a CRM you don’t have to maintain. It connects to Gmail, your calendar, LinkedIn and X. Then it builds the context itself: who you know, what you’ve talked about, which follow-up slipped. No human enters the data.
For you, YouSpot is not an alternative. Shah himself says sales teams belong on HubSpot. But it is a signpost. HubSpot is letting its own co-founder undermine the assumption the whole industry stands on. That someone has to enter, clean and update the data. One week, two directions, one conclusion: CRMs are no longer built for clicking. They’re built for agents.
Software an AI agent can’t buy and use on its own will disappear from the market.
Now the good news for those of you on HubSpot. What Claudeforce is selling as exclusive comes as standard on your platform. HubSpot’s MCP server has been generally available since April. Free on every tier. The Claude connector switches on in minutes. It respects the permissions each user already has in HubSpot. No migration, no new licence, no waiting.
The category Salesforce just advertised for millions already exists on your platform. The only thing missing is the one thing nobody sells.
So what is this “skill” Benioff rented Wall Street for? A written playbook. An instruction precise enough for a machine to run: what to check, where to look, what to do, and when to ask a human.
Our skills didn’t come out of a lab. They grew out of implementation projects with our customers. We use them internally every day. At FrodX, meeting prep is one prompt. The customer summary, open deals and recent activity arrive by email or in Teams, together with the meeting reminder. The same prep used to mean some twenty minutes of digging through the CRM, the calendar and old emails. Which is why it often didn’t happen at all. People improvised in the meeting. Same with sales calls and follow-ups. After a Teams call you get the minutes, the tasks that follow from them, and an analysis of how the conversation went. Even coaching for running the next meeting better, or for handling escalations.
The technology was the easy part. The harder question was what to write down in the first place. Which steps actually move a deal? Where does a salesperson lose hours? Which data has to be fresh for the forecast to mean anything? That’s not engineering. That’s knowing how sales works.
Here’s the bigger picture. Once connectors are standard and the interface is a conversation, engineering skill starts counting for less on the implementation side. Configuration that used to mean a project becomes a prompt. The difference between providers won’t be “who can set it up”. It will be “who knows what to write down”.
Today you can still buy a CRM as an IT project. In two or three years, my bet is you won’t. You’ll buy a written record of your own processes. And an advisor who knows which processes are worth the effort. One who has seen them at dozens of companies. The question you ask your partner will flip. No longer “what does the implementation cost”, but “which of our processes can you turn into a playbook an agent will run tomorrow”.
We’ve been investing in that position for ten years. There were years when business consulting was harder to sell than licences and go-lives. Now Benioff has paid for the campaign. One we could never have afforded ourselves.
The man from the start of this column missed nothing. His decision has just been confirmed from the least expected direction. The category is real; both camps put their names to it the same week. The connector already exists on his platform. And the playbooks were never for sale anyway.
Which of your processes is ripe for writing down first? I’d start with meeting prep. Lowest risk, savings in the first week, and every salesperson feels it. Automate further only after that.
The show was for investors. The message is for you.