CRM’s AI race moved from slogans to surfaces

Salesforce turned AI into the category price of admission

The fortnight was not quiet. Salesforce dominated the tape with AI, capital allocation and market reaction all pointing in the same direction. On 27 Aug, it reported record second-quarter fiscal 2027 results, raised its full-year outlook and expanded its AI partnership with Anthropic. On 28 Aug, further reports said Agentforce ARR had passed $1.5bn, while the stock rose 22% after strong Q2 earnings and a positive second-half outlook. On 5 Sep, press coverage said Agentforce and Data 360 ARR had increased by more than 200%.

The product message hardened fast. On 29 Aug, Salesforce integrated its CRM into Claude and launched the Claudeforce AI sales plugin with Anthropic. On 31 Aug, the company announced Claudeforce more formally, and on 6 Sep it introduced AI document tools for regulated workflows. That last move matters because it shifts the AI story away from generic sales productivity and towards compliance-heavy buying committees.

Salesforce also used its balance sheet as part of the message. On 2 Sep, it invested $166m in HiBob, valuing the HR tech company at $3.2bn. On 4 Sep, it led a $550m Series C for Wonderful at a $5bn valuation and also led the $166m HiBob round. At the same time, on 4 Sep, press reports noted pressure on profit margin guidance linked to Claude usage. The market is being asked to believe two things at once: AI is lifting growth, and AI is now a material cost line.

The homepage war was the clearest signal

The most revealing Salesforce move was not a funding round or an earnings recap. It was the homepage. On 27 Aug, the hero headline changed from “Get started with the #1 Agentic CRM” to “The #1 AI meets the #1 CRM.” On 5 Sep, it changed back to “Salesforce. The #1 Agentic CRM.” On 6 Sep, it returned again to “The #1 AI meets the #1 CRM,” with subtext updated to “Humans with Agents drive customer success together.”

That sequence says the positioning is still being tuned in public. “Agentic CRM” is the category-making phrase. “AI meets CRM” is the safer buyer-language phrase. Salesforce tested both within ten days, around the same period it was announcing Anthropic work, Agentforce momentum and regulated workflow tooling.

These timing gaps are the market’s useful tells. A press release says what a company wants to announce. A homepage reversal says what it is still deciding. For founders selling against Salesforce, the lesson is not that Salesforce lacks conviction. It is that even the category leader is still negotiating the line between agent hype and enterprise-legible AI.

HubSpot is building the connective layer around agents

HubSpot’s fortnight was quieter than Salesforce’s, but it was not passive. On 27 Aug, HubSpot launched an AEO tool for AI visibility in search. On 31 Aug, it added a product page titled “AI Instructions.” On 3 Sep, it launched a product page for a Claude connector, and press coverage discussed Agent Hub as a customer intelligence initiative. On 5 Sep, further articles covered HubSpot’s new AI Agent platform and its pricing model.

The commercial surface moved too. On 31 Aug, HubSpot added a new CTA, “Get a demo of HubSpot’s premium software.” On 4 Sep, it announced a Snapchat partnership to help businesses convert leads into customers and a TikTok integration to unify campaign management. That puts HubSpot’s AI push in a different lane from Salesforce. Salesforce is selling AI inside the enterprise operating model. HubSpot is tying AI to demand generation, campaign management and the practical conversion loop.

There were investor-facing notes as well. On 28 Aug, HubSpot’s stock rebounded 7.9% after earnings that exceeded expectations, with articles noting fresh institutional buying. On 3 Sep, reports said the CFO used 1,077 shares to cover RSU taxes, and on 7 Sep, further reports said CEO Yamini Rangan sold 2,622 shares for tax obligations while the CFO sold more than 1,000 shares on the same day. Those sales were reported as tax-related transactions, not strategic commentary.

The challengers are shipping around the edges

Attio looked like the fastest surface-level mover. On 27 Aug, it launched a ClickUp product page and removed an AndOpen page. On 1 Sep, it created new product pages for CallPal and MCP. On 2 Sep, it launched Super Link In and added a London Talent Partner careers page. On 3 Sep, it added Reply Flux. On 4 Sep, it introduced a 20% off annual promotion. That is a four-day run of app, product and packaging changes, ending with a price incentive.

Close used customer proof and hiring as its signal. On 28 Aug, it added Devvo as a case study, describing how Devvo turns dormant leads into an extra 85 car sales a month. On 29 Aug, it added Giveffect and Flatiron School case studies, including a Flatiron School story reporting a 75% conversion increase over 60 days after rebuilding its sales motion. On 1 Sep, Close launched a page for Android calling sound and stability improvements, and on 2 Sep it listed two remote US Senior Backend Engineer roles focused on building AI capabilities into the CRM to reduce manual data entry.

Pipedrive spent the period cleaning and localising its content estate. On 27 Aug and 28 Aug, it launched new Czech and German sales automation pages and removed older versions. On 2 Sep, it removed seven blog pages and five product pages, including pages around AI customer service and Black Friday marketing. On 3 Sep, it removed more CRM and sales and marketing resource pages, added a Videoboost customer case study, and released version 6.7.0 of its client-php library with new changelog endpoints for deals, persons and organisations, plus billing-related parameters. This was less headline-led than the AI announcements elsewhere, but more operational.

The real split is distribution versus workflow

The category is separating into two motions. Salesforce and HubSpot are trying to make AI native to the revenue platform and the customer acquisition system. Salesforce did it through Anthropic, Claudeforce, Agentforce, Data 360 and regulated document workflows. HubSpot did it through Agent Hub, AI Instructions, Claude connectivity, AEO, Snapchat and TikTok.

Attio, Close and Pipedrive are not matching that volume of platform news, but they are narrowing the surface area. Attio added a sequence of app pages and then a 20% annual discount. Close published conversion-led customer proof and opened AI backend roles. Pipedrive pruned older pages, refreshed localised sales automation pages and exposed more developer-facing changelog detail.

That split is important for operators. The large platforms are using AI to expand the centre of gravity. The smaller vendors are using shipping cadence, integrations, proof and pricing to stay specific. In this market, specificity is no longer a positioning preference. It is the defence against being absorbed into someone else’s AI platform story.