Written by
Patricia Diaz-Hymes
Founder & Principal Product Marketing Manager
On Oct 6, Sierra announced the Personal Agent Protocol, an open standard it is developing with Meta and partners including Genesys, Shopify, Stripe and Walmart, covering how a consumer’s personal AI agent identifies itself and gets things done with a business. On Oct 9, Sierra published a draft, nicknamed Poppy, and named 35 more design partners, among them OpenAI, Visa, Mastercard, Bank of America, Notion, Zapier, Zendesk, Okta and Klaviyo.
The draft starts with one file. A company publishes it at a standard address on its website, and it tells personal agents how to interact with the company, how customers sign in, and what an agent is allowed to do. An agent can begin as a guest, which Sierra says may be enough to check stock or a returns policy, and the customer signs in through OAuth when a task needs their account. The company then picks the route: its regular website, its APIs (built on standards such as MCP and OpenAPI), or its own agent for tasks that need a conversation. Meta’s own announcement, as reported by CMSWire, argues that a personal agent shouldn’t have to interpret a support article when it can ask a business’s agent directly.
What it means for a GTM team. Every example so far is consumer (returns, flight rebooking, a mortgage pre-approval), and nothing yet shows B2B adoption. Even so, the draft forces a set of decisions into the open that usually sit in nobody’s job description: what an anonymous visitor can learn without filling out a form, which tasks an outside agent may complete, and whether it deals with your site, your API or your own agent. Worth checking now: what your pricing, packaging and policy pages tell a guest; whether your help content answers questions plainly enough for a machine to use; and who in your company would decide what goes in a file like this if engineering shipped it tomorrow. Sierra says design workshops and a reference implementation come over the next month.
HubSpot reorganizes its product teams around outcomes instead of Hubs
On Oct 6, HubSpot CEO Yamini Rangan told employees the company is cutting about 7% of its workforce, nearly 660 people, and reorganizing product teams around four customer outcomes (generate demand, win deals, delight customers, scale growth) rather than its Hubs, according to the full message as reproduced by PPC Land and Boston.com’s report. Rangan wrote that HubSpot has moved from building software that helps customers grow to delivering outcomes for them with AI, and that the cuts are not driven by AI-related efficiencies.
What it means for a GTM team. The message doesn’t say Hubs will be renamed or repriced. Still, HubSpot customers should expect features built by outcome-based teams to cross the old Hub lines more often, and it’s fair to ask your account team how that will show up in packaging at renewal. For any PMM team, Rangan’s line that customers don’t think in Hubs and features works as a test of your own site: does the navigation follow your product lines, or what a buyer is trying to get done? If PMM is staffed by product, check which buyer outcomes have no clear owner.
Claude moves into Google Docs, Sheets and Slides
On Oct 6, Anthropic released Claude for Google Workspace in public beta on all paid Claude plans: a sidebar in Docs, Sheets and Slides, plus connectors for creating and editing Google files from the Claude chat. In Slides it builds from the deck’s theme and layouts, and it shows suggested edits and new slides before you accept them. In the sidebar, Claude sees only the open file and any connectors you turn on; working from Claude’s own chat, it can reach any file your Drive permissions allow. Teams can save preferences and turn a working process into a skill that runs the same way across the three apps.
What it means for a GTM team. Gemini already sits natively in Workspace, so for many teams this adds a second assistant to the same files. For anyone who builds decks, one-pagers and pipeline sheets in Google, production gets faster and the inputs carry more weight. Worth checking: whether your slide master is current, which “pitch deck” a Drive search turns up first, and whether there’s one approved messaging doc you’d want an assistant to read before anything else.
How this week connects. All three stories put a machine in front of material your team already wrote: the policies a customer’s agent reads, the outcome language a vendor now organizes around, and the templates and files an assistant builds from.
My take
A personal agent that reaches your site under Poppy will read a file somebody at your company wrote. Claude, sitting in your Slides, will build from whatever template and files are in your Drive. And HubSpot just reorganized its product teams because, in its CEO’s words, customers don’t think in Hubs and features.
Every time I’ve been the first product marketer in the room, the first real job was getting a handful of decisions made out loud: who we’re for, what a prospect should be able to learn before talking to sales, which story we lead with when the buyer wants an outcome rather than a SKU. Those answers usually lived in a few people’s heads and in a dozen decks that each told a slightly different version (that’s GTM debt in its most ordinary form, and plenty of companies with smart, busy marketing teams carry it).
What’s new is that machines now need those answers written down, in one place, before they can act on your behalf. The teams that already did that work will plug into all of this quickly, and everyone else will learn, one agent at a time, which version of their story got picked up.
Deciding what you’d let an agent say and do for a customer is the positioning work most of us have been meaning to finish anyway.
