Skip to content

Season 2, Issue 1 18 November 2026

GTM Whisperer

One GTM market move each issue, and what to do about it. by Shashwat Ghosh

Harness, Headless and Humanness: The Re:found HubSpot

The company that named inbound marketing now calls that work chapter one, and its co-founder describes three ways into the product. What it means for B2B demand, and what you should rebuild first.

Is inbound marketing still working in B2B in 2026?

Not in its old form. Buyers still come in, but I think more of them now meet an AI answer first and arrive later as a buying committee. HubSpot's refounding on 12 November is the strongest signal yet. Nobody has that figured out for your buyers yet, so the checklist below starts with a test you can run this week.

What happened

On 12 November, co-founder Brian Halligan returned as HubSpot's CEO and called it a refounding. Its other co-founder, Dharmesh Shah, said it in one line: "20 years ago, it was about helping organizations leverage inbound marketing. Today it is about helping them leverage AI."

For a B2B marketer, that is the clearest sign yet that the inbound playbook needs rebuilding. The people who wrote it are rebuilding theirs.

The demand signal

I think fewer buyers now reach your site through the free search traffic that inbound was built on. HubSpot's outgoing CEO, Yamini Rangan, told analysts (transcript, page 4): "AI is changing how customers discover HubSpot and how they make purchasing decisions." Pipedrive's CEO wrote to his staff: "More small businesses are asking AI for recommendations when choosing software. As a result, the free traffic that powered our growth has declined sharply and the paid traffic we bought to replace it costs more."

When free traffic falls and paid traffic costs more, every lead from inbound gets more expensive. I think that holds for most companies that depend on it, not only for software vendors. HubSpot's own new customer numbers show the slowdown, though it has other causes too. Treat it as a signal, not proof, and check your own organic traffic and the pipeline it brings before you act.

HubSpot, second and third quarter 2026

Net new customers in the quarter

  • Q2 20267,000
  • Q3 20265,300

Source: HubSpot call transcript, page 15

Three doors: harness, headless, human

Four days into the refounding, Shah described HubSpot in three words: headless, harness and human. I read them as three doors a buyer can use to reach any company, including yours. Each door changes how demand reaches you, so I take them in order of how much they change it.

Harness: your CRM runs your demand

My reading of HubSpot's new tagline, "THE CUSTOMER RELATIONSHIP HARNESS": your CRM becomes the place AI agents run your demand from, so clean, consented customer data becomes the edge, not more content. Halligan shared the plainest explanation, written by Chris Tveter, who runs AIRops, a HubSpot partner. Tveter wrote: "Whichever AI wins, it plugs into that harness, works from that information, takes action on it, and stays within the rules you set."

For a marketer, that means the AI that follows up leads, scores accounts and writes outreach will only be as good as the records it works from. Duplicate contacts, stale fields and missing consent will make it fail, at speed and at scale. Start with two things: the fields your team scores and routes on, and a consent record for every contact you email.

Headless: your first visitor is an AI assistant

Headless means HubSpot used without its own screens. Other software, AI agents included, works with it directly. Shah's warning to software makers, on X: "If you aren't making your software product composable and callable by AI harnesses, you're going to miss this wave."

I think the same now applies to your marketing. A buyer asks an AI assistant which products fit, and the assistant reads the market for them. It can only use what it can read: facts, prices and proof on pages anyone can open. In my experience it skips forms and gated PDFs. Picture a pricing page with plan names and no number. A person might book a call to find out. The assistant moves on and reports "pricing not published" next to your name.

Human: the committee arrives last

Shah's third door is "Human: Thoughtful, classic interface to HubSpot", and he has said he will write more about it. My reading: people still make the final call, but they now arrive last, and as a group, holding a shortlist that the other two doors have already shaped.

So the job is to equip the committee, not to capture one lead. Rangan told analysts (transcript, page 4): "Win rates are nearly three times higher in deals with five or more buying committee members, when we engage multiple stakeholders." That is one company's data and a correlation, not proof. Still, a form captures one name, and a committee needs more. Finance needs a cost case, IT needs security answers, and the people who will use the product need proof from a similar customer. Give one champion a pack they can forward without rewriting it.

Reading it through EPIC

EPIC is my framework for choosing how a B2B company wins GTM. It rates four motions from 1 to 10: Ecosystem and ABM (E), Product-Led Growth (P), Inbound and Outbound (I) and Community-Led (C), and names the one to lead with.

If inbound is under pressure and the harness is still a big unknown, I would not wait for either to settle. I would let the other motions carry more of your demand for now.

I would put E first. Named accounts, partners and referrals put you in front of several people at the same buyer, and none of that depends on being named in an AI answer. A practical start is your 20 best-fit accounts and the three partners closest to them.

P comes next, where your product allows it. A self-serve trial or a live demo lets a buyer, or their AI, test you without a form.

C matters in categories where buyers ask peers before they ask vendors. I think peer trust is hard for an AI answer to fake.

I stays for the cheap fixes: pages an AI can read and quote, and the customer data underneath your CRM. Put more into it once you see how the harness plays out.

What I would not do

I would not cut content or swap it for SDRs, but I would not add more of it either. Its job changes from earning clicks to earning citations, and an assistant can only quote what you publish. The extra effort belongs in clean, consented customer data. Nor would I cut inbound's budget on this news alone. Keep it flat, and put the next hire or the next spend into E. Fix one page, then measure it for 90 days by how often AI assistants name you. The checklist and the sheet below are where to start.

Do this before Friday

This week's sheet

Agent-first inbound audit

What you end up with: one list of the eight places a buyer or their AI meets your company, a yes or no for whether an AI can quote each one, and the one fix you will make this month.

How to fill it

  1. Ask three AI assistants your buyer's category question, and note who they name.
  2. Open each of the other seven places as a buyer would. Mark yes if an AI could quote a clear fact from it, no if not.
  3. Pick one "no", give it an owner and a date, and recheck it in your weekly pipeline meeting.

On a phone the table shows three of its six columns. The download has all six.

PlaceWhat an agent or buyer finds todayCan an agent quote it (yes/no)OwnerFix this monthDate checked
AI answer to your category question
Pricing page
Product or docs page
Comparison or alternatives page
Case studies and proof
Review sites
Demo or contact route
Follow up pack for the committee

Example of one filled row (invented, to show the level of detail): Case studies and proof. Found today: three customer logos and a quote, no figures. Can an agent quote it: no. Owner: marketing head. Fix this month: add one result with a number to each case study page. Date checked: 18 Nov.

Download as CSVDownload as ExcelCopy for ChatGPT

When you are done, run EPIC with your deal source and channels

Also this week

Former Freshworks executives launched an AI-native CRM

Dreamteam, an AI-native CRM started by former Freshworks executives, came out of stealth on 16 November with $53M. Girish Mathrubootham's Together Fund was its first institutional investor, and the product offers "a first-class API and Model Context Protocol (MCP) server for AI agents".

Why I am watching it: that is the headless door, built into a new CRM from day one, so an agent can work with it directly instead of through its screens. It has only now come out of stealth, and I have no results to report yet. Keep a close watch on what it ships next and what its first customers say in public.

Source: Dreamteam press release, GlobeNewswire, 16 November 2026 | Business Standard, 16 November 2026

Freshworks moved its field sales to the faster business

Freshworks' newer business, software for IT and employee service (EX), is now larger than customer support (CX), the business it was founded on, and grows several times faster, Fortune reported. In May its CEO told staff: "Our Field team will be clearly focused on EX".

Freshworks, two businesses at two speeds

EX share of total ARR

  • At the 2021 IPO35%
  • 30 September 202659%

Source: Fortune, 3 November 2026

ARR at 30 September 2026, US$ millions

  • EX, IT and employee service$588M
  • CX, customer support$404M

Source: Fortune, 3 November 2026

ARR growth, year on year

  • EX, IT and employee service22%
  • CX, customer support4%

Source: EX: Fortune, CX: Futurum Research

What I would do: in that position, stop calling the original business the growth engine. Split the quota so no rep sells both, give the older business a profit target, and lead your message with the faster one.

Source: Fortune, Nick Lichtenberg, 3 November 2026 | Freshworks CEO note to staff, 6 May 2026

Explorium is shutting down. Check where your account data comes from

Explorium, a B2B data provider for enrichment and prospecting, is shutting down. Its notice to users says exports end on 2 December and data is deleted by 31 December.

Explorium shutdown: days after the 3 November notice

What stops, and when

  • Notice; automated billing stops (3 Nov)day 0
  • Platform and data exports end (2 Dec)day 29
  • All account data deleted (31 Dec)day 58

Source: Explorium notice to users

What I take from it: the risk sits in any account list, enrichment field or signal that lives in one supplier's system and nowhere else. When that supplier stops, your CRM keeps the old values but nothing refreshes them, and scoring and routing drift without anyone noticing. Three things I would do this month:

  • Export what you can while exports still work. For Explorium users that is 14 days after this issue.
  • Tag every enriched field with where it came from and the date.
  • Set a second-source rule: no field your team scores or routes on depends on one provider.

I am not naming a replacement. Choose on your own data, and test before you switch.

Source: Founders' farewell post | Notice to users

Canva folded its growth team into product

Canva is merging its Product, Growth and Platform teams into one product organisation, co-founder and CEO Melanie Perkins wrote on 4 November, because "our growth has slowed for the first time in a decade".

My reading: Canva does not link this to self-serve; that link is mine. In product-led growth, a separate growth team usually owns sign-ups, activation and upgrades. When AI gives a user a finished result inside the product, the path from free to paid runs through the product itself, and a growth team outside it has less to own. If you run a self-serve motion, check who owns activation and upgrade, and whether that one team can change the product to move them.

Source: Melanie Perkins' note, Canva newsroom, 4 November 2026

FAQ

What does headless HubSpot mean?

HubSpot used by other software, AI agents included, without opening its screens.

What is a customer relationship harness?

HubSpot's new name for itself: the place where a company's customer records sit, so that any AI can work from them within the rules the company sets.

Sources

Archive