SandsDX
Back to Research
AI & Automation7 min read

What a Customer Harness Is, and What HubSpot's Refounding Means for B2B Revenue Teams

A customer harness is the software around AI agents that act on customer relationships. What HubSpot's November 2026 refounding means for revenue teams.

Page Sands ·

Fifteen years building B2B SaaS revenue systems: a Microsoft preferred vendor, Drift, Avalara, Blackbaud, ConnectWise. About →

A customer harness is the software around AI agents that act on customer relationships: the customer data they read, the tools they use, the rules and approvals that govern them, and the record of what they did and whether it created revenue.

The term went mainstream on November 12, 2026, when HubSpot named co-founder Brian Halligan CEO again and called it a refounding. His description of the plan, from HubSpot’s announcement: “Chapter one was creating the inbound marketing category… Chapter two was building an easy and powerful CRM platform… Chapter three is building the customer relationship harness that will enable millions to grow better.”

This piece covers what a harness is, what a customer harness adds, and what changes for the people who run pipeline.

What a harness is

A model on its own predicts text. It cannot send an email, update a record, or remember last week. A harness is the software that wraps the model and makes it an agent that can do work.

LangChain’s definition is the cleanest: Agent = Model + Harness. The harness is every piece of code, configuration and execution logic that is not the model. Microsoft’s documentation lists what it does: it drives model and tool calls, manages state and context, applies approval policies, and keeps a multistep task moving.

In practice a harness has five parts:

  1. Context. What the agent can read on each turn: records, documents, prior steps.
  2. Tools. What it can act with: send, update, route, search.
  3. Rules. What it may and may not do, and when a human must approve.
  4. The loop. How a task keeps going across steps until it is done or stopped.
  5. The record. Logs of what it did, so someone can check it.

What a customer harness adds

A customer harness is a harness pointed at customer relationships. The context is your CRM: accounts, contacts, deals, conversations. The tools are the things revenue teams do every day: outreach, follow up, routing, qualification, renewal. The rules decide which of those actions an agent takes alone and which wait for a person.

The fifth part matters most for revenue leaders. A customer harness should show whether the agents created pipeline. Activity is easy to count. Opportunities created, meetings held and win rate are the numbers that tell you the harness works.

What changes for revenue teams

The CRM was a system people logged work into. In a harness, it becomes the system agents work from. Three consequences follow.

Your definitions become instructions

An agent decides who to contact from your ICP fields, when to escalate from your stage definitions, and how to credit a deal from your lead source rules. Where those are vague, agents act vaguely. Where they disagree between teams, agents act on whichever version they read.

Your story gets repeated at scale

Agents write from the messaging they find in your systems. Gartner’s 2025 survey of 632 B2B buyers found 69% already see inconsistencies between a supplier’s website and what its sellers say. An agent reading three versions of your positioning will send all three. When agents send that copy, the repositioning lag reaches buyers faster and in larger volume.

Buyers already do their research with AI

The same Gartner survey found 61% of B2B buyers prefer a rep-free buying experience, and Forrester’s 2025 Buyers’ Journey Survey found 94% of business buyers use AI in their buying process. Buyers do much of their research with AI before they contact a vendor. A customer harness gives the seller agents for the same stages of the deal.

What to check before you buy one

I run my own consultancy on a harness, described in How I run SandsDX on a GTM harness. Most of the work in building it went into writing down the rules and definitions the agents follow. Connecting the model took little time.

Before evaluating any vendor’s harness, check four things in your own stack:

  • One written ICP. Can you hand an agent a definition specific enough to build a list from?
  • Stage and source definitions that sales, marketing and RevOps agree on, with entry and exit criteria.
  • An approval rule for anything that reaches a customer or changes a deal.
  • A pipeline measure that can show whether agent work created opportunities, not just activity.

The full question list for vendors is in How to evaluate a customer harness.

HubSpot’s own repositioning lag

HubSpot now has the problem its customers have: a new story to carry across every surface buyers read. I measured how many of its public surfaces tell the customer harness story five days after the announcement, in HubSpot’s customer harness story, five days in.

Frequently Asked Questions

What is a customer harness?

A customer harness is the software around AI agents that act on customer relationships. It gives the agents customer data to read, tools to act with (email, CRM updates, routing), rules and human approvals that govern what they may do, and a record of what they did and whether it produced pipeline or revenue. HubSpot's returning CEO Brian Halligan used the phrase "customer relationship harness" on November 12, 2026 to describe the company's next chapter.

What is an agent harness?

An agent harness is everything around an AI model that turns it into an agent that can do work: tool execution, memory and context, the loop that keeps a multistep task moving, permissions, and feedback. LangChain summarizes it as Agent = Model + Harness.

What does HubSpot becoming a customer harness company mean for my revenue team?

It signals that the CRM is becoming the place where agents run as well as where people log activity. For revenue teams the practical effect is that the quality of your CRM data, your written definitions (ICP, stages, lead source) and your approval rules start to decide what agents do on your behalf. Teams with inconsistent definitions will get agents that act inconsistently.

Do I need a customer harness?

Only once your definitions are written down and your data can be trusted. A harness runs on your ICP, stage definitions, routing rules and messaging. If those disagree across systems, an agent will repeat the disagreement at higher volume. Fix the definitions first, then let agents act on them.

Put this to work on your GTM

Thirty minutes. Bring the number you need to move and I'll tell you what I'd check first in your system.

Schedule a Conversation

Not ready to talk? Run the AI Visibility Baseline on your company →