INBOUND 2025: AI Won’t Save Your Messy CRM — But This Will
Everyone at INBOUND 2025 was buzzing about AI. HubSpot unveiled 200 new features, moving AI from “front-end magic” into the backbone of the CRM. It was dazzling.
But here’s the inconvenient truth: AI is useless if your CRM is still a mess. Yamini Rangan, CEO of Hubspot, herself acknowledged it — and HubSpot doubled down on foundational CRM tools because without a clean core, you’ll never unlock AI’s full potential.
200 Shiny New Features, Shift to Foundational Use Cases
Hubspot product & engineering team’s velocity is impressive - seems to keep improving! A few highlights:
AI Field Creation: auto-generates fields from internal + external data, cutting manual work.
Data Hub: CDP-like data unification (a la Autopilot / Ortto), with duplicate logic, scoring, and visibility into weak spots.
Campaign + Segment Builder: finally flexible enough to match how you want to work.
Buying Groups + CPQ + AEO Reporting: more flexible quoting, plus tools that reflect how companies actually buy, not how we wish they did.
Hybrid Human + AI Teams: Breeze Agents and other AI assistants tackling admin work that wasn’t possible before LLMs.
These are real leaps forward. But HubSpot is also making a statement: it’s investing as much in the foundation—data, processes, system architecture—as in the advanced features, openly acknowledging that everything else depends on it.
“My CRM is a Mess…” — And That’s Okay
I led user groups titled exactly that. Dozens of CRM leaders showed up and admitted their systems weren’t working. Brave? Absolutely.
Why? Because it’s easier to bask in the glow of AI hype than to say out loud: my CRM is broken.
And yet, those are the folks I’m betting on. They’ll do the hard, unglamorous work of cleanup and process change. They’ll defy the odds — even though 50%–70% of CRMs still fail to deliver real business value (Forrester, HBR, CIO Magazine, Salesforce).
Why CRMs Fail (and How to Defy the Odds)
From our sessions, the mess clustered into two big dimensions:
½ human issues — adoption, ownership, discipline.
½ data & integration issues — duplicates, incomplete fields, conflicting sources, brittle connectors, half-finished syncs.
Within that, three themes came up again and again:
Human & Buy-In: “What’s in it for me?” matters at every level. Reps want comp and workflow alignment; executives want a business case to justify investment. Without both, systems limp along. The fix: tie metrics to each audience. Sometimes that means painful manual work first to prove the value before resources get unlocked.
Data Quality & Structure: Messy data doesn’t just pollute reports — it breaks workflows, clogs automations, and erodes trust. Patterns included missing fields, broken parent/child hierarchies, and “data remorse” for properties no one will ever use. The fix: simplify ruthlessly. Keep only what matters, archive the rest, enforce governance with dashboards, and lean on HubSpot’s new Data Hub or a Master Data Management (MDM) solution for hierarchies and control.
System Integration & Source of Truth: A CRM without revenue or service data is just a partial view. Teams debated how to bring ERP, OMS, or homegrown data into HubSpot (or Salesforce, Sugar, etc.) reliably. The consistent lesson: identify the few true source systems, then build durable, bidirectional syncs so Sales and Service aren’t left flying blind.
The Twist
Even when problems looked “human,” the fixes often came from technology: dashboards to surface usage gaps, automation to enforce standards, integrations that make the right path the easy path.
It’s never purely human vs. technical. It’s both — and the companies that recognize that are the ones who beat the odds.
The Human Stakes of AI
Zach Kass and Dharmesh Shah didn’t just talk about features — they raised questions that cut to how we define ourselves:
Will AI erase jobs, or force us to rethink the role of work in our identity?
Will automation let us focus on our greatest talents — or just raise the bar higher?
Technology has already tied identity more tightly to work through adoption and wealth creation. If AI keeps advancing, will it loosen that grip — shifting meaning back to family, hobbies, and passions outside of work?
Both Zach and Dharmesh have motivation to influence how these questions play out, but they also bring a depth of exposure and insight few others have. That makes their perspective worth listening to — not because they hold the answers, but because they help shape the questions.
The takeaway wasn’t neat or reassuring. It was thought-provoking, even unsettling. And that same tension shows up in CRM: human adoption and technical design constantly pulling against each other. The tools will keep advancing. Success still depends on how people adapt and define what “good” looks like.
So What Now?
AI is arriving fast. But no amount of automation matters if your CRM is messy. The companies that win will be the ones that get the foundation right and lean into AI. One without the other is wasted effort.
That’s why every successful CRM transformation follows the same arc:
Audit — map processes and surface the root causes, not just the symptoms.
Roadmap — prioritize fixes and investments against what matters most: revenue, impact, engagement, cost, risk, and feasibility
Build — implement with agility on a solid framework, minimizing throwaway work and turning every sprint into a force multiplier.
Final Word
At GMA Growth Engines, we live in this tension every day: cleaning up CRM tech stacks while building growth engines that actually work.
If your CRM feels more like a roadblock than a launchpad, check out our Force Multiplier article — and let’s talk about making AI work for you, not against you.
