If you feel like every SaaS product suddenly slapped “AI” on its homepage, you’re not imagining things.
But underneath the buzzword bingo, there is a real shift happeningespecially when you listen to the people actually running category-defining companies.
On a recent SaaStr deep dive, three heavy hitters shared how AI is reshaping their worlds:
Tooey Courtemanche, founder and CEO of construction-software giant Procore;
Eran Zinman, co-CEO of monday.com; and the leadership at HubSpot, where the board chair and executive team are steering a CRM and marketing powerhouse into an AI-first future.
What’s interesting isn’t just that all three are “using AI.” It’s how they’re using it,
what they’re refusing to do (for now), and where they think the real long-term value lies.
Spoiler: it’s less about clever chatbots, and more about data, workflows, and pricing models that don’t blow up your unit economics.
The Big Picture: From AI Feature to AI Fabric
Across Procore, monday.com, and HubSpot, AI is no longer treated as a sidecar feature.
It’s becoming part of the fabric of how customers plan projects, run revenue teams,
support clients, and make day-to-day decisions.
Consider a few themes:
- AI is embedded in workflows, not parked in a separate “lab” UI. Monday.com talks about AI “blocks” built directly into work managementtask assignment, risk detection, reportingso users don’t have to learn a new tool.
- Proprietary data is the real moat. HubSpot’s leadership points out that incumbents with rich CRM, marketing, and sales data have a serious edge: their AI doesn’t start from a blank page; it starts from years of customer behavior.
- Pricing is still experimental. Procore isn’t charging separately for AI yet, monday.com is betting on consumption-based AI pricing, and HubSpot is mixing per-seat with usage-based credits for its AI agents.
Put simply: AI is moving from gimmicky add-on to the invisible engine underneath the product.
Lesson 1: Procore and the Rise of Vertical AI
Procore lives in one of the most complicated and labor-constrained environments on the planet: construction.
You’re juggling contractors, subs, schedules, inspections, budgets, weather, and thousands of documentsoften on tight timelines and thin margins.
Tooey Courtemanche’s take is refreshingly blunt: construction doesn’t have enough skilled people to do everything the industry demands.
AI, in his view, isn’t a “nice-to-have”; it’s a survival tool that lets teams do more with the people they already have.
Procore’s AI efforts focus on tasks that are painful, repetitive, and high-risk if done poorly, such as:
- Reviewing submittals and RFIs automatically, flagging risk or missing information before it hits the field.
- Scanning and organizing drawings, specs, and change orders so project teams can actually find what they need.
- Predicting schedule slippage or cost impact based on historical project patterns and current site data.
Notably, Procore isn’t rushing to carve out a separate AI SKU and meter every prediction.
For now, AI is mostly bundled into the existing value proposition, with the expectation that once customers see measurable impact on margin and risk,
there will be room for more explicit AI-driven line items.
The broader lesson for SaaS founders: in vertical markets, AI that quietly absorbs complexity and labor will win over flashy features every time.
Your AI roadmap should probably start with, “What are people constantly overworked and behind on?” not “How do we add a chatbot?”
Lesson 2: Monday.com, Unstructured Data, and Automation 3.0
Monday.com has become a poster child for AI-at-scale in horizontal SaaS.
Co-CEO Eran Zinman has shared that the platform has already logged tens of millions of AI-driven actions,
and the most popular use cases aren’t necessarily glamorousthey’re about taming ugly, unstructured data.
Think about what lands in a typical work management system:
- Resumes by the thousands.
- Invoices, purchase orders, PDFs from vendors.
- RFPs, contracts, and status reports in mixed formats and random folders.
Monday’s AI shines in turning that mess into structured, queryable informationscreening CVs, parsing documents, tagging items,
and letting teams search and act on data that used to be buried in attachments.
On top of this, monday.com has leaned hard into an AI-first product portfolio:
AI-infused work management, CRM, service management, and dev tools; “digital worker” concepts; and AI agents that can assign people, detect risks,
and resolve requests without constant human babysitting.
The pricing decision is especially interesting. Rather than charging per seat for AI, monday.com ties many AI capabilities to consumption
(e.g., number of AI actions), which aligns better with the reality that successful automation might actually reduce hiring in some areas.
That creates a cleaner story for customers: you pay for the work the AI is doing, not the number of humans you didn’t hire.
Their momentum shows up in the numbers: AI-powered products and features are cited as a key driver in monday.com’s revenue growth,
partner program expansion, and even stock performance, as investors increasingly shift interest from pure AI infrastructure to AI-enabled software.
Lesson 3: HubSpot, Data Moats, and AI Agents
HubSpot sits in the beating heart of go-to-market data: emails, calls, meetings, website visits, deals, tickets, campaignsyou name it.
That’s a goldmine for AI, and HubSpot’s leaders know it.
The “Chair of HubSpot” framing around this SaaStr discussion points back to Brian Halligan’s long-running view: in AI, both big incumbents and startups can win
but only if they have real distribution or real data, preferably both.
In practice, HubSpot’s AI push has three big pillars:
-
Assistants embedded everywhere. Tools like Content Assistant and ChatSpot help draft emails, blogs, reports,
and even CRM updates directly from natural-language promptsspeeding up sales and marketing workflows without forcing users to leave the HubSpot interface. -
Agentic AI that actually does work. More recent capabilities, such as Breeze AI Agents, can handle tasks like prospecting,
responding to customers, and filling gaps in knowledge bases, essentially acting as specialized digital teammates for small businesses. -
AI built on top of a connected data hub. At INBOUND and beyond, HubSpot has emphasized that AI is only as smart as the data it sees,
so they’re investing in unified data infrastructure and hybrid pricing models that combine seat licenses with usage-based credits for high-value AI agents.
The board and leadership structure matters here too: HubSpot’s chairperson and directors include operators with deep enterprise, SaaS, and AI experience,
and the company is increasingly positioning itself as a platform where AI can safely act on sensitive go-to-market data with strong governance,
permissions, and trust controls.
What All Three Agree On: The New Rules of AI in SaaS
Procore, monday.com, and HubSpot sell very different products, but their leaders are surprisingly aligned on several rules of the AI era:
1. AI Must Remove Friction, Not Add It
AI features that require users to adopt a totally new workflow usually fizzle.
Monday.com’s most successful AI work sticks to things customers already dolike parsing PDFs or assigning tasksand simply makes them 10x faster.
Procore and HubSpot follow similar logic: AI gets dropped into existing processes (reviewing submittals, logging CRM notes, answering tickets),
not launched as a separate “AI mode” the user has to remember to turn on.
2. Proprietary Data is the Differentiator
Foundation models are increasingly commoditized. What isn’t commoditized is the vertical and customer-specific data Procore has on construction projects,
monday.com has on cross-functional workflows, and HubSpot has on deals, contacts, and campaigns.
That data is what makes their AI predictions, summaries, and agents actually usefuland it’s a big reason incumbents can still win in an environment
where startups move fast but don’t yet have deep data moats.
3. Pricing and Value Are Moving Targets
No one has fully cracked AI pricing:
- Procore is essentially “giving away” AI for now to accelerate adoption and prove value.
- Monday.com is betting that consumption pricing will age better than per-seat, especially as AI shrinks team sizes in some workflows.
- HubSpot is running a hybrid model with seats plus credits and is tracking what percentage of work their AI agents can reliably handle (in some cases, more than half of tickets).
For SaaS founders, the message is: expect to iterate. Start with simple, transparent pricing, measure usage and outcomes obsessively, and be ready to evolve
as AI moves from “premium” to “table stakes.”
4. Humans Still Matter (Sometimes Even More)
All three companies position AI as an amplifier of human teams, not a replacement.
Salesforce, HubSpot, and others talk openly about “hybrid workforces” where AI agents handle the heavy admin and pattern-matching,
while humans stay focused on relationships, creativity, and judgment.
That’s not just good PRit’s a practical necessity. The more AI takes on high-leverage actions (like replying to customers or editing construction schedules),
the more you need humans to design guardrails, review exceptions, and decide where automation stops.
What This Means If You’re Building or Buying SaaS
Whether you run a startup or you’re deciding which tools to roll out across your company, the combined perspective from Procore, monday.com, and HubSpot
points to a few practical moves.
Start With One Painful Workflow
Don’t try to “AI-ify” your entire product in one release. Instead:
- Find the most hated repeated task (invoice classification, lead routing, status reporting, ticket triage).
- Instrument it: how long it takes, error rates, how often it blocks other work.
- Build or buy AI that directly attacks that workflow and ships inside the tools your team already uses.
This is essentially monday.com’s playbook for automation 3.0 and Procore’s approach to construction workflowstight focus, clear value, visible time savings.
Design Around Your Data, Not Around a Demo
The slickest demo in the world falls apart if your data is scattered, incomplete, or stuck in legacy systems.
HubSpot’s push toward connected data hubs and context-aware AI is a reminder: AI needs clean, unified data to be useful.
Before shipping big AI features, ask:
- What proprietary signals do we already have that others don’t?
- How do we centralize and normalize that data?
- What’s the smallest AI feature that would prove our data advantage?
Build Trust, Not Just Features
As AI gains more autonomyespecially in agentic models that send emails, update records, or make recommendationscustomers will care as much about
permissions, transparency, and audit trails as they do about speed.
HubSpot’s emphasis on trust, governance, and user control is very likely a preview of where enterprise buyers will set the bar.
That means your roadmap shouldn’t just have “auto-reply to customers” as a bullet pointit should include “show exactly what the AI did and why.”
Real-World Experiences: Bringing These AI Lessons Into Your SaaS (Extra Deep Dive)
To make this less abstract, let’s look at how teams inspired by Procore, monday.com, and HubSpot are actually putting these ideas into practice.
Experience #1: A Vertical SaaS Startup Stops Drowning in Documents
Imagine a startup building workflow software for specialty clinics. Their customers manage piles of referrals, lab results, insurance forms, and visit notes.
The founder originally thought “AI” meant a cute chatbot on the homepage. After hearing the Procore story, the team reframed the problem:
the real value was in never losing track of a critical document again.
They rolled out an AI-powered intake pipeline:
- Every uploaded PDF is auto-classified (referral, lab result, insurance appeal, etc.).
- Key fieldspatient, provider, service, datesare extracted and normalized.
- High-risk items (e.g., time-sensitive appeals) are flagged and escalated.
They didn’t charge a separate “AI fee” at first. Instead, they measured reduction in lost documents and time-to-schedule.
When customers saw tangible wins (fewer angry calls, faster reimbursement), the startup had the credibility to introduce a higher-tier plan
with more advanced AI automationmirroring Procore’s “prove value first, price later” mindset.
Experience #2: A Mid-Size SaaS Company Rethinks AI Pricing
A 300-person SaaS company in HR tech watched monday.com’s shift to consumption-based AI and realized their own per-seat AI pricing was working against them.
Their customers wanted broad automation, but every time the AI saved enough time to reduce headcount,
the buyer had to justify why they were still paying per “AI seat.”
Inspired by monday.com’s approach, they:
- Moved from per-seat AI pricing to bands of monthly AI actions.
- Bundled a reasonable amount of AI usage into every standard plan.
- Positioned higher tiers as “automation expansion,” not “more licenses.”
Churn among AI adopters dropped, even though some customers reduced employee licenses.
Net revenue still grew because the perceived fairness of pricing improved, and AI became something customers wanted to expand, not ration.
Experience #3: A GTM Team Builds a Human-Agent Hybrid Stack
A B2B SaaS sales team already using HubSpot noticed a familiar pattern: reps were spending more time updating records than talking to humans.
Drawing on HubSpot’s AI agents and broader industry conversations, they created a “human-agent” hybrid motion:
- AI assistants drafted outreach emails and suggested follow-ups based on CRM history.
- Agentic tools handled routine ticket responses and knowledge-base lookups.
- Reps focused their energy on discovery calls, negotiation, and expansion opportunities.
After three months, their metrics looked a lot like the trend lines in HubSpot’s own AI reports:
more pipeline per rep, shorter sales cycles, and higher CSAT, with fewer nights spent catching up on admin.
The biggest lesson from this experiment wasn’t technicalit was cultural.
The team had to explicitly define which tasks belonged to agents and which were “human only.”
Once that line was clear, adoption soared, and the fear of “AI taking jobs” was replaced by “AI taking the worst parts of the job.”
Taken together, these experiences echo what the leaders of Procore, monday.com, and HubSpot are saying in their own ways:
AI in SaaS works best when it’s focused, embedded, fairly priced, and built on real data and trust.
Everything else is just noise.
