AI adoption is moving quickly, but for many teams, the biggest challenge is not a lack of tools.
It is uncertainty.
Teams are asking practical questions:
- Can we use AI if our CRM data is not perfect?
- Where does our customer data go when we use AI tools?
- Should we start with AI agents, or something smaller?
- How do we make sure AI helps our team instead of creating more work, more risk, or more bad outreach?
These are the right questions to ask.
In a recent conversation, Measured Results CEO Christopher Antonopoulos and Nathanael Yellis, Senior Solutions Architect at HubSpot, discussed what is really holding teams back from AI adoption and how organizations can move forward in a more practical, lower-risk way.
The core takeaway: AI enablement should not start with the technology. It should start with the business objective, the process, and the data required to support a useful outcome.
That means getting clear on three things before jumping into AI agents or automation:
- The use case you are trying to improve
- The data required to support that use case
- The guardrails needed to reduce risk and protect trust
1. What’s Holding Teams Back From AI?
Many teams delay AI adoption because they believe they need to clean up everything first.
That instinct makes sense. If the CRM is messy, workflows are outdated, or data lives in disconnected systems, it can feel risky to introduce AI into the mix.
But waiting for a perfect CRM cleanup can also become a blocker.
The better starting point is often smaller and more focused. Instead of trying to “clean out the whole garage,” teams can identify a specific segment, use case, or process where AI can create value with the data already available.
There is also a second, more structural blocker: disconnected business data.
If HubSpot or another CRM is only being used as a point solution, AI may not have enough business context to produce useful outputs. For example, if customer status, lifetime value, renewal data, or sales activity are not connected, AI will have a harder time helping teams identify the right leads, customers, or next steps.
Key takeaway: You do not need a perfect CRM to start with AI, but you do need a focused use case and enough connected data to support it.
2. How to De-Risk AI Adoption
AI risk is real.
Teams are right to be cautious about exposing confidential data, creating unintended automation, or allowing AI to take action without the right oversight.
The answer is not to avoid AI completely. The answer is to start in the right place.
“You don’t start with the tech solution, you start with the business context and the business process.”
A lower-risk AI deployment begins with business context:
- What objective are we trying to improve?
- What process already exists today?
- Where could AI responsibly accelerate or improve one step?
For example, a company trying to improve customer renewal rates does not need to start by deploying a fully autonomous AI agent. It might begin with a standard renewal workflow, NPS survey process, or customer health scoring approach. Then AI can be layered in to help analyze patterns, summarize risk signals, or surface accounts that need attention.
Key takeaway: Do not start by asking, “What can we feed into AI?” Start by asking, “What business process are we trying to make better?”
3. Use AI as a Librarian for Your CRM
One of the most practical ways to start with AI is to use it as a research assistant or “librarian” for your CRM.
Instead of asking AI to take over a process, ask it to help uncover patterns that would be difficult or time-consuming for a person to find manually.
“Use a tool as a librarian to do the research.”
For example:
- Which customers canceled last quarter?
- What attributes did those customers have in common?
- Were there changes in survey responses, payment behavior, support tickets, or engagement before cancellation?
- Are there warning signs we can use earlier in the customer lifecycle?
This is where AI can help teams get to the starting line faster. It can help surface patterns, summarize large volumes of information, and point teams toward better questions.
It does not replace strategy. It supports it.
Key takeaway: AI can be valuable before it ever automates a customer-facing action. Start by using it to analyze, summarize, and surface insight.
4. Not All AI Is an Agent
One of the reasons AI conversations become confusing is that people often use the word “AI” to describe very different things.
In business software, AI can show up in several ways.
First, AI can act as a system interface assistant. This might mean asking questions about CRM data, surfacing records, or helping users navigate information more quickly.
Second, AI can act as an accelerator. This includes things like record summaries, email drafts, call note summaries, or agenda generation.
Third, AI can act as an agent. This is where AI begins performing parts of a process semi-autonomously or autonomously, such as answering support questions, researching prospects, or generating outreach.
That third category has the highest potential reward, but it also carries the highest risk.
For most teams, the best starting point is not a fully agentic workflow. It is usually a smaller AI feature or assistant-level use case that proves value, builds trust, and creates a foundation for larger deployments later.
Key takeaway: You do not have to start with AI agents. Start with the level of AI that matches your process maturity, data readiness, and risk tolerance.
5. Where Does Your AI Data Go?
Data governance needs to be part of the AI conversation from the beginning.
Before introducing a new AI tool, teams should understand:
- Whether customer, financial, or confidential business data is protected
- What data the tool can access
- Where that data is stored
- Whether data is used to train external models
- What agreements are in place with AI providers
- Whether customer, financial, or confidential business data is protected
“Knowing where the data is stored sounds like a pretty basic question. But frankly, it’s wide open.”
This is especially important when teams adopt smaller AI tools, browser extensions, call recorders, productivity apps, or external platforms without fully understanding how data flows through them.
AI creates opportunity, but it also creates new connection points. Those connection points need governance.
The same discipline that companies applied during the shift to cloud software now needs to be applied to AI adoption: clear controls, approved tools, security review, and internal guidelines for how company data can and cannot be used.
Key takeaway: AI enablement is not just a marketing, sales, or operations initiative. It also requires data governance and security alignment.
6. Don’t Use AI to Fake Sincerity
AI can help teams move faster. But speed is not the same thing as value.
One of the clearest risks in sales and marketing is using AI to create fake personalization at scale.
Messages that claim “I read your article,” “I watched your video,” or “I loved your case study” can quickly damage trust if they are not true. AI can make those messages easier to produce, but that does not make them more effective.
“Don’t use this tooling to lie.”
Poor outreach has always existed. AI simply makes it easier to do poor outreach faster and at a larger scale.
The goal should not be to use AI to pretend a message is personal. The goal should be to use AI to support relevance, consistency, research, and timing while preserving the sincerity of the interaction.
Key takeaway: Do not use AI to manufacture false familiarity. Use it to create more relevant, respectful, and useful interactions.
The Bigger Lesson: Start With the Outcome, Not the Tool
AI adoption works best when it is tied to a real business objective.
That might be:
- Improving renewal rates
- Prioritizing sales outreach
- Reducing time spent on repetitive service tickets
- Finding patterns in customer data
- Improving segmentation
- Summarizing CRM activity
- Supporting better campaign planning
- Creating more consistent sales or support processes
The common thread is that AI should be applied to a defined business problem, not introduced as a disconnected experiment.
When teams start with the tool, they often end up “death scrolling” through prompts, models, and possibilities without a clear result.
When teams start with the business objective, AI becomes easier to evaluate:
- Did it save time?
- Did it improve consistency?
- Did it surface insight?
- Did it reduce manual effort?
- Did it help the team make a better decision?
- Did it improve the customer or prospect experience?
That is where AI enablement becomes practical.
How Measured Results Helps Teams Approach AI Enablement
Measured Results helps organizations identify where AI can create practical value across their CRM, marketing, sales, service, and revenue operations processes.
That work starts with questions like:
- What business outcome are we trying to improve?
- What data is required?
- Where does that data live today?
- What risks need to be addressed?
- What AI features or agents are appropriate for this use case?
- What guardrails, training, and monitoring need to be in place?
- How will we measure whether AI is actually helping?
The goal is not to use AI for the sake of using AI.
The goal is to help teams move faster, make better decisions, improve consistency, and reduce manual effort without creating unnecessary risk.
Ready to explore AI enablement for your team?
Measured Results can help you identify the right use case, data requirements, and activation path so your team can start with AI in a practical, responsible way.
