Growth AI Blog

Before You Buy Another AI Tool, Use the Stack You Already Have

Erica Salm Rench — Wed, 04 Nov 2026 16:29:55 GMT

How many browser tabs do you have open right now? 20? 30?

I see you. We've all been there.

And we’ve all got a version of this. A browser that looks like a cry for help. A new “AI-powered” app for every task, because surely the answer is one more tab away. (Right??!!!)

But something keeps coming up on calls with marketing and association leaders: teams are searching for capabilities they may already have. The software is there. The subscription is active. The AI features are available. Yet the actual process still runs on exports, copy-and-paste, and someone remembering to follow up.

Before buying another AI tool, it’s worth asking: What could we do with the stack we already have?

The “we never turned it on” gap

Sometimes the adoption problem really is a setting. More often, “we never turned it on” means nobody has made the feature part of how the team works.

A person tried it once. The output was generic. Nobody knew which information it could access. The team went back to the familiar process, and the feature became one more thing they were technically paying for.

HubSpot is a useful example. Its AI settings control access to generative features and the data those features can use. Some AI settings are even on by default. So the right question is broader than “Did we flip the switch?” It’s “Can our people use this, with the right context, to finish a real task?” HubSpot’s AI settings guide explains those controls.

Availability still matters. Specific capabilities can depend on subscriptions, permissions, configuration, and HubSpot Credits. “Built in” doesn’t automatically mean everything is included in your plan. HubSpot’s adoption guide outlines those requirements.

That’s a reason to investigate your account before shopping for another one.

Start with the process that keeps eating your afternoon

“We should use more AI” is a difficult assignment. “We need to get useful follow-up to webinar attendees without spending an afternoon assembling it” gives a team somewhere to start.

Pick one recurring task with a clear beginning and end. For a marketing or membership team, that might be event follow-up, preparing for a sponsor conversation, drafting a renewal message, or pulling together a weekly engagement summary.

For a sales team, it could be finding people in your ICP not yet in your database, automatically drafting personalized outreach to those people, and then automatically sending follow ups after that initial outreach. (Plus adding them to your database, LinkedIn profile and all.)

For architecting additional time saving and transformational workflows: watch how the work actually happens. Where does someone export a spreadsheet? Where do they copy information between systems? Which questions require three tabs and a Slack message? Where does a finished draft sit because no one owns the next step?

Those details tell you much more about the opportunity than a list of AI features.

Audit your existing stack before adding to it

A useful review can start with five questions:

What outcome are we trying to improve? Choose something you can observe, such as faster follow-up, fewer manual handoffs, or more relevant member conversations.
What can our current tools already do? Check the actual subscription, available AI features, permissions, usage costs, and existing integrations.
What information is missing? Identify the records, definitions, approved content, and organizational context needed to produce useful work.
Who will own the process? Name the person who reviews output, handles exceptions, and makes sure the work gets finished.
How will we know it helped? Compare the current process with a small pilot. Measure total effort, including review and corrections, alongside the quality of the result.

If drafting becomes faster but checking the output takes longer, you need to know that. If the team saves time but the follow-up becomes less relevant, that matters too.

A good pilot gives you evidence for the next decision. It may show that an existing feature is enough. It may reveal a specific gap that makes another tool worth buying. Either outcome is more useful than another subscription nobody quite owns.

The context work is part of the work

Your CRM may hold useful history, but it doesn’t automatically contain everything your team knows. Your priorities, member needs, brand voice, and rules for handling an exception may still live in someone’s head.

That’s why getting value from AI involves organizing knowledge and helping people apply it. I explored this in Context Is Queen: What One Prompt Built in HubSpot. A short prompt can sit on top of a much larger foundation of context.

At GrowthHubAI, our HubSpot optimization work and organizational AI work address both sides: making systems more useful and equipping people to use them in their everyday work.

If your team is looking at tool number 48, and has 134 tabs open, let’s first look at the work you need to accomplish and the capabilities you already have. Book a strategy call to talk through where to start.

Tab 135 can wait.

Context Is Queen: What One Prompt Built in HubSpot

Erica Salm Rench — Fri, 30 Oct 2026 18:01:05 GMT

One prompt. A multi-email marketing campaign, a matching landing page, reusable copy, and the HubSpot segment for launch.

That was the result of a recent project for my personal brand, Erica Salm Rench. With a single command, I had Claude build a fully designed email campaign in HubSpot, branded to my ESR website, draft a landing page to match, create copy I could repurpose across individual sends, and create the segment I wanted to reach.

The work was already taking shape inside the CRM. The assets needed a little tweaking, but by golly, not much.

That last part is what made me stop and pay attention. The distance between having an idea and having something nearly ready to launch had gotten a whole lot shorter.

The one-prompt result had a much longer backstory

Did I set my systems up for success first? Yes. Did I provide detailed, extensive context? Also yes.

The command was short because I had already invested in the foundation.

This is the part of the AI conversation I keep coming back to. We share the impressive output, and everyone understandably wants to know what we typed. The more useful question is what the system already knew when we typed it.

Context is Queen, my friends.

For sales and marketing operations, useful context can include who you serve, what you offer, how you sound, what you want a campaign to accomplish, and how your CRM is organized. It also includes the boundaries: what the AI can do, what it should leave alone, and what needs your review.

A campaign has to make sense across several connected decisions. The audience, offer, email copy, landing page, and next step all need to work together. Giving AI that shared foundation helps it carry the same intent through the work.

I've done similar work in ChatGPT, too. My example happened to use Claude. The context investment is something I would prioritize with either tool.

Nearly ready still needs an operator

I reviewed the work and made adjustments before it would be ready to send.

For any campaign like this, I want to check the message, the design, the links, and the audience. I want to know that the segment reflects the people I actually intend to reach and that the landing page delivers on the email's promise.

That review takes judgment. So does deciding what the campaign should accomplish in the first place.

The time savings give me more room for those decisions. Instead of starting every asset from a blank page, I can spend more of my attention evaluating whether the pieces add up to a useful experience for the person receiving them.

The opportunity goes beyond campaign production

The same setup can help me examine attribution across my funnel and surface recommendations for answer engine optimization, or AEO: how clearly my organization can be understood and found through AI-generated answers.

That opens up a different set of questions. Which channels are contributing to pipeline? Where are people dropping out? What questions should my website answer more clearly?

The quality of those insights still depends on the information available. Connected systems and useful context give the AI more to work with; someone who understands the business still needs to interpret the findings and decide what to do next.

I was also tapped to join an early tester group for pre-release features at a company with well-known AI functionality. I can't share the details, but the experience reinforced how much is still changing in this space.

For anyone working in growth operations, I think the value of knowing what to ask, supplying the right context, and recognizing a useful answer will keep growing alongside the tools.

Start with the foundation

If you're wondering how to get more out of AI in your own sales and marketing operations, look at one workflow your team repeats. What would a capable new teammate need to know to do it well?

That is a useful starting point for building your context. Capture the goal, the audience, the process, the relevant systems, and the decisions that need a human. Then put that foundation to work and improve it as you learn.

My one-prompt campaign was a satisfying moment because I could see that investment paying off. Campaign, landing page, copy, segment: nearly ready, together, inside HubSpot.

If you're on HubSpot or looking to get started with HubSpot, I'd love to talk about what this could look like for your organization, including the strategic insights your systems could help surface.

When Answers Cost $20 a Month, What's Actually Scarce?

Erica Salm Rench — Thu, 22 Oct 2026 19:27:51 GMT

The three new bottlenecks of AI-powered work, and why the one nobody talks about matters most.

My freshman year of high school, a teacher drew a simple chart on the board. It showed a curve that started with a smiley face, dropped into two frowns, and climbed back up to a smile. Lately it has been on my mind quite a bit; (thanks, Mr. S.).

That's because of a question I keep asking the association and business leaders I work with: when knowledge is abundant and you can do almost anything for $20 a month, what's actually scarce (and most important)?

The short answer is that knowledge is no longer the bottleneck. Time, judgment, and knowing what you don't know are.

Knowledge Isn't the Bottleneck Anymore

For most of modern work, expertise was expensive. You paid for it in tuition, in consultants, in agencies, and in years of experience. Access to the right answer was a competitive advantage on its own.

That has changed. A single AI subscription now gives anyone a research assistant, a writer, an analyst, and a first-draft designer. The cost of producing an answer is approaching zero.

When the cost of something drops that far, its value moves somewhere else. In my work with teams adopting AI, I see it moving to three places.

Bottleneck 1: Time, and Which Questions Deserve It

AI will answer almost any question you ask it. That sounds like it saves time, and it does. But it also shifts where your time matters.

The real work is now deciding which questions deserve your hours in the first place. A team that uses AI to answer the wrong question faster has not gained anything. It has only arrived at the wrong place sooner.

The leaders getting the most out of AI spend more time up front defining the problem, and less time producing output for its own sake.

Bottleneck 2: Judgment, and What You Put Your Name On

You can now produce an unlimited number of assets: blog posts, emails, social content, reports, proposals. Volume was never the hard part, and it certainly isn't now.

Every one of those assets still needs a human to decide whether it actually represents you and furthers your mission. That requires a clear point of view, an understanding of your audience, and the discipline to say no to content that is technically fine but doesn't sound like you.

As the volume of AI-generated content grows, that human review becomes the thing that separates organizations people trust from organizations people scroll past.

Bottleneck 3: Knowing What You Don't Know

This is the bottleneck people talk about least, and it is where that high school chart comes in.

The Four Stages of Competence

The chart is the Four Stages of Competence, a learning model commonly credited to Noel Burch of Gordon Training International in the 1970s. It describes how people move from ignorance to mastery:

Unconscious incompetence: You don't know what you don't know. You're smiling because you have no idea what you're missing. (Sometimes this place of blissful ignorance can feel pretty good, depending on the situation... like I will remain at this stage forever when it comes to how many calories are in a beignet. This is fine for me when it comes to this realm of knowledge.)
Conscious incompetence: You can now see the gap. This is the first frown, and it's uncomfortable.
Conscious competence: You know what to do, but doing it well takes real effort and focus. Still a frown, but a productive one.
Unconscious competence: The skill is second nature. This is the smile that was earned.

Some tips when we are applying these concepts to AI, and how to move to the happiest quadrant:

Decide the goal before you prompt. Before you open an AI tool, memorialize what GOAL the answer will inform and help you get closer to. If you can't name one, the question probably isn't worth your time yet.
Define what "ours" looks like. Document your voice, your point of view, and your standards so that human review has something concrete to measure against.
Ask AI to find your blind spots. Instead of only asking for answers, ask what an expert would check, what's missing from your thinking, and what could go wrong. Then verify what it tells you.
Treat discomfort as a signal. If learning something new with AI feels effortless, you may be in stage one. A little friction often means you're finally seeing the gap.

Here's that chart (thanks, Mr. S):

The happiest stage and the most dangerous one look exactly alike.

Why AI Makes Stage One More Comfortable

Here is the problem. AI makes unconscious incompetence feel better than ever.

When you ask an AI tool a question, you get back a confident, polished, well-organized answer. It reads like expertise. But if you don't know enough about the topic to spot what's missing, that polish hides the gap instead of revealing it. You don't know which follow-up question to ask, so you never ask it.

The same thing happens at the organizational level. A team rolls out AI tools, output goes up, everyone feels productive, and nobody notices the strategy, data, or expertise that isn't there. The smile at stage one looks exactly like the smile at stage four. Only one of them was earned.

Moving from stage one to stage two is uncomfortable, and moving through stage three takes work. But that dip is the only path to real competence, with or without AI.

The Bottom Line

When answers are cheap, the advantage goes to the people who can see the edges of what they know.

AI has made knowledge abundant. What it hasn't made abundant is focus, judgment, or the self-awareness to notice what you're missing. Those are now the things worth investing in, for individuals and for whole organizations.

Frequently Asked Questions

What are the four stages of competence?

The four stages of competence are unconscious incompetence, conscious incompetence, conscious competence, and unconscious competence. The model describes how people move from not knowing what they don't know, to recognizing the gap, to building a skill with effort, to performing it automatically.

Who created the four stages of competence model?

The model is commonly credited to Noel Burch of Gordon Training International, who developed it in the 1970s. It is also known as the conscious competence learning model.

Why does AI make blind spots harder to see?

AI tools produce confident, fluent answers regardless of how complete they are. If you lack the expertise to evaluate an answer, its polish can make you feel informed while hiding what's missing, which keeps you in the unconscious incompetence stage longer.

What are the biggest bottlenecks when using AI at work?

Once knowledge is cheap and widely available, the main bottlenecks become time (choosing the right questions to answer), judgment (deciding what output truly represents your organization), and self-awareness (knowing what you don't know).

If your team is producing more than ever with AI but isn't sure it's producing the right things, that's exactly the work we do at GrowthHubAI. Let's talk!

Meet the AI Growth Operator

Erica Salm Rench — Sat, 10 Oct 2026 18:49:42 GMT

One thing I’ve been thinking about a lot lately: who actually owns growth at your organization?

Not just marketing. Not just sales. Not just membership. Growth itself: the work that happens between those functions, when a stranger becomes a prospect, a prospect becomes a member, and a member becomes an advocate.

That is where I have spent much of my career. And it is where I see an increasingly important role taking shape: the AI Growth Operator.

What is an AI Growth Operator?

An AI Growth Operator connects an organization’s growth strategy, people, knowledge, and systems, then uses AI agents to help that work happen. The role spans marketing, sales, membership, and operations because growth rarely stays inside one department.

The job includes deciding what matters, creating shared organizational context, improving the systems behind relationships, and helping the team use AI in useful, repeatable ways.

The title matters less than the responsibility. Someone needs to own the connections.

Why this idea resonates with me

I started thinking about this after listening to Greg Isenberg discuss a related idea on The Startup Ideas Podcast: the Marketing Engineer. What resonated with me was the emphasis on building the actual workflows and AI agents that help a business grow.

My conversations with association leaders keep bringing me back to something broader than marketing alone. They describe disconnected tools, valuable knowledge scattered across individuals, and people experimenting with AI without a shared foundation.

I have scaled agency operations to support 1,200 customers, helped grow rasa.io from zero to more than 300 customers, and helped grow Sidecar’s learner audience fivefold. Across those experiences, growth has depended on connecting good people with the right systems and a clear direction.

AI changes what we can do with those connections. It does not make the connections any less important.

Discovery and attention are changing

People can now research organizations through AI-generated answers as well as search engines, social channels, communities, and conversations. That adds a new question to the growth playbook: can these tools understand what your organization does and recognize its expertise?

HubSpot’s Loop Marketing framework reflects this broader shift toward connected, adaptive marketing. For me, the practical implication is that a website needs to do more than look good. It needs to explain the organization clearly, answer meaningful questions, and support an ongoing relationship.

That is part of the work of answer engine optimization, or AEO. Search still matters. So does being a useful, credible source wherever someone starts their research.

Attention also depends on a recognizable human voice. I see an opportunity for leaders who can express a real point of view, share useful experience, and show up consistently. AI can help amplify that work, but the expertise and judgment need to remain real.

What the work looks like in practice

Start with leadership and opportunity

Work with senior leaders to understand where the organization wants to grow, what is getting in the way, and where AI could make a meaningful difference. Move beyond a list of interesting tools to a set of useful priorities.

Build shared context

Give AI the organizational knowledge it needs: goals, audiences, voice, processes, priorities, and boundaries. Interview the people doing the work so the foundation reflects what actually happens across the team.

Shared organizational context is how we begin moving beyond scattered individual prompting toward a system the whole team can use.

Connect engagement systems

Clean up the data, clarify workflows, and connect the information behind marketing, sales, and membership. HubSpot optimization can be the starting point when the biggest opportunity sits in the systems you already have.

The goal is a foundation that supports better decisions and more relevant conversations.

Put AI agents to useful work

Agents can help with research, reporting, preparation, and other repeatable work. They need relevant context, clear boundaries, and someone responsible for reviewing and improving what they produce.

I think of it as agentifying the tedious parts of marketing and sales so your people can spend more time on strategy and, even more importantly, talking to members.

Help the team keep building

Train people around their own context and workflows. Maintain the knowledge. Refine the agents. Keep the work connected to the organization’s priorities as those priorities change.

Some organizations need a focused implementation. Others need fractional marketing leadership, fractional AI leadership, or ongoing agent support to keep the work moving.

Who owns this at your organization?

Call the role an AI Growth Operator, a growth marketer, a revenue operator, or something that fits your team. What matters is that someone takes responsibility for the whole picture.

Who connects your growth goals with your people’s knowledge, your systems, and the way AI gets used every day?

If the answer is not yet clear, that is a useful place to begin.

At GrowthHubAI, my team and I help organizations make those connections. Talk with Erica and her team about the next chapter of your growth.