Ask any revenue team today and you’ll hear it. “What are the best AI tools right now?”
It sounds smart. It’s actually the problem.
The AI market is moving fast enough that even experienced operators are getting swept up in the excitement, swiping the credit card on shiny platforms, and hoping the results justify the spend.
Most of the time, they don’t.
Not because the tools are bad, but because the buying process is broken. There’s no universal answer to which AI tools are best, because the right AI strategy depends entirely on where your company is, not where you want to be.
Blake Tiemeyer, Director of Growth Acceleration at General Atlantic, and Amy Kramer, Operating Partner for Go-to-Market at Level Equity, sat down with York IE’s Mike Veilleux on our State of the Industry: Value Creation webinar to talk through exactly how they evaluate AI investments across their portfolios.
Start With the Problem, Not the Tool
Amy said it directly when asked about the most common AI evaluation mistake she sees: “Most teams are asking the wrong question.”
Teams hear about a tool, get excited and reverse-engineer a use case. That’s backwards.
The right starting point is your problem list, not the demo. And the problems worth solving depend entirely on your company’s stage and posture.
Blake ran a study of more than 250 portfolio companies and found that AI positioning mirrors strategic positioning almost exactly.
Slower-growing companies that are in a defend-and-extend mode are using AI to protect the base: retention, churn reduction, customer support efficiency. Hyper-growers are deploying it aggressively at the top of the funnel to scale demand gen and pipeline creation.
The companies that get into trouble are the ones who get these backwards, a struggling company trying to use AI to blow up what’s working, or a fast grower getting so cautious about doing it right that they lose the speed advantage AI was supposed to give them.
Amy sees companies buy sophisticated data orchestration platforms when what they actually needed was basic enrichment. They buy AI SDR tools when their CRM data is too messy to support accurate targeting. The tools aren’t wrong. And if the underlying data isn’t clean, no tool will save you.
Productivity vs. Performance: Know Which One You’re Solving For
Not all AI investments are created equal, and treating them the same is one of the fastest ways to misread results.
Amy draws a clean line between two categories: productivity gains and performance gains. Productivity is about speed and efficiency. Performance is about outcomes, conversion rates, win rates, expansion revenue.
“Was it productivity? Was it performance? Was it to improve engagement?” she said on the webinar. “You have to come up with the hypothesis of what you used that AI tool for.”
For productivity plays, buying off-the-shelf tools is usually the right call. The ROI math is straightforward and the use cases are proven. For performance plays, Amy recommends a crawl-walk approach regardless of stage.
Before investing in a tool, validate the hypothesis manually.
“It’s okay if you’re copying and pasting something from ChatGPT into your email just to see,” she said. “Let me have clear KPIs to say does this improve conversion rates? If so, great. Now I want to invest in a tool that can do this for me much faster.”
Build a Testing Framework Before You Buy Anything
Here’s the thing about AI investments: most teams don’t actually know if they’re working.
Amy asked a portfolio company to walk her through their testing framework for an AI tool they were actively running. The answer was gut feel. They were moving fast, learning nothing. That’s not a technology problem. That’s a process problem.
This is the paradox that comes up constantly: the teams that get the most from AI are usually the ones that slowed down first.
Forcing adoption without structure produces surface-level compliance and real resistance underneath. You have to create the space to learn before you expect people to perform.
A real testing framework has four components:
A clear hypothesis. What specifically do you expect this tool to change?
Defined KPIs. What does success look like at 30, 60, and 90 days?
Leading indicators, not just lagging ones. Blake put it plainly on the webinar: “Some of the companies we work with have enterprise sales cycles of 18 months, so you need to have some way of assessing productivity now.” Are response rates improving? Meeting bookings increasing? Deliverability up? Track those now. Wait for pipeline contribution later.
A control group. AB test against your old workflows. Without a baseline, you can’t prove anything.
Blake takes it further: “Defining success is really important. It’s easy to swipe the credit card on a new product because it’s exciting and shiny, but if we don’t know what success looks like, it’s hard to actually hold ourselves accountable.”
“We’re learning” is not a framework. Fast iteration requires structure to mean anything.
The “Hear It Three Times” Mantra
One of the most practical takeaways from the webinar has nothing to do with spreadsheets or scoring models. It’s pattern recognition.
Blake’s rule, which he shared he’d also heard from a peer in the industry: if a tool’s name comes up three times across portfolio companies or trusted conversations, it’s worth booking a demo.
“If I hear the name three times, I set up a demo,” he said. “Then I can build the knowledge base, the case study with other portfolio companies I can introduce as champions, and then we can just send that out broadly.”
This matters for operators managing multiple portfolio companies at once because it compresses the evaluation timeline. Rather than every company running independent experiments on the same tools, pattern recognition at the portfolio level surfaces what’s working faster.
If you don’t have that cross-portfolio view, build a lighter version of it. Stay in peer groups. Share notes with counterparts at other companies. Follow operators who are posting real results, not vendor marketing. When you hear the same name from sources you trust, that’s your signal.
De-Risk the Purchase Before You Make It
Even with a solid hypothesis and a real testing framework, the AI tool market in 2026 carries structural risk that most buyers aren’t accounting for.
The pace of innovation is fast. Platforms that look differentiated today may be table stakes in six months, or absorbed into a larger stack entirely. Blake is direct about what’s coming: “I think there’s a massive consolidation coming.”
His tactical recommendation: protect yourself contractually. “Go for shorter terms, whether that’s six-month contracts or even monthly contracts. Building in trial periods that are pretty extensive, you can de-risk some of these net new product buys.”
On the build vs. buy question, Blake leans toward buy for most go-to-market use cases, particularly right now. Building and maintaining a custom AI solution takes resources away from your core product. The exception is niche use cases that don’t exist in the market, where a custom build would deliver disproportionate value specific to your business.
Amy’s framework ties back to the productivity vs. performance split: “From a productivity standpoint, buying tools makes sense. They exist and there are clear use cases. From a performance standpoint, we take a crawl-walk approach.”
The Bottom Line
There’s no single AI strategy that works for every company. What works is understanding your stage, your constraints and making tool decisions that match your reality, not someone else’s.
The teams getting the most from AI aren’t the ones who bought the most tools. They’re the ones who bought the right tools for the right reasons, at the right time, with a clear definition of success.
That discipline isn’t glamorous. But it’s what separates the companies that are actually accelerating from the ones that are spending a lot of money to stay in place.
