Conversation
Highlights

How AskElephant Converts 85% of AI Pilots to Production While the Industry Struggles at Single Digits

Woody Klemetson spent years scaling sales teams—100 people at Divi, then 350 at Bill.com after the acquisition. He learned how to merge cultures, build systems, and drive results at scale.

Then he walked away to solve a harder problem: why do AI pilots fail?

Not the technology question. The implementation question. Why can every revenue leader articulate what they want AI to do, but less than 5% of pilots make it to production?

In a recent episode of BUILDERS, Woody explained how AskElephant hit 400% growth in their first year with zero marketing spend and, more critically, why 85% of their pilots convert to production. The answer has nothing to do with having better models and everything to do with understanding what “AI ready” actually means.

The Diagnostic Question That Reveals Everything

When prospects tell Woody they’re AI ready, he has a single diagnostic: “We just asked, walk us through your current customer journey.”

The answer reveals everything. “They say, well, we don’t. We have cell stages,” Woody explains.

Sales stages aren’t a customer journey. They’re CRM hygiene. And if teams haven’t documented actual workflows—the decisions, handoffs, and judgment calls that happen between stages—AI has no training ground.

“A lot of people, they’re not ready for AI,” Woody says. “Some of their systems aren’t in place. They don’t know what it is. They’ve been reliant on people that haven’t documented systems.”

This is why AskElephant starts some customers with dictation. Not voice-to-text transcription as a product feature, but as “the gateway drug to get people just starting to use AI to start to open their horizon on how they should think.”

The point isn’t the feature. It’s training humans to articulate their processes out loud so AI can learn from them.

The 40-Hour Reality Behind 5% Forecast Accuracy

Most founders assume AI should work from the first prompt. ChatGPT makes it look easy—ask a question, get an answer.

Production systems don’t work that way.

AskElephant runs 27 different CRM agents. Together, they deliver forecasting “within 5% accurate of where we end for the month” without human input. “It still took 40 hours to build and give it the context and train the system to do that,” Woody explains.

Here’s what 40 hours of context-building actually means: each agent handles a discrete job. Contact creation. Data enrichment. ICP scoring. Churn monitoring. Stage progression logic. Each learns from the previous layer’s context. The 27th agent is only accurate because agents 1-26 have already processed and structured the data it needs.

This is compound context architecture—not one smart agent, but a system of specialized agents where each adds a layer of understanding. “I think that’s the misconception is if you’re using AI, you’re like, well, it didn’t do it. And it still takes effort and human capital to get the AI and system built.”

The companies that convert pilots understand this from day one. The ones that don’t expect magic and abandon the project when magic doesn’t happen.

Why 85% Conversion Starts With One-Minute Wins

Industry standard for AI pilot conversion: single digits. AskElephant: 85%.

The architecture is counterintuitive: “Dream big, implement small.”

First pilot doesn’t save 27 hours per week. It saves one minute per day per person. “If I save one minute person per day, that’s a good start,” Woody says. “And now we’re on average saving 27 hours per week person.”

The progression is deliberate. Start with automated CRM notes—what humans currently write. Then layer in notes humans wish they’d written—the context they skip because it takes time. Then automated field updates nobody does consistently. Each step proves the system works before adding complexity.

“Within 30 minutes, you now never have touch your CRM again,” Woody explains about their onboarding. “The AI is managing your CRM for you. It’s looking at customer context, it’s updating the fields, it’s adding ICP scores, it’s monitoring for churn alerts.”

But the 30-minute implementation only works because they’ve already scoped to one discrete problem with documented inputs and outputs. The temptation—for both vendor and customer—is to scope big and demonstrate comprehensive value. That’s where pilots die.

Big projects fail because the last 20% takes 99% of the effort. Small wins that deliver real outcomes build trust that funds the next layer.

Partner-Led Distribution as Primary GTM

AskElephant’s 400% growth came entirely through partners. Not traditional channel partners selling on commission, but consultants whose businesses were being disrupted by AI.

The insight: Salesforce and HubSpot implementation partners were becoming AI strategists whether they wanted to or not. Sales coaches needed to extend beyond training into system implementation or risk becoming irrelevant. These professionals had built practices around delivering transformation, but their clients wanted results that stuck, not slideware.

“These partners who started off as salesforce partners, HubSpot partners and other tools, they’re the ones that are leading the charge,” Woody explains. “Our consultants and contractors, what they’re doing now is they’re now getting longer renewals because they’re not just going and teaching the system like a coach, sales coach would. They now get to help you implement that system and help hold you accountable to that system so you actually get the outcomes.”

The model works because AskElephant solves the partner’s business model problem. A sales coach used to train a team and leave. Contract ends. With AskElephant, they implement AI systems that enforce the training, creating ongoing engagement and measurably better outcomes for their clients.

Result: partners become the primary sales motion because AskElephant makes them more valuable. They’re not incentivized by commission—they’re incentivized by their own client retention and expansion.

Hiring for Agent Orchestration, Not Quota Carrying

Over half of AskElephant’s non-engineering team uses Cursor daily. This isn’t a technical curiosity—it’s the hiring filter.

“I’ve been hiring ops minded sellers, I’ve been hiring tech minded sellers,” Woody says. The requirements haven’t changed for human skills: “They have to be able to talk to humans, they have to be able to connect, they have to be curious, they have to know how to listen.”

What’s changed is everything between those human touchpoints. “They’re not building lists anymore. They don’t create sows, they don’t create the handoffs with other things. They’re not creating product scopes, they don’t use a solution engineer to sell the tool.”

The transformation: “You no longer get to just be the silver tongue salesman that has a solutions engineer, has like 27 people helping you sell. It’s now 27 agents helping you sell for you.”

This isn’t aspiration. It’s operational reality. Their website—built in one week for $2,000 in tokens with every motion graphic coded—demonstrates what this looks like in practice. “My VP of marketing is coding,” Woody notes.

The filter for hiring: can this person think in systems, implement technical solutions, and orchestrate agents while maintaining high-touch relationships? If they can’t manage the orchestration layer, they can’t scale in this environment.

Revenue Outcome Systems: Category Positioning Through Problem Clarity

AskElephant calls their category “revenue outcome systems” to distinguish from “revenue operating systems” language others use.

The distinction matters. “I am calling it an outcome system because I think that’s actually what we care about as companies,” Woody explains. “We don’t care about do we operate this month. We care about do we get the outcomes that we want.”

But Woody’s realistic about category creation: “I believe it’s actually happening with or without me. Categories usually are not created by one individual and they’re created by what the market is asking for.”

The underlying problem exists regardless of terminology: “Infrastructure for a hybrid human AI team doesn’t exist.” Every revenue leader uses 15-20 disconnected tools trying to make revenue predictable. None were architected for AI to operate alongside humans as a peer, not a feature.

On whether they need Gartner validation: “We already know people are using those tools less and consulting less and switching more and more to the deep type of research that they’re doing through GPT and Claude to really understand how to solve these problems that they have every day.”

The category will form through problem recognition, not analyst validation. AskElephant’s job is implementation success at scale, not winning the terminology war.

The 20x Efficiency Thesis

Woody’s long-term vision isn’t about replacement—it’s about compression. “We believe humans are still the future. Maybe that shocks people.”

The thesis: “I believe that each human will be around 20x more efficient.”

Not 20x more automated. More efficient. The work that gets replaced is the coordination layer—the administrative tax of working in systems designed for compliance, not outcomes.

“That’s where I think the future is going to just feel where you just get to show up and be able to do your work,” Woody explains. “We have prototypes within 15 minutes of us talking about a problem and then you iterate and iterate to truly get the things that you need.”

The infrastructure they’re building: “Customers who want to talk to humans should be able to talk to humans, and customers who don’t want to talk to humans shouldn’t have to talk to humans.”

This is the actual implementation challenge. Not whether AI can do the work, but whether you can build systems where AI and humans operate as peers with clear handoffs based on customer preference and outcome optimization.

For revenue leaders watching pilots fail, AskElephant’s playbook is clear: document before you automate, scope to one-minute wins, build agents that compound context, and scale from proven trust. The companies shipping AI aren’t the ones with the most ambitious roadmaps—they’re the ones converting small wins into systematic transformation.