How To Stop Buying AI Tools and Start Identifying the Workflows That Still Run on Duct Tape

Look under the hood of most current revenue operations, and you often won't find seamless automation. You'll find duct tape. While AI is frequently marketed as a magical business transformation, many operators find the most reliable value in treating it as a ruthless execution layer. When applied to go-to-market systems with a clear problem in mind, AI tools can cut down manual reporting and help teams make decisions faster with less effort.

Jonah Sigel, Chief Revenue & Marketing Officer at Rysun Labs, approaches AI tools from an operator's vantage point. His career spans senior roles at Amazon, Starbucks, and Holt Renfrew, alongside managing corporate venture capital at Central Ventures and serving on the NextPoint advisory board. Sigel says his background shapes his view of the technology as a practical conduit that has to prove its worth in revenue generation, not just slideware. He points to the fact that many organizations in verticals like CPG and retail still rely on disjointed technology stacks. Revenue teams report they're often stitching together data from multiple systems just to extract basic operational insights, offering a prime opportunity for AI to prove its value.

"Depending on where teams are in their journey, they may have access to data, but it's coming from disparate systems. Finding inventory, or getting finance people the information they need right now, is brutally painful." Some of the most effective teams bypass top-down overhauls. Instead, they find value by empowering the frontline workers who are already connecting disparate systems by hand, and breaking down silos. Stripping away the mystique of the technology can help operators use AI to compress timelines and complete heavy-volume work faster and cheaper.

• Meet the MacGyvers: Stripping away the mystique of the technology can help operators use AI to compress timelines and complete heavy-volume work faster and cheaper, particularly when the tools are put in the hands of the frontline teams already doing the work. "It starts with going to the MacGyvers of your organization that are using duct tape and bubble gum to do their jobs right now, empowering them with the right tools," Sigel says. "Ensuring that they know which part of their job remains creative, human, and individualized, and where AI can take a lot of the meaty nastiness out of their workflows."

• Solving the SKU slog: Execution starts with hard math: defining exact efficiency gains for data-heavy grunts. In these kinds of high-volume workflows, the technology can shorten timelines for tedious work. AI now crushes massive data processing tasks—like overhauling Shopify storefronts or generating retail catalog descriptions—in a fraction of the time. "If you have 10,000 SKUs and you need to update them, what used to take somebody literally weeks or months to enter new SKUs with descriptions and photographs was really tedious," Sigel says. "Somebody now can take a picture of a shirt, instruct an LLM to put it on a model and a white background, and build that into an algorithm. That type of work is now done in hours, not weeks and months."

• Stitching the stack: That same logic can also be applied to how some teams view their core systems. Instead of treating every reporting request as a bespoke exercise, organizations can use AI-enabled stacks to unify existing data sources. "If all of a sudden we have access to data from all of our ERPs in one dashboard, instead of having to piece these together 38 different times with duct tape, that's going to save us time and increase efficiency," he says. "It's going to drive incremental revenue."

Against that backdrop, the AI vendor market has grown crowded. Sigel points out that the volume of offerings makes it difficult for buyers to distinguish between tools that solve specific problems and those trading on generic automation claims. Navigating this market starts with clearly identifying the problem to be solved, then choosing partners that can show exactly how they address it. He expects that over time, market survival will favor vendors focused on concrete operational gaps.

• The Shoptalk shelf life: "I just came back from Shoptalk and every single booth had the word AI in the title. I think 70 to 80% of those companies in 12 to 18 months won't exist anymore," he predicts. "The strong—not necessarily the big, but the strong—will survive, just as they did in e-commerce."

• Math over magic: To avoid buying into the noise, he advises teams to demand trackable ROI metrics before adopting new tools. For individual purchases, his recommendation is similarly pragmatic: justify spend in terms of hours saved and outcomes achieved, not generic innovation language. "If you tell your CEO that you want to start paying $200 a month for Claude, great. What does that output look like?" he says. "If I can say I'm going to achieve these five things, saving me this number of hours, and my hourly rate is Y, all of a sudden you've got things that are trackable, traceable, and meaningful."

As the tooling advances, the pace of change makes it helpful to stay focused on a clear north star—what the business is actually trying to fix. For many organizations, empowering the workforce starts with a clear governance strategy that both protects corporate data and drives the right outcomes. Sigel notes that this should include giving frontline teams access to AI tools, while also setting guardrails for where and how they're used. Setting those parameters helps workers develop the kind of judgment and skills needed to use AI effectively, rather than relying on blanket bans or ad hoc experimentation, a pattern echoed by recent research on how employees build judgment in the AI era.

• The Amazon approach: "Back in the day, when I worked at Amazon, when you were on the network, you couldn't go to Costco or Best Buy," he recalls, noting that several large corporations currently ban ChatGPT. "I don't think the answer is banning. I think the answer is understanding the right use case and setting up the right parameters."

• Babysitting the bots: For Sigel, that governance layer works hand-in-hand with the human judgment required to keep outputs reliable. Without human review, speed can just as easily amplify existing mistakes rather than fix them. "AI is not smart on its own. Somebody needs to issue a prompt," Sigel stresses, highlighting the need for catching AI errors. "All of these things require someone looking at something for at least a minute to recognize when a hallucination is not real. It's critically important that you have the right people using the right lens to ensure that things are being used properly."

Sigel sees this as a shared responsibility between management, which sets policy and endorses appropriate use, and the frontline workers on the ground. The "MacGyvers" understand where the AI tools can take the tediousness out of workflows without losing human oversight. That shared responsibility is what keeps the technology grounded.

AI can compress timelines, unify fragmented data, and strip the tedium out of high-volume workflows. But the operators who extract the most value are the ones who start with a specific problem, measure the outcome in hours and dollars, and keep a human close enough to catch what the model gets wrong. The technology works best when it's treated as an execution layer, not a strategy. The strategy still belongs to the people who know where the duct tape is.