AI for Founders Logo
All Episodes
Delightful Procurement: The CFO That Never Sleeps - Alex Yakubovich from Levelpath
April 2, 202600:59:29

Delightful Procurement: The CFO That Never Sleeps - Alex Yakubovich from Levelpath

with Alex Yakubovich, Levelpath

Delightful Procurement: The CFO That Never Sleeps - Alex Yakubovich from Levelpath

0:000:00
Listen on SpotifyApple Podcasts

Show Notes

Your company's biggest expense is not payroll. It is the money leaking out because nobody actually knows what they are spending. Every time someone swipes a corporate card because the official purchasing process is too painful, that is rogue spend. That is your cash walking out the door with no ID badge.

Alex Yakubovich has been in procurement technology since 2014. He and his co-founder Stan built Scout RFP, scaled it to a $540 million acquisition by Workday, spent a few years inside one of the world's largest enterprise software companies, and then went back to a blank page. Not because they needed the money. Because the problem was still broken.

Levelpath is what they built when they had no constraints - a platform designed from scratch to make procurement delightful. Not just faster. Not just cheaper. Delightful. This episode is about what that word actually means when you are building software for some of the largest enterprises on earth, and why the founders who are going to win with AI are the ones obsessing over customers - not models.

From Scout RFP to Levelpath: Why They Went Back

The Scout RFP acquisition was not the end of the story. Alex and Stan spent several years at Workday after the deal closed, learning what enterprise procurement looked like at global scale. They saw the complexity. They saw what was still broken. And they could not stop thinking about it.

The thing that kept pulling them back was simple: people did not love procurement. The tools were painful. The processes were opaque. Buying something at a large company required knowing who to ask, what forms to fill out, which system to use, and then waiting anyway. The Levelpath thesis was a blank page question: what if you built the entire procure-to-pay workflow from scratch, AI-first, with delight as the north star?

Their customers now include Western Union, American Airlines, New York Life, and Biogen. None of them have left.

Frameworks from This Episode

These frameworks have been added to the AI for Founders Frameworks Library. Filter by Operations or Alex Yakubovich to find them.

Name the Customer

The single filter that kills bad experiments before they waste resources. If a product manager cannot name a specific customer who will benefit, kill the feature.

  • •As experiments get cheaper, the danger is running more of them without a compass. Volume of experiments without customer grounding creates noise, not learning.
  • •The filter: before any feature, experiment, or initiative, someone in the room must be able to say 'customer X will love this because Y.'
  • •If the answer is abstract - 'our market,' 'the enterprise segment,' 'users in general' - that is a signal the experiment is internally motivated, not customer motivated.
  • •Alex sees this pattern in big companies and small ones alike: dashboards full of market data, nobody in the room can name the customer.
  • •The companies that win with AI will run more experiments than ever - but filtered ruthlessly through the customer lens.

The Delight Filter

Every product decision passes through one question: does this reduce cognitive load, or does it add to it? Delight is a feeling, not a feature.

  • •Beautiful icons that users cannot identify are not delightful. They are a tax on attention. Label the icon, or remove it and just use the word.
  • •Cognitive load reduction is the foundation. Surprising delight is the ceiling. You cannot get to the ceiling without the foundation.
  • •Levelpath's design principle 'surprisingly delightful' means the platform does something useful the user did not think to ask for - like auto-generating a negotiation strategy the moment you open a contract.
  • •Delight in software is the same as delight in hospitality: anticipating needs, pacing the experience correctly, and making the user feel like the product was built specifically for them.
  • •Alex's north star question: 'Does the customer feel like we took the thing in their head and made it real?'

Context Over Model

The AI model you use matters far less than the quality and completeness of the context you feed it. Managing context is the real competitive advantage.

  • •Levelpath does not require customers to have perfectly structured data. Messy contracts, missing metadata fields, unstructured files - the AI interrogates all of it and surfaces risk anyway.
  • •As models improve, the custom tooling and prompting tricks built around weaker models becomes obsolete. What does not become obsolete is rich, well-maintained context.
  • •When tariffs change overnight, Levelpath can tell you which contracts are exposed - even if you never created a 'tariff' metadata field. The context does the work.
  • •The platform's job is to help customers manage what context goes in, and make the AI output reflect their specific policies, procedures, and workflows.
  • •Customer education at Levelpath focuses on one thing: the better you manage your context, the better the platform works for you.

The Build vs. Buy Audit

Every software contract should now be evaluated against three questions before renewal. The answers will surprise most teams.

  • •Question 1: Do you still need this at all? AI from major providers has absorbed the functionality of many point solutions. Some software you are paying for is now redundant.
  • •Question 2: Do you need as many seats? If people can access the capability through an AI-native interface, the seat-based pricing model collapses.
  • •Question 3: Can you build it yourself now? IT departments can build things in weeks that used to take years. The bar for 'worth buying' just got much higher.
  • •Alex's caution: enterprise software that survives this audit does so because of depth, compliance, integration, and the operational risk of building incorrectly - not because of basic functionality.
  • •The companies that will lose are those whose products could be vibe-coded in two weeks. The companies that will win are those whose products could not.

Founder Experiment: Run a Rogue Spend Audit

Every person in your company is already doing procurement. Every time someone buys something on a corporate card because the official process is too painful, that is rogue spend. Here is how to find it, quantify it, and close the leak.

  1. 1Pull 90 days of corporate card transactions. Export to a spreadsheet or pipe into an AI tool. Ask: which of these purchases went through no approval process? That is your rogue spend baseline.
  2. 2For each rogue spend category, ask: why did someone buy this outside the system? The answer is almost always 'because the official process was too slow or too painful.' That friction is your product design problem.
  3. 3Drop your messiest contracts - PDFs, scanned docs, anything - into an AI tool and ask: what are my top five areas of financial risk in these contracts that I have not explicitly tagged or tracked? You do not need clean metadata. Let the AI interrogate the unstructured text.
  4. 4Pick one vendor relationship where you are not sure if you are getting what you paid for. Ask an AI to compare the contract terms to the last three invoices. Flag any discrepancies. This is the manual version of what Levelpath automates.
  5. 5Run the Build vs. Buy Audit on your five largest software contracts. For each one, answer: do we still need it, do we need as many seats, and could we build a functional replacement in 30 days? The answers will reshape your software budget.

Stretch goal: Identify one process in your company that currently requires someone to "go call Fred" to get information. Map the time cost of that process over a year. Then spec out what it would take to eliminate the Fred call entirely - through better data organization, a purpose-built AI query, or a platform like Levelpath. The gap between what it costs now and what it would cost to fix is your ROI case.

Key Terms

These terms have been added to the AI for Founders Glossary. Search by Alex Yakubovich to filter them.

Rogue Spend: Purchases made outside a company's official procurement process - typically on corporate cards or personal expense reports because the approved system is too slow or painful. Represents cash leaving the organization with no oversight.
Procure-to-Pay (P2P): The full procurement workflow from identifying a need and sourcing vendors through to purchase order, invoice receipt, and payment. Levelpath automates this entire workflow in a single platform.
Purchase Order (PO): A formal document issued by a buyer to a vendor authorizing a specific purchase at agreed terms. The PO creates a contractual obligation and is used to validate incoming invoices.
Three-Way Match: The process of verifying that a purchase order, a goods receipt, and a vendor invoice all align before payment is approved. When they do not match, it signals a billing error, delivery discrepancy, or potential fraud.
AI-Native Company: An organization that applies AI to automate its internal processes as a default, not as an exception. Distinct from an AI-native product - a company can sell AI products without being AI-native internally, and vice versa.
AI-Native Product: Software whose core value proposition changes week over week as new model capabilities become available. The tell: customers receive meaningful new capabilities on a near-weekly basis without requesting them.
Cognitive Load: The mental effort required to use a piece of software. High cognitive load - unlabeled icons, ambiguous navigation, unpredictable behavior - is the primary source of user frustration in enterprise tools.
Surprisingly Delightful: A Levelpath design principle where the platform proactively surfaces something useful the user did not think to ask for - such as a contract negotiation strategy auto-generated upon opening a document.
Contract Metadata: Structured fields attached to a contract document - vendor name, value, expiration date, key terms. AI procurement tools can surface insights even when metadata is incomplete by interrogating the unstructured contract text directly.
Rogue Spend Audit: A review of corporate card and expense transactions to identify purchases made outside the official procurement process. The first step in understanding and controlling unmanaged spend.

Tools from This Episode

Levelpath

AI-native procure-to-pay platform for large enterprises. Handles sourcing, purchase requests, purchase orders, and invoice matching in a single platform. Surfaces contract risk and vendor exposure without requiring clean metadata. Customers include Western Union, American Airlines, New York Life, and Biogen.

levelpath.com

Links & Resources