Redefining Procurement Processes with Hyperbots PR/PO Co-pilot

Introduction to Procurement Challenges in Finance

Featuring Dave Sackett, VP of Finance at Persimmon Technologies, in conversation with Shalini Urankar, CEO of IkigAI Advisory.

Shalini Urankar: Welcome everyone. I'm joined today by Dave Sackett, a finance leader with deep expertise in AI, ERP systems, and operational finance. He currently serves as VP of Finance at Persimmon Technologies and has held several roles at companies such as ULVAC Technologies, Visibility Corporation, and NOAA Biomedical. In addition to his corporate roles, he co-founded AI One and advises multiple AI-focused startups.

Dave, today we're going to unpack a challenge that every finance leader knows too well, which is procurement. We're diving into Hyperbots PR and PO co-pilot (Procurement Co-pilot), which is transforming a traditionally manual, error-prone process into a streamlined, intelligent workflow.

Before we dive into the solution, though, let’s talk about the current reality. From your CFO lens, what makes the PR-to-PO process such a persistent bottleneck? Why is it such a headache?

Why the PR-to-PO Process is a Persistent Bottleneck

Dave Sackett: It's usually a very manual system with many internal steps, making it challenging for many companies. That’s one reason we're looking to Hyperbots to upgrade our solution. Some of the pain points include entering purchase requisitions manually. There are a lot of forms and fields to fill out, which takes time and can result in keying errors due to the manual nature of the process.

Often, bottlenecks occur on the approval side. Right now, we’re using SharePoint on-premise, which is being phased out, so we need another workflow engine. Agentic AI has a built-in workflow engine. Without proper workflows, people rely on emails for approvals, which can be delayed. People aren’t always notified, and this slows everything down.

There are also budget blind spots. People submit purchase requisitions without knowing what the budget is, and without automatic checks, which leads to surprises and overspending.

Duplicate purchase requisitions are another problem. People may not know what's already been submitted or who is working on it. Without visibility, this leads to too much inventory, different pricing, and a lot of confusion.

Wrong GL account codes are common. Often, purchasing or supply chain staff just guess the correct code, which finance then has to reclassify, disrupting the audit trail. There's also poor status visibility. People constantly ask accounts payable for updates, and the lack of clarity slows everyone down and creates tension.

Lastly, there’s a weak audit trail. With many people involved across different platforms, the documentation is scattered, which can cause SOX compliance issues and policy gaps.

How Hyperbots Procurement Co-Pilot Transforms Procurement

Shalini Urankar: Very true. I’ve encountered those issues myself in different companies I’ve worked with. That said, I’ve seen many procurement tools that promise automation but still rely heavily on manual keying. From your perspective, how would you describe Hyperbots’ Procurement Co-Pilot? How is it different?

Dave Sackett: It’s an agentic AI assistant that really helps. It can automatically extract purchase requisition fields using Vision-Language Models (VLM) and Large Language Models (LLM) to read contracts and statements of work, pre-filling about 80% of the required data, which saves a lot of time.

It runs live budget checks, which I find very valuable. It ensures there’s no duplication and that all entries comply with policy. Because it’s an AI assistant, it can cross-check data across multiple sources, which is something that typically needs a dedicated staff member.

It can also generate and send purchase orders using branded templates. Approvals happen in minutes with no manual keying or emailing involved, being automated through AI.

The AI assistant learns continuously from every approval, denial, or human correction. It incorporates those learnings into future recommendations. So over time, the bot becomes smarter and more efficient at handling purchase requisitions and orders.

Step-by-Step Workflow of Hyperbots Procurement Co-Pilot

Shalini Urankar: That’s great, it being a tool that learns as it goes. It doesn’t get better than that, right? When you're in the CFO seat, you care not just about efficiency but also about accountability and auditability. It’s great that this tool addresses both.

Let’s talk about workflow a bit more, because I know that’s where things can get tricky. I remember one quarter where we had $400,000 in open POs and no one had a clear view of the process. Can you walk us through Co-Pilot’s workflow from start to finish? What exactly is happening under the hood?

Dave Sackett: Sure. Here’s a sample workflow. Of course, this will vary by organization, but it gives a general idea of the functionality.

• Step 1 – Contract or statement of work extraction: The AI uses models to pull amounts, line items, payment terms, and other relevant data directly from the documents into the purchase requisition.
  
  
  

• Step 2 – Perform field enhancement and autofill: Purchase requisition screens open already populated with vendor info, item descriptions, pricing, tax, and dates.
  
  
  

• Step 3 – Validation and reasoning with AI: It checks the math, looks for anomalies, and provides transparent explanations for flagged issues, with guidance on how to fix them. It also checks for duplicates and scans inventory. The Co-Pilot cross-references live open PRs with stock. For budget utilization and GL coding, it does real-time checks and recommends GL accounts.
  
  
  

• Step 4 – Flexible approval process: Integrates with Slack or Teams, routes workflows, and supports category/cost center approvals with reminders and escalation.
  
  
  

• Step 5 – PO creation and dispatch: Uses custom or predefined templates, with options for auto-emailing to suppliers or human review first.
  
  
  

• Step 6 – Vendor portal collaboration: Suppliers can view POs, purchase history, and collaborate on payment timing.
  
  
  

• Step 7 – PO close automation: Integrated with Hyperbots Invoice Processing Co-Pilot to close POs once fully billed and delivered.
  
  
  

• Step 8 – Audit trail and notifications: Every action is hash-chain logged, timestamped, and sent as alerts.
  
  
  

Tangible Benefits and Key Metrics

Shalini Urankar: Wonderful. It’s great that you can take a phased approach until you're confident about the tool’s accuracy. It sounds like you did just that by evaluating the tool before fully adopting automation. In your experience, where do you see the biggest gains? Is it time, accuracy, or control? If you had to boil it down to key metrics or outcomes, what stands out?

Dave Sackett: The biggest benefit is time savings. Employee time saved is probably my number one metric. For example, filling out a purchase requisition might take 15 minutes manually. With this solution, it drops to five minutes, three times faster.

There’s also a major time gain in PO creation and dispatch. Approval workflows that used to take days can now happen in minutes, up to 90% faster.

There are also intangible benefits. Human error decreases, accuracy increases, audit trails are more robust, and approvals move faster with clear communication. POs can be auto-closed, and everything can be checked against budgets to prevent overspending. It even improves KPIs.

To give a concrete example, employees can create purchase requisitions in five minutes, POs land in the vendor’s inbox quickly, and auditors get a one-click audit trail as a result of this solution.

Shalini Urankar: That’s great. I can vouch for that. In my previous company, we projected nearly a 60% reduction in PO cycle times, and there was a material impact on early pay discount capture, which is huge.

Shifting Ownership from Manual to AI-Driven Processes

Shalini Urankar: Dave, in most companies, AP and procurement typically carry the operational burden. I’ve watched finance teams act like traffic controllers for routine requests. How will the Co-Pilot change ownership of that effort, and where will it be concentrated going forward?

Dave Sackett: That’s a great point. In my organization, supply chain and accounts payable are separate functions, but they will both be responsible for the areas where the AI Co-pilot operates. Both teams will be involved and have a say in how improvements happen.

In terms of tasks, right now, entering a purchase requisition is a manual job. With Co-Pilot, AI extracts and autofills the required data, saving time.

Validations, such as checking math or policy compliance, currently require manual review. With AI, these are automated. The AI reads the policy and uses judgment to ensure compliance.

It checks for duplicates, scans stock, and reviews all related requests. Before Hyperbots, people were doing ad-hoc searches. Now the AI ensures a thorough and consistent check every time.

For budget checks, I currently use spreadsheets to track fixed asset budgets. This tool can link to that data and verify whether the expense was planned. If it’s not part of the plan, a human can step in and assess whether it should be pushed to a future budget period.

As for assigning GL codes, the supply chain team currently does this manually. With AI, you train it once, and it remembers the correct account. If there’s a deviation, it flags it, acting as an assistant to keep everything accurate.

Approval routing is now done via email or SharePoint. Soon, it will be integrated with Slack and Teams workflows.

Today, buyers manually generate and email POs. We have several buyers at Persimmon who do this regularly. In the future, AI will automate the posting and dispatch of POs to suppliers.

Tracking status and closing POs is currently the responsibility of individual buyers. We’re moving toward automation that includes vendor portals and automatic PO closure.

Shalini Urankar: That’s a big shift from manual push to automated pull. Teams can now focus on reviewing exceptions instead of chasing every step of the process, which is a huge improvement.

Key Features That Set Hyperbots Apart

Shalini Urankar: What features do you think move the needle in the Co-pilot?

Dave Sackett: One of the key features is pre-trained extraction from contracts, being able to pull relevant data without a human having to do it manually. That alone saves a lot of time and effort.

It’s already pre-trained into the model, so there’s no extra setup required; it comes with Hyperbots. Automatic purchase requisition autofill is built in. It pre-populates as much data as possible, checks the math, evaluates the language, and flags any anomalies using simple explanations. 

It also performs duplication checks, scanning the entire entity for duplicate purchase requisitions. If it finds one, it can stop the process in real time. Budget control is real-time too; it can flash a red banner to alert users when something hasn’t been budgeted, helping teams avoid policy violations and overspending.

Adaptive GL recommendations ensure the correct GL account is selected, reducing the need for finance to reclassify entries. This can increase GL coding accuracy to up to 99%. You can also use custom or predefined purchase order templates, keeping your company’s brand consistent across documentation. The system handles automatic PO dispatch and closing, managing workflows, and ensuring everything is routed and tracked properly.

Flexible workflows and notifications mean approvals can happen in context, with better communication around status updates. Everything, including the audit trail, is captured in Hyperbots.

It also supports multi-entity operations. Compliance is easier to manage, and you get a single dashboard to view your data quickly—some of the key takeaways of the solution.

Why Choose Hyperbots Over Traditional Procurement Tools?

Shalini Urankar: That sounds like music to a controller’s ears… Why choose Hyperbots over other e-procurement or P2P tools on the market?

Dave Sackett: Yes, I have prior experience with other RPA implementations, but Hyperbots is different. One of the biggest differentiators is field extraction, where it uses expert-level vision-language LLMs as part of the solution, unlike traditional RPA tools that rely on fixed templates or manual training. Agentic AI has built-in reasoning. It checks for duplicates, scans inventory, and makes decisions like a human would—unlike tools that just pull data from an MRP database.

When it comes to budget and GL intelligence, Hyperbots offers real-time, adaptive learning. Traditional RPA tools operate on static rules, for example, "this vendor always maps to this GL account." Hyperbots is far more flexible.

It also automates the entire PO process by generating and dispatching POs all in one place rather than through scattered emails.

Another big differentiator is the learning loop. Every transaction teaches the bot your specific company’s workflow. It adapts instead of sticking to a one-size-fits-all model. The audit evidence is hash-chained and searchable, unlike other systems where it's often scattered.

From a pricing and rollout perspective, Hyperbots operates on a pay-per-PO model. My previous RPA implementation required a huge upfront investment and ongoing maintenance. Hyperbots is different, where you pay per transaction, which makes scaling more cost-effective. Deployment takes just three weeks.

You don’t need seat licenses either. Everyone in the company has access at no additional cost; it’s all usage-based. You only pay for what the bot does, and I find that incredibly appealing.

Hyperbots Procurement Co-Pilot: The CFO’s Elevator Pitch

Shalini Urankar: That’s great to hear. If a fellow CFO asked for an elevator pitch… what would you say?

Dave Sackett: Hyperbots Procurement Co-Pilot shrinks purchase requisition creation time to just five minutes, turns approved requests into vendor-ready POs in minutes, and gives finance real-time budget control. It’s auditable, reduces errors, removes delays, and lowers costs; all in one streamlined solution.

Conclusion

Shalini Urankar: I'm enjoying this new universe of intelligent tools. Thank you so much for walking us through it. The Procurement Co-Pilot delivers not just speed but also structure; something modern finance teams truly need. It’s been a pleasure talking to you. Thank you again for joining us today.

Dave Sackett: Thank you, Shalini.