Faster go-live with Hyperbots pre-built connectors

Brian Kalish: Good day, everyone. It's great to be with you again. I'm Brian Kalish. Today, we have a wonderful opportunity to share some insights from both finance and technology perspectives. I'm joined by my good friend, Dave Sackett, Vice President of Persimmon Technologies Corporation. First of all, thank you for joining us today, Dave.

Dave Sackett: Thanks, Brian.

Brian Kalish: We've received feedback that our audience wants to understand how Hyperbots’ integration supports faster onboarding. The word "integration" often raises concerns—sometimes it’s not about cost but about time, which is just as valuable. So, unless there’s anything you'd like to add, let’s dive in.

Dave Sackett: I’ll just say, my background is in finance. If someone in finance is watching, I recommend including your IT counterpart. We’ll be discussing several technical aspects that your IT team will find important.

Brian Kalish: Great point. Let’s get into it. When we talk about integration, we're referring to leveraging existing ERPs. A frequent question is: What ERP systems does Hyperbots already have ready-made integrations with?

Dave Sackett: Hyperbots comes pre-built with bi-directional connectors for SAP S/4HANA, Oracle NetSuite, QuickBooks, Microsoft Dynamics 365, Sage, Acumatica, and Deltek Costpoint. In my case, we went live with Hyperbots using Epicor Kinetic, which is also supported. All use the same canonical schema. Once one co-pilot is live, others connect through the same pipeline. New connectors roll out every quarter. It’s in Hyperbots’ interest to support as many ERPs as possible for smoother transitions.

Brian Kalish: That makes sense. So, how do the co-pilots work if a connector doesn't already exist for an ERP?

Dave Sackett: We use a flexible integration toolkit. If your ERP exposes REST, SOAP, OData, SQL views, or even flat files, we connect an adapter to translate those calls into Hyperbots’ event bus. Internally, the co-pilot still interprets standard events like "purchase order created" or "invoice posted," so the AI logic remains consistent.

Brian Kalish: From a finance standpoint, does Hyperbots charge extra to integrate with unsupported ERPs?

Dave Sackett: No, there's no additional license or consulting fee. Building a connector is included in the subscription price. Since Hyperbots operates on a pay-per-transaction model, it's in everyone’s interest to complete onboarding quickly.

Brian Kalish: That’s reassuring. In terms of time, how long does it take to build a new connector?

Dave Sackett: For a new ERP like Epicor, it typically takes 2 to 4 weeks. Week one involves API mapping and authentication. Week two builds the CRUD calls—Create, Read, Update, and Delete. Weeks three and four handle replay tests and user acceptance testing (UAT). Only highly customized ERPs push the timeline to the upper end. Most are simple and can be connected easily.

Brian Kalish: Good to know. How frequently does Hyperbots write data into the ERP?

Dave Sackett: By default, copilots post immediately after the AI makes a decision. For example, an approved invoice or journal entry is posted within minutes. You can configure it to run hourly or at the end of the day if that aligns better with IT requirements. It’s adjustable via a configuration slider.

Brian Kalish: How does Hyperbots technically integrate with an ERP?

Dave Sackett: There are three layers: connectors that speak the ERP's native protocol, a schema mapper that standardizes everything into the canonical graph, and an event bus that each co-pilot subscribes to. For on-premise ERPs, we install a lightweight agent in the DMZ. Only an outbound HTTPS tunnel is required—no inbound firewall rules—so IT security remains intact.

Brian Kalish: That underscores the importance of finance and IT collaboration. Each field has its language, and bridging that gap is essential. How did that collaboration work for you?

Dave Sackett: It was a team effort. IT managed the sandbox and access configurations. Finance handled application logic and user-related questions. So, responsibilities were naturally divided between the two teams.

Brian Kalish: Often, finance drives the initiative, but technical gaps can be barriers. Still, that shouldn’t stop us from adopting tools that help us make better decisions.

Dave Sackett: Agreed. Hyperbots is also very hands-on during implementation. It's truly a team approach.

Brian Kalish: How long does a typical integration take?

Dave Sackett: Typically 2 to 4 weeks, covering sandbox mapping, UAT, replay testing, and production cut-over. Most delays are due to customer testing windows. The technical work itself fits within a single sprint.

Brian Kalish: So this isn’t a replacement—it’s an enhancement, right?

Dave Sackett: Correct. It enhances the ERP. For example, I use it to automate accounts payable by connecting Hyperbots with Epicor Kinetic.

Brian Kalish: From your perspective, what should customers know in advance?

Dave Sackett: Connect with people who’ve already implemented it. Understand what worked well and what didn’t. Learn from their experience to avoid common mistakes.

Brian Kalish: What does Hyperbots require from a company during implementation?

Dave Sackett: Four things: API and service credentials, a sandbox UAT environment with realistic data, a 90-minute field mapping workshop, and one business unit to sign off on golden sample tests.

Brian Kalish: So, how do you verify that the integration is truly reliable?

Dave Sackett: We follow a three-step process: schema handshake, golden sample replay in UAT, and a 48-hour shadow run posting to a sub-ledger. Production is approved only if there are zero errors and latency is under 60 seconds. Initially, finance approves every transaction. Once stable, the process becomes automated.

Brian Kalish: What about ERP upgrades or migrations?

Dave Sackett: Most upgrades are enhancements rather than structural changes. Hyperbots’ connectors are versioned. During migration, we activate dual-write mode to post to both old and new ledgers. This ensures data integrity. Once validated, the old endpoint is retired. There’s no downtime and no data loss.

Brian Kalish: How do you ensure ERP transactions are successful?

Dave Sackett: Every post is followed by an immediate readback through the ERP API. The connector compresses and compares each field. Any mismatch triggers an auto-retry and is flagged in the exceptions queue. Nothing posts to the general ledger unless verified.

Brian Kalish: The system improves over time. Once the AI learns, it benefits everyone. That’s exponential learning.

Dave Sackett: Absolutely. That's why we chose Hyperbots. As our volume increases, we don’t need to hire more staff. Once the bot is trained, it processes everything accurately, around the clock, with strong audit trails.

Brian Kalish: It’s not about replacing people with AI, but enabling them. As we say, AI won’t replace people—people who use AI will replace those who don’t.

Dave Sackett: Exactly. The future of finance will rely heavily on AI. People must get comfortable using it.

Brian Kalish: Hyperbots’ finance-first approach is what makes it powerful. They’re solving real-world finance problems, not just building AI for its own sake.

Dave Sackett: Right. AP is just the beginning for us. We plan to expand into cash flow management, collections, purchasing, and more.

Brian Kalish: Thank you, Dave. To the audience—feel free to reach out to Dave, to me, or to the team at Hyperbots. It’s a powerful tool for improving efficiency, accuracy, and ROI. The future is here.

Dave Sackett: Thanks, everyone. Thanks, Brian.