Der HubSpot-Blueprint: Aufbau eines von der Buchhaltung geleiteten KI-Kraftwerks

HubSpot (NYSE: HUBS) is the agentic customer platform that delivers seamless connection for customer-facing teams. With a unified platform that includes AI-powered engagement hubs and a Smart CRM, HubSpot helps over 288,000 customers in more than 135 countries connect and grow better. This connected ecosystem is further supported by over 2,000 App Marketplace integrations, a global community network, and premier educational content.

Committed to a "walk the walk" mentality, HubSpot ensures its internal operations are as tech-forward as the solutions it provides to customers. By leveraging its own platform to prioritize innovation and transparency at scale, HubSpot serves as its own best use case for how a unified, AI-powered strategy can drive global growth.

To honor this culture, Director of Accounting Blake Winchester refused to settle for reactive, spreadsheet-bound workflows or wait for IT-led automation that typically required significant technical maintenance. By taking the lead with FloQast AI Agents, HubSpot’s accounting team built the automation themselves. This shift transformed manual processes into scalable, audit-ready workflows, eliminating the internal friction of workbook assembly that previously drained the team's time. Today, the accounting function operates as a strategic powerhouse moving at the true speed of the business.

‍

The Mission: Bringing AI Innovation Into Accounting

As a company, HubSpot is deeply invested in AI across its product and operations. For Blake Winchester, Director of Accounting, that raised a clear question. If AI is transforming how the business operates externally, how should it reshape accounting internally?

“HubSpot leans into AI in our product, and internally, we want to walk that walk as well,” Blake explained.

To meet that standard, the accounting team needed to move beyond manual, spreadsheet-driven workflows and rethink how work gets done. The goal was not just incremental efficiency, but building a function that could operate with the same speed and adaptability as the broader business.

The "Build vs. Buy" Paradigm

A critical part of this mission involved evaluating the "Build vs. Buy" dilemma. While HubSpot possesses elite internal talent, the company’s own AI Task Force—the group specifically tasked with implementing AI across all departments—delivered a clear verdict: FloQast Transform was the superior path.

Reflecting on the decision, Blake noted that the AI Task Force recognized the high cost of "homegrown" technical debt. “Our internal AI Task Force that looks at these things—they said, ‘This is a better option than us trying to build this for you,’” Blake explained.

The reasoning was rooted in long-term sustainability. Previous internal automation efforts had proven difficult to scale. “It’s taken a lot of technical lift, a lot of development time, and a lot of maintenance at times,” Blake noted. By choosing FloQast, HubSpot avoided the "Maintenance Trap" of a custom build. Instead of waiting for a long IT development cycle, the accounting team was empowered to be the architects of their own workflows, deploying audit-ready automation that moves at the speed of the business without draining internal resources.

The Challenge: Spreadsheet Friction at Global Scale

Despite HubSpot's forward-thinking culture, Blake Winchester approached the AI landscape with caution. As an existing FloQast customer, HubSpot had already streamlined their month-end close. However, after navigating nearly two years of AI buzzwords and market hype., Blake remained skeptical that AI could handle the granular complexity of accounting across more than 15 global subsidiaries.

‍

{emphasize}

“I walked away from our initial conversations with a healthy level of skepticism. Many tools make bold claims, so my message to the team was simple: ‘Show me the actual output.’”

— Blake Winchester, Director of Accounting, HubSpot

{emphasize}

‍

The "noise" Blake had to cut through was rooted in systemic friction that traditional automation had failed to solve:

• The Maintenance Trap: Legacy automation required heavy IT lifting and constant upkeep.
• The Information Gap: Accountants were acting as "detective researchers," manually digging through AP mailboxes for context.
• Workbook Friction: “Assembling the same workbooks twelve times a year for fifteen different subsidiaries is a significant drain on talent. We needed to eliminate that friction so our team could focus on higher-value analysis.”

The Turning Point: Proving AI in Real Workflows

The turning point for Blake’s skepticism came during a Proof of Concept (PoC) for accrual workflows that revealed how the technology actually worked. Blake was not looking for another generative text tool. He wanted a technical engine capable of handling HubSpot's unique "flavor" of accounting. During the walkthrough, he saw that FloQast Transform translated accounting logic into repeatable, auditable Python code.

‍

{emphasize}

“During the demo, I asked the team to walk me through the mechanics. Seeing that the system produces repeatable Python code was the turning point. It ensures the output is consistent and reliable, regardless of when the workflow is triggered.”

— Blake Winchester, Director of Accounting, HubSpot

{emphasize}

‍

The realization unlocked a new level of automation for the team. Instead of relying on static workflows, HubSpot’s accountants could interact with the system to refine workbooks in real time. Because the automation is built on traceable code and live formulas, the logic behind every output remains transparent. It allows automation to support accounting workflows while keeping accountants in control of the underlying logic.

The Catalyst: An AI Agent Hackathon

With confidence in how the system worked, HubSpot needed a way to move from understanding the technology to applying it in practice. To accelerate adoption, Blake organized a five hour accounting AI hackathon. The goal was simple. Give the team hands-on experience building automation using their own workflows.

Instead of watching a demo, accountants were actively building agents using real accounting processes. Each participant brought a workflow they understood deeply and worked through how it could be translated into automation.

‍

{emphasize}

“The collaborative energy in the room was incredible. You could see the team coaching one another on prompt engineering and logic. By the end of that session, our communal knowledge regarding AI application was remarkably strong.”

— Blake Winchester, Director of Accounting, HubSpot

{emphasize}

‍

The format removed the abstraction that often surrounds AI. Rather than discussing possibilities, the team saw exactly how their logic could be implemented and refined in real time.

By the end of the afternoon, the team had moved from curiosity to capability. They saw firsthand that accountants could design and deploy automation themselves.

Automation Built for Audit Confidence

As a public company, HubSpot requires automation that is not only transparent, but also scalable and audit ready across global operations. A major hurdle for AI adoption in finance is the trust gap. If an auditor cannot trace the logic, the tool creates more work than it saves. HubSpot prioritized an approach that internal and external auditors could understand and verify.

‍

{emphasize}

“For a public company, auditability is non-negotiable. Because the logic is traceable and the output is consistent, both our internal and external auditors can gain the necessary comfort to rely on these automated processes.”

— Blake Winchester, Director of Accounting, HubSpot

{emphasize}

‍

Rather than introducing a separate layer of automation, FloQast AI Agents operate directly within the spreadsheet environment. Reviewers can click any cell to see how a calculation was performed, making it easy to validate outputs during both internal review and external audit. 

‍

‍

Accrual & PO Enrichment Agent

Recognizing that every department has its own "flavor" of accruals, HubSpot built an agent to standardize logic across the board while automatically enriching the process with uploaded Coupa invoice and PO data. By providing the critical context—such as unapproved invoices and open PO balances—before the review even begins, the agent is projected to reduce preparer time by 50% and allows the team to focus on investigating variances rather than formatting cells.

Prepaid Reconciliation Agent

HubSpot transitioned from a manual process where three staff members spent six hours monthly assembling 15 individual subsidiary workbooks to a single, automated run. Now, one person runs the agent to produce a single consolidated workbook for all 15 entities in just 1.5 hours per month—a significant gain in efficiency and data centralization.

What began as a proof of concept is now a model for how automation can operate reliably within a controlled, audit ready environment.

Strategic Impact: From Preparer to Reviewer

With FloQast AI agents handling much of the manual preparation, HubSpot’s accounting team has fundamentally shifted how work gets done. 

Instead of spending time assembling data, accountants now begin their work at the point of review and analysis. This allows the team to focus on identifying variances, investigating anomalies, and applying judgement where it matters most.

‍

{emphasize}

“We’ve redefined our roles: AI serves as the new preparer. Our staff and seniors have transitioned into first-level reviewers, allowing leadership to focus more deeply on high-risk areas and strategic initiatives.”

— Blake Winchester, Director of Accounting, HubSpot

{emphasize}

‍

The result is a team focused on higher value outcomes:

• Exception-Based Management: Seniors now begin their day by reviewing outputs instead of verifying formulas and links.
• Strategic Business Partnership: Accountants spend more time working with department leaders to understand spend trends rather than gathering raw data.
• Managing Scalable Complexity: By automating baseline workbook assembly, the team has the capacity to manage increasing complexity within the prepaids area, specifically as more vendors transition to consumption-based billing models. This allows the agent to handle the heavy lifting while the team focuses on these evolving, high-touch financial models.

Cultural Impact: Building the Accountant Architect

The most significant shift at HubSpot hasn't been in the software, but in how the team defines expertise.  Blake is intentionally raising the bar, moving beyond the era of the Excel expert toward a model centered on AI fluency and toward the era of AI Fluency.

‍

{emphasize}

„Die Definition eines Experten auf diesem Gebiet hat sich geändert. Wir suchen nicht mehr nur nach Excel-Kenntnissen, sondern auch nach KI-Kenntnissen — der Fähigkeit, Agenten aufzubauen und dieses technische Wissen im gesamten Unternehmen zu teilen.“

— Blake Winchester, Direktor für Buchhaltung, HubSpot

{emphasize}

‍

Diese Veränderung verändert auch die tägliche Arbeitsweise des Teams. Wie Blake es ausdrückt: „Wir konzentrieren uns auf die berufliche Entwicklung. Unserem Team die Möglichkeit zu geben, eigene KI-Agenten zu entwickeln, ist ein großer Gewinn für ihre Karriere und das Unternehmen.“

Was einst eine Funktion war, die sich auf die Ausführung konzentrierte, wird zu einem Innovationszentrum innerhalb des Unternehmens.

Das HubSpot-Playbook: 3 Schritte zur KI-Transformation

Blake Winchesters Rat für andere Buchhaltungsleiter:

1. Beginne mit einem PoC „Zeig es mir“: Wählen Sie einen aufwändigen Prozess aus und fordern Sie einen Machbarkeitsnachweis an, der zeigt, wie die Logik tatsächlich funktioniert.
2. Desinfizieren Sie zuerst den Prozess: „Tabs sammeln sich an und der Flaum bleibt im Laufe der Jahre dort. Gestalten Sie den Prozess neu, bevor Sie ihn in den Agenten einbinden.“
3. Fly Low mit dem Team: „Verbringen Sie Zeit mit den Erstellern, um genau zu verstehen, wie die Arbeit erledigt wird. Die besten Ideen für Agenten kommen von den Leuten, die der manuellen Reibung am nächsten sind.“

Bei HubSpots Ansatz geht es nicht nur um die Implementierung von KI. Es geht darum, die Arbeitsweise von Buchhaltungsteams neu zu definieren.

Durch die Kombination von Transparenz, Kontrolle und praktischer Umsetzung haben sie ein Modell geschaffen, bei dem Buchhalter die Automatisierung nicht nur nutzen, sondern auch umsetzen. Dies ermöglicht es dem Team, Arbeitsabläufe kontinuierlich zu verbessern, sich schnell an neue Geschäftsmodelle anzupassen und Abläufe zu skalieren, ohne den manuellen Arbeitsaufwand zu erhöhen.

Da immer mehr Teams diesem Weg folgen, wird sich die Rolle des Rechnungswesens von der Ausführung von Workflows hin zur Entwicklung von Systemen weiterentwickeln, die Effizienz, Genauigkeit und bessere Geschäftseinblicke fördern.