Glonce

HSP GRUPPE deploys ChatGPT Enterprise across 81 tax firm groups

toolsThe Glonce editorsOpenAI🔥53Glonce rating53Impact45Novelty30Relevance88Surprise25Players80Credibility78
HSP GRUPPE deploys ChatGPT Enterprise across 81 tax firm groups

OpenAI published a customer case study on HSP GRUPPE, a network of legally independent German tax advisory, auditing and law firms that has invested in digitizing and standardizing its processes for more than two decades. HSP shares a ChatGPT Enterprise workspace with Kanzleipakt, covering 81 organizational groups. Leadership framed the rollout as an organizational transformation rather than a software deployment, embedding ChatGPT Enterprise into tax advisory, legal research, client communication, financial analysis and knowledge sharing, backed by monthly AI forums and internal data protection and governance rules; professional review and final responsibility stay with the relevant tax, legal or accounting professional. Successful use cases were standardized into shared custom Agents, including AI Client Communication, which drafts client messages for tone and clarity, and the Booking Assistant SKR03 and SKR04, which supports classification of bookkeeping entries; the final decision in both cases remains with the employee. Individual professionals also built on these foundations: Managing Partner Frank Heibel uses ChatGPT as what he calls a technical sparring partner for complex tax questions, Partner Magdalene Posnak cut the time to evaluate multiple real estate investments from nine hours to around two, Managing Director Marco Sell uses AI-powered dashboards with clients during meetings, and lawyer and tax advisor Jan-Henrik Leifelt uses it to support, not replace, his professional judgment. Over the six months from February 1 to July 14, 2026, the shared workspace reached 84% weekly active usage, with 755 weekly active users among 913 unique users, and the headline stat block cites more than 500,000 ChatGPT conversations for the period, while the body text separately states that users exchanged more than 500,000 messages with ChatGPT over the same period; the source does not clarify whether the two figures describe the same underlying count. In an internal survey, 98.6% of employees reported higher productivity and 84.6% reported improved work quality; 95.9% reported weekly time savings, with 63.5% saving at least two hours a week and 25.7% saving at least five; 79.7% reported better client service and 78.1% reported higher job satisfaction. HSP has not yet assessed the full financial impact of the rollout and plans to do so later this year, but a deliberately conservative internal scenario estimates about 40,000 hours of additional annual capacity, roughly 28,000 of them billable specialist work and about 12,000 in administration, client service and support; at conservative hourly rates, HSP puts a theoretical annual revenue potential on that capacity at approximately 3.8 million euros, which the source stresses is a capacity scenario, not realized or guaranteed revenue. HSP says the point is not headcount reduction: the network firms are already at capacity with substantial backlogs, so time saved is meant to go toward more client work, shorter turnaround times and stronger client service. HSP is now piloting ChatGPT Work, first with a small group of developers and administrators before a wider rollout, to test how agentic AI can automate complex workflows while balancing governance, quality and cost. Carsten Schulz points to year-end accounting as an example: today, accountants often find missing information only when preparing annual accounts months after bookkeeping is done, and HSP is exploring whether ChatGPT Work could continuously review bookkeeping, flag gaps throughout the year and proactively request documents from clients. Schulz says HSP identified 88 automation opportunities but considers that framing too narrow, arguing the real question is redesigning processes rather than automating existing ones. Together with Kanzleipakt, HSP plans to set up a joint DATEV-centric AI, firm process and development team to turn successful decentralized experiments into documented, scalable reference workflows as ChatGPT Work and agentic capabilities mature.

Key facts

  • HSP GRUPPE, a network of independent German tax advisory, auditing and law firms, shares a ChatGPT Enterprise workspace with Kanzleipakt covering 81 organizational groups.
  • Over six months (February 1 to July 14, 2026) the workspace hit 84% weekly active usage among 913 unique users, with the case study citing over 500,000 ChatGPT conversations in the headline stats and over 500,000 messages exchanged in the body text for the same period.
  • 98.6% of surveyed employees reported higher productivity and 95.9% reported weekly time savings; Partner Magdalene Posnak said ChatGPT cut her multi-property real estate investment analysis from nine hours to around two.
  • A deliberately conservative internal estimate puts additional annual capacity at about 40,000 hours (roughly 28,000 billable, 12,000 administrative), worth a theoretical, unrealized annual revenue potential of approximately 3.8 million euros at conservative hourly rates.
  • HSP is piloting ChatGPT Work with a small group before wider rollout and, with Kanzleipakt, plans a joint DATEV-centric AI and firm process team to scale successful experiments into documented workflows.

Why it matters

The case study is OpenAI's evidence that enterprise AI adoption can move past individual productivity gains into an organizational capability inside a regulated, judgment-heavy profession. HSP built shared custom Agents on top of ChatGPT Enterprise rather than leaving usage to ad hoc habits, and it is now testing whether agentic tools can restructure entire workflows, such as year-end accounting, instead of just speeding up existing tasks.

Who it affects

Tax advisory, audit and law firms weighing AI adoption, particularly in Germany's DATEV-centric accounting ecosystem that HSP and Kanzleipakt plan to build a joint AI team around. It also matters to OpenAI's enterprise customers more broadly, since the story doubles as a template for governance, custom Agents and staged rollout that other professional services firms could copy.

How to use it

HSP paired ChatGPT Enterprise with monthly AI forums for sharing use cases and built two shared Agents: AI Client Communication for drafting client messages, and the Booking Assistant SKR03 and SKR04 for classifying bookkeeping entries. In both cases, professional review and final responsibility stay with the relevant employee. ChatGPT Work is being piloted first with a small group of developers and administrators before wider access, a staged approach the firm recommends generally: pilot new capabilities small, then expand.

How solid is it

This is a case study OpenAI published about its own customer, built on HSP's self-reported internal survey data rather than independent verification. HSP itself says it has not yet assessed the full financial impact of the rollout and plans to do so later this year, and it explicitly labels the 3.8 million euro figure a theoretical capacity scenario, not realized revenue. The source also leaves an internal inconsistency unresolved: the headline stat block cites 500,000-plus ChatGPT conversations for the six-month period, while the body text separately cites 500,000-plus messages for the same period, without clarifying whether these are the same count under different labels.

Risks and caveats

The productivity and satisfaction figures come from a survey of HSP's own employees promoted in a vendor-authored case study, not an outside audit, and the capacity and revenue estimates are described as deliberately conservative projections rather than measured outcomes. The source gives no headcount, revenue or size figures for HSP or Kanzleipakt as organizations, so the reported usage and capacity numbers cannot be benchmarked against the firms' overall scale. HSP's governance model keeps professional review and final responsibility with human specialists throughout, which the source presents as the safeguard against over-reliance on the tools.

“We identified 88 opportunities for automation. But I think that's already thinking too small. The real question isn't how we automate today's processes. It's how we redesign them. That's what opens the door to a completely new world.”

— Carsten Schulz