HSP GRUPPE uses ChatGPT Enterprise to expand tax advisory capacity
HSP GRUPPE, a network of legally independent tax advisory, auditing and law firms, has used ChatGPT Enterprise across a shared workspace with Kanzleipakt covering 81 organizational groups. From 1 February to 14 July 2026, the workspace recorded more than 500,000 ChatGPT conversations, 84% weekly active usage, and 755 weekly active users. HSP estimates that AI-supported work could create more than 40,000 hours of additional annual capacity across the firm network.
The figures reflect a transformation programme rather than a simple software deployment. HSP has spent more than two decades digitising and standardising professional firm processes, embedding quality management and continuous improvement before generative AI entered its operations.
Enterprise AI embedded in professional workflows
HSP has introduced ChatGPT Enterprise for tax advisory, legal research, client communication, financial analysis and knowledge sharing. Monthly AI forums let employees exchange practical use cases, while internal requirements govern the use of client-related information, data protection and confidentiality. Professional review and final responsibility remain with the relevant tax, legal or accounting professional.
The network has also standardised repeatable applications into shared custom Agents. AI Client Communication supports first drafts, structure, clarity and a consistent client-oriented tone. The Booking Assistant SKR03 & SKR04 supports the preparation and classification of specific booking questions, with the employee making the final professional decision.
These uses are designed to reduce repetitive preparation work and make proven practices available beyond individual AI specialists. Partner Magdalene Posnak said that evaluating multiple real-estate investments fell from about nine hours to around two, leaving more time for client advisory. Managing Director Marco Sell uses AI-powered dashboards to explore financial scenarios with clients during meetings.
Usage, quality and capacity indicators
In HSP’s survey, 98.6% of employees reported higher productivity and 84.6% reported improved work quality. Weekly time savings were reported by 95.9% of respondents; 63.5% saved at least two hours a week and 25.7% saved at least five. The firm network also reported better client service for 79.7% of respondents and higher job satisfaction for 78.1%.
Its conservative scenario divides the estimated 40,000 additional hours into roughly 28,000 hours of productive, generally billable specialist work and about 12,000 hours for administration, client service and support. HSP estimates theoretical annual revenue potential of approximately €3.8 million using conservative hourly rates, while stressing that this is a capacity scenario rather than realised or guaranteed revenue.
Testing agentic workflows with ChatGPT Work
HSP has begun a small pilot of ChatGPT Work with developers and administrators before a wider rollout. The aim is to understand how agentic AI can automate complex workflows while balancing governance, quality and cost. The platform’s direction follows the model outlined in OpenAI ChatGPT Work agentic workflow model, where agentic systems are intended to carry work through to an outcome rather than merely return an answer.
Year-end accounting is one workflow under consideration. HSP is exploring whether ChatGPT Work could review bookkeeping continuously, identify missing information during the year and proactively request documents, so preparation is further advanced before an accountant opens the file.
The organisation has identified 88 automation opportunities and plans, with Kanzleipakt, to establish a joint DATEV-centric AI, firm process and development team. For professional-services businesses, HSP’s experience points to a practical priority: build governance and shared learning around AI, test workflows in small groups, and retain qualified professionals’ final judgement as capacity expands.