How AI managers are redesigning workplaces with CompanyGPTs

A current study by McKinsey shows that 55% of employees have a positive attitude towards GenAI tools and would also like to work with them. In order to avoid the use of public ChatGPT accounts, i.e. shadow IT, companies should quickly create the appropriate framework conditions for AI-supported work. The increasing number of AI managers is a promising sign that this need is being met.

AI managers and their innovative field of activity

Generative AI means one thing above all for companies: numerous work processes can be completely rethought and made significantly more productive – provided the right tools are available and employees are open to working with them in an exploratory way. In practice, the latter usually proves to be the smaller hurdle: Knowledge workers often recognize the benefits of generative AI very quickly. It is much more challenging for decision-makers to find a Platform environment that can be used immediately with the company’s own data and at the same time enables the long-term implementation of a dynamic AI strategy. This is where hands-on AI managers come into play: with their technical understanding and intensive involvement with AI solutions, they build the bridge between technology and organization – making AI a real innovation opportunity for companies.

These tasks and skills make successful AI managers:

Credible combination of technical know-how and change management expertise to successfully lead teams through transformations
The ability to drive AI initiatives in parallel and measurably in the short, medium and long term
Strong risk awareness for data protection, ethical and technical issues – combined with a pragmatic view for realistic solutions
Sound knowledge of existing AI solutions – supplemented by own practical application experience

Milestone: Choosing the right CompanyGPT

For many companies, the introduction of a CompanyGPT marks the first decisive milestone on their AI journey. However, implementation rarely runs smoothly: according to Gartner For example, only around one in two companies that have opted for Microsoft Copilot actually use the solution across the board. There are also hurdles with ChatGPT in enterprise use – starting with the cost of data protection-compliant on-premise processing, which starts at around 50,000 euros.

A key task for AI managers is therefore to thoroughly evaluate potential solutions, to represent their assessments internally and to assert them against supposedly “established” providers. This is because specialized AI partners often offer a more intelligently put together technology bundle that enables an uncomplicated entry into new ways of working – and at the same time ensures the necessary scalability.

Benchmarking CompanyGPTs: 9 evaluation criteria

The following evaluation criteria are a good starting point for a structured comparison of CompanyGPTs:

1. Availability of multiple LLMs	LLMs such as ChatGPT, Mistral and Claude are becoming increasingly sophisticated and perform tasks such as searching, coding and text generation with varying degrees of success.
2. Secure integration of company data	In addition to hosting, simple upload options, API integrations, authorisation concepts and content monitoring are essential aspects for the secure handling of generative AI and company data.
3. Productivity tools	The availability of flexible work surfaces (canvas) and the potential integration of specialised AI assistants, such as automated email dispatch or data transfer to other third-party systems, should be available and expandable.
4. Multimodality	Depending on the company and use cases, the interpretation and processing of different data types such as text, images and video can be an important decision-making criterion.
5. KI-Workflows	An administration console in which AI processes can be transparently created and managed enables long-term, individually flexible and centrally managed CompanyGPT use.
6. Memory & Deep Research	For good search results and sophisticated knowledge processing, CompanyGPTs must have memory and deep research functionalities.
7. External KI Chatbots	CompanyGPT environments, which also enable the activation of AI chatbots for use cases at the customer interface, ensure the long-term implementation of an AI strategy.
8. Testing environment	CompanyGPTS will continue to evolve. Testing new features in a test environment enables companies to use AI in a controlled manner.
9. EU-Hosting-Options	GDPR-compliant hosting of CompanyGPTs can be most reliably ensured by operating on servers within the EU.