adesso Blog

The release of ChatGPT in November 2022 marked the start of unprecedented hype around an AI technology. It took just five days to break the one million user mark - now there are over 100 million and the technology has become an integral part of many people's everyday lives. But at adesso, it's now: goodbye ChatGPT, hello adessoGPT! Over the past few months, our Group IT and colleagues from the Microsoft Business Line have developed our own adessoGPT based on OpenAI on Azure - by adessi, for adessi.

In this blog post, I would like to show you how we can use adessoGPT in our day-to-day work and which hacks improve the results of our prompts.

adessoGPT is an advanced AI technology in the form of a language model that is able to generate extensive texts and react to user input. The abbreviation "GPT" stands for "Generative Pre-trained Transformer" and adessoGPT is a special implementation of this model for dialogue and interaction with natural language. With our adessoGPT, we have all the functions we know from ChatGPT in a data protection-compliant format that fulfils all compliance requirements.

Use cases for adessoGPT

adessoGPT can be used in a variety of ways and the great thing about it is that, in addition to the general productivity increases (e-mail, communication, templates, etc.), there are also specific use cases for all areas. The following overview shows possible application scenarios of adessoGPT for various adesso departments and is intended to serve as a guide. These use cases individually and collectively lead to significant increases in productivity with comparatively low risks and costs.

Framework for the perfect prompt

The result of adessoGPT essentially depends on the text input. This text input is referred to as a prompt in this context. The following framework for the perfect prompt (developed by Dr Daniel Mühlbauer) improves the results when working with adessoGPT enormously. The framework consists of the following seven components:

  • 1. Role: AdessoGPT should be assigned to roles. We should be specific and clearly name and describe the role ("e.g. recruiter"). We can also add the main tasks of the role to make the answer more precise.
  • 2. Context: Give adessoGPT as much context as possible and describe it precisely. Contexts can be: our own position and tasks; the situation we are currently in; our own company can also be a context.
  • 3. Personas: The target group about which we want to obtain a result should be described precisely. Possible target groups are: Employees, applicants, social media networks, stakeholders, etc. We can also describe the interests and age groups of the personas.
  • 4. Goal/result: We should formulate a clear goal/result of what we want to achieve with adessoGPT (formulate overarching goal + describe further sub-goals).
  • 5. Tonality: Funny, informal, professional, formal, etc.
  • 6. Boundaries: No discriminatory content, the result should not violate laws, etc. We should show the limits of what we do not want to see under any circumstances.
  • 7. Reflection: Ask adessoGPT to explain the results for us by explaining how they arrived at the result and why it is the best possible result.

The following framework can simply be copied and pasted into adessoGPT. It then only needs to be adapted to the desired use case.

  • 1. You are [insert name of role]. Your task is [insert a short description of the main task here].
  • 2. Icurrently work as [insert role title here] in [insert a description of your professional situation here]. My tasks are [insert typical tasks here].
  • 3. My current target groups are [insert a brief description of the target groups here] aged [insert relevant age group here]. My target groups are particularly interested in [insert interests of target groups here]. My relevant stakeholders are [insert brief description of stakeholders here]. My stakeholders are particularly interested in [insert stakeholder interests here].
  • 4. My goal is [insert clear objective for adessoGPT here]. The [insert main focus of your objective here] is particularly important.
  • 5. The result should be formulated [insert relevant tonality here].
  • 6. Please make sure that the result does not violate [insert relevant moral boundaries here] or [insert relevant laws here].
  • 7. Justify your answer and explain why it is the best possible outcome.

It is important to mention that not every prompt needs to be formulated in such detail. The prompt should be formulated depending on the individual objective.


CustomGPTs offer the possibility to create your own adessoGPT bots according to your own ideas, plans and wishes. With CustomGPTs, anyone can build their "own" AI in just a few steps. For example, companies can train a CustomGPT model to communicate in a specific jargon or to respond to a specific type of enquiry that is typical for their business. In the long term, adessoGPT will also make it possible to create CustomGPTs for specific use cases. The following overview shows five possible CustomGPTs for adesso's everyday work:

  • Sales (objection handlingGPT): Training with best practices for objection handling, expertise from sales training etc. for negotiations with customers.
  • Human Resources (RecruitingGPT): Training based on successful application texts to be able to generate customised application letters from uploaded CVs/professional experience in a matter of seconds.
  • Corporate Marketing/Communications (CommunicationGPT): Training based on specific texts to generate a desired writing style (e.g. adesso writing style for employer branding). Books and poems by specific authors can also be used as training data to generate a specific style.
  • Proposal Management (ProposalGPT): Training based on successful proposals to match tenders even better with the adesso portfolio and to create customised proposals.
  • All areas (MentalHealthGPT): Training based on best practices on the topics of health, well-being, mindfulness, etc. to promote general well-being and cope with stressful situations.


It is undisputed that ChatGPT and adessoGPT have changed the world and everyday working life and will continue to do so in the future. The application examples described in this blog post represent only a small selection of the possible application scenarios for adessoGPT. The framework outlined above serves as a basic framework for optimising your own prompts, but the actual use should depend on the specific objective. Despite all the potential that adessoGPT offers us for everyday life, the importance of the human factor must not be neglected. On the contrary: people should be empowered to work with AI, whereby the aim should always be to achieve an interplay between AI and people in which people automate time-consuming routine tasks so that they can concentrate more on strategic issues and real human interactions.

Would you like to find out more about exciting topics from the world of adesso? Then take a look at our previous blog posts.

Also interesting:

Picture Erik  Langnickel

Author Erik Langnickel

Erik Langnickel has been Direct Search Recruiter at adesso since February 2023. In his current role, he is responsible for recruiting for the Leipzig, Dresden, Jena, Berlin & Potsdam locations. He also works in an LLM working group at adesso and specialises in unusual search methods such as recruiting via Twitter, GitHub or Stack Overflow.

Save this page. Remove this page.