Skip to main content

Automatic request answers / Question answering

With the advent of generative text systems it is possible to automatically answer request on certain topics with the help of artificial intelligence.

Text prompt​

It is essential to create the right text prompt so a language model is able to answer precisely.

By adding all relevant information to a prompt but with no content that shouldn't be exposed, a large language model is able to create very good support answers for simple topics.

It can be challenging to restrict the model in such a way that only relevant information is returned when a prompt is processed and no additional data originating from the language model itself.

Bias and wrong generated data​

It's essential to note that while generative text models are powerful, they are not without challenges, such as the potential for generating incorrect or biased information. To maximize their usefulness and mitigate these issues, organizations often employ a combination of automated responses and human oversight, ensuring that critical decisions or sensitive issues are handled by human agents, while routine and repetitive tasks are automated. This hybrid approach strikes a balance between efficiency and quality.

Main Benefits​

  • Instant Responses: Users receive immediate answers to their queries, enhancing the overall user experience and reducing wait times.
  • Cost-Efficiency: They help organizations reduce labor costs associated with customer support and repetitive inquiries.
  • 24/7 Availability: These answers provide support or information at any time, even outside regular working hours, improving customer service accessibility.

Example: Party by the barn​

The prompt in this example contains all relevant information on a particular topic. E.g.:

The topic is: A party in the barn on Nov 27th 2023 10 o'clock without cars.
Please answer the question below as a short e-mail:

When this is concatenated with a question of a potential attendee like

When exactly is the party and where?

The language model can elaborate on the topic and is able to answer the question based on the information given in the topic.

Try it out by opening the example in Examples → E-Mail → AI Auto Answer.