Jaap de Koning of Witteveen+Bos wrote an interesting article about AI and tendering in Cobouw magazine.
He discusses the possibilities of using AI in tendering 🚀🤯 but he also describes a number of concerns and comments, such as:
"It is only a matter of time until EMVI writers will discover that EMVI plans can also be written very quickly by ChatGPT. After all, an EMVI plan contains answers to questions raised in a tender. And Chat GPT can then be a great help in writing a nice plan, only the question is whether the tenderer is able to assess the plan properly and check whether it really comes from the tenderer."
and on the struggle of humans coping with the amount of unstructured data (documents, drawings, text, images etc):
"In this struggle (what would the client actually mean by all this info, what is really important and what is not?), a digital AI powered tool could provide a lot of support when dealing with tenders."
ChatGPT is a powerful large language model developed by OpenAI that can generate human-like text. It is trained on a massive dataset of internet data and has the ability to understand and respond to a wide range of topics and questions.
One potential use for ChatGPT in tendering is to assist with the creation of responses to Request for Proposal & Information (RFP and RFI) and Request for Quotation (RFQ) documents. The model can analyse a RFP/RFI/RFQ and generate relevant and compelling responses, saving time and effort for the tendering team. Additionally, ChatGPT can assist with the creation of technical documents, such as EMVI documents and proposals, by providing accurate and detailed descriptions of products and services.
ChatGPT is trained on public data and therefore gives generic answers. So you won't be very distinctive with that. Added value is really only there when you can create output written in the context of the tender AND in the tone of your organisation AND based on previous (winning) plans and answers that you have submitted in similar tenders or subjects. It becomes much more unique, specific and of higher quality. But even then, you will still have to make adjustments as a human being. However, this can already be a big time saver on the time that goes wasted into looking up, validating and fine-tuning information that you will include in the plan or response.
So you will want to train your own GPT model on your own data. To make it specific, but also because your previous plans and answers are competition-sensitive information that you don't want to give (away) as input on a public (Chat) GPT API or service. You would want a private model in a private environment, with a powerful (domain-specific) model as a basis tutored on your own data.
This allows you to add extra information (to give the model more context (customer or tender specific) and with that, it's going to output super specific and relevant, texts and answers.
While ChatGPT and LLMs can be a valuable tool for tendering teams, tender managers and EMVI writers should not be worried about being replaced by the technology. ChatGPT is designed to assist with the tedious and time-consuming tasks, allowing professionals to focus on more strategic and creative work. Additionally, ChatGPT's output still requires review and editing by a human to ensure accuracy, completeness and compliance with the organisations standards.
Overall, ChatGPT is a powerful tool that can streamline the tendering process and assist with the creation of high-quality documents. However, it should be used as a supplement to, not a replacement for, the expertise and experience of human professionals.
At Brainial we are creating exactly that smart digital tender assistant to assist you with dealing with tenders, read more about our AI powered Tender Assist solution
Read the full article in Cobouw here:
https://www.cobouw.nl/311419/gaat-digitalisering-ons-helpen-bij-aanbestedingen
