This month’s blog has been adapted from our quarterly newsletter – please sign up using the sign-up form below, if you’d like to receive industry insights and tendering recommendations straight to your inbox!
Some do’s, some don’ts and some important points to consider…
As the use of AI continues to be talked about across industries and day-to-day life, we are taking this opportunity to highlight the risks of relying on AI when generating written content for tenders and bids.
We are mindful that, despite the caution we counsel on using AI for writing, some businesses will still turn to AI as part of their bidding process. We have therefore included some recommendations in the form of Do’s and Dont’s, to try and help you mitigate the risks we perceive, where a bot may be considered as an option to replace a good bid writer/manager… 😬
Of course, AI is not static, systems change and may improve. Today’s blog is written in the context of our experiences up to today.
AI & LLMs
Generative AI in the context of writing bids and tenders refers to the use of Large Language Models (LLMs) such as ChatGPT and Claude (there are others!). The basis of the writing outputs of LLMs is predicting the likely next words, based on the sources, context and prompts given by the user. An LLM does not “think” about the wording it produces, it uses probability calculations to predict what is likely come next.
When you are crafting a bid that makes your buyer/evaluator understand and connect with your offer, your bid writer will actively “think” about what is motivating the buyer in this tender process. They will develop content that engages with the evaluator. They will select the evidence and describe approaches that are most relevant to the buyer, their organisation and past experience with the subject matter of the tender.
Your writer will also challenge you and your team to think about relevant examples that resonate and connect. They will make you work hard to achieve top scores! Therein lies the essence of what will make your bid stand out.
We heard on a recent procurement (buyers) webinar, that bid submissions are starting to all sound more alike – making it hard to differentiate between bidder offers.
If you use AI (an LLM) to write your submissions, make sure you don’t allow it to make your bid read like all the others – an average bid will attract average scores.
The Do’s
- If you decide to use an AI/LLM solution, make sure you understand the security and confidentiality of using your chosen solution. If you use a generic open solution, you may be risking putting your content into the model for future use/farming/replication. Consider if this risks releasing sensitive or confidential data or IP?
- Challenge and robustly interrogate any output or content that an AI solution generates for you. Generative AI is widely recognised as often “hallucinating” when creating content – giving the impression of factual information that, quite simply, is not factual.
- When planning your reviews of AI generated content, include a verification review (of facts, assumptions) AND an accuracy review (is your business being reflected accurately?).
- Arrange for a human with a fresh pair of eyes (ideally someone within your business or close to your business) that understands your business USPs and culture to undertake content reviews. They will add much more value to a review than a machine. It’s important to remember tenders are still being evaluated and awarded by humans.
- Use LLMs to support you in being organised, if you need help with that. This could include inputting key dates and using it to create a planner, asking it to help structure a project plan.
The Don’ts
- Don’t use an LLM to write your all of your answers in their entirety. LLM’s have a habit of being very generic, and that’s not what makes a winning bid.
- Don’t take the content generated at face value. Remember responses should be a direct answer to the question – using the specification as a guide.
- Don’t use the AI to do the reading for you. The danger is you’ll miss something important and you need to be au fait with the full requirements and specification in order to put forward a winning bid.
- Don’t upload an entire bid pack to an open source AI/LLM. You may be in breach of the terms of accessing the tender documents.
- Don’t forget to keep a detailed log of how you have used AI in your bidding. Most public sector tenders now require the completion of an AI declaration.
Our experience & findings
The views and recommendations in this blog have been developed from the first hand experience of members of the Tender Victory team.
We have trialled AI/LLM solutions, one of which was a closed solution, specific to the bid writing profession. We had the permission of our clients, who were curious to see the results. Our key findings were:
- that the solution could not be relied upon to accurately create responses from past content sources (of our own accurate work);
- the amount of review and re-work required far outweighed the time it takes us to put pen to paper (fingers to keyboard!) and write fresh content, that responds directly to a) the question, and b) the specific buying organisation.
- that our clients were disappointed with the outputs and preferred we use our traditional approaches to planning and writing.
We found the AI/LLM bid writing process to be inaccurate and inefficient.
In the bid reviews that we undertake for clients when delivering our tender review service, we are noticing an increase in the use of AI generated content. Be mindful that if we can spot this, so can the buyers and their evaluators.
AI-free writing services!
At Tender Victory, we don’t use AI as a writing tool. Where we use AI solutions for research, findings are verified with the client or by using independent, reputable websites, before we rely on them.
If you have used AI to plan or draft your bid, but now realise it needs human interaction to improve its readability and connection with the evaluator – get in touch!
Disclaimer:
The team fully appreciates that there are some highly valid, industry specific, role specific applications of artificial intelligence that are ground breaking and life saving – we applaud such developments. As in such instances AI is being used as a tool to be used by humans to enhance their work. It’s our worries about the replacement of creativity, critical thinking and the ability to challenge information presented, that keeps us awake at night.
Yes, we used AI to create the “covers band” image 😉