The Impact of AI on RFP Writing

When ChatGPT steamrolled into everyone’s consciousness in late 2022, the predicted impact of AI on RFP writing wasn’t pretty.

“AI will make RFP writers obsolete.” “AI will kill the RFP writing business”. “AI will mean anyone can write an RFP in no time at all.”

It’s a bunch of months later and these predictions haven’t come to pass. Perhaps disappointingly, it’s still not possible to simply push a button and have AI write the perfect bid response!

That doesn’t mean RFP writers and suppliers aren’t integrating Artificial Intelligence into their work and their services. However, what’s occurring is more of a tool evolution than a writing revolution.

Yes, AI is useful for writing a first draft of something, scanning an internal database or the web for winning solutions, executing repetitive tasks and streamlining formatting. However, none of those is really new. Tools to do that have been around for quite a while now.

What AI hasn’t been able to do is replace the subject matter expertise, personality, security and overall organizational competency that comes from humans continuing to put many hours into some very key tasks like:

1. Interpreting what the issuer is looking for in a winning RFP bid.

Perhaps if every RFP request was well written and every issuer had a history of choosing bidders based on a consistent set of criteria, then AI could interpret what most issuers are seeking. However, that simply isn’t reality. Reality is that poorly written requests for proposal are common (this a key complaint of bidders, in the same way that many issuers complain about the quality of the bids they receive). Also, it’s not uncommon for issuers and their teams to issue RFP’s without being clear on their solution expectations – making it even more difficult/impossible to interpret the components of a winning bid. Finally, buyers needs and teams change – making it difficult to continuously use data from the past to predict success in the future. Given these hurdles, it seems unlikely AI will ever simply be able to pop out the perfect, winning RFP response.

2. Developing (and documenting) a unique winning solution.

Winning solutions are unique in two ways. First, they need to fit with the unique products, services, competencies, skills and competitive advantage of the bidding organization. It won’t matter what the winning solution looks like if it doesn’t fit with what the bidder can deliver.

Second (obviously), winning solutions need to fit with the unique characteristic of the issuer. Every organization is different, so every solution likely embodies some form of ‘customization’ – which could come in many forms, from cost to timing to management process to products and services that are part of the solution.

Sophisticated bidding organizations already use tools to library their solutions and profile their customers and competitors. So, while matching a doable, winning offering to any customer isn’t automatic, it’s certainly easier for some. However, the sophistication required to automate solution development is beyond the realm of the majority of RFP respondents. Without teams (or at least one highly skilled subject matter expert!) of people to define the customer, their needs and the best internal solutions available/possible, most RFPs could never be written. Significant investment in internal systems will be required before AI takes over the RFP writing process. In many cases, this investment will likely never happen.

3. Writing an accurate, personalized and secure RFP response.

ChatGPT’s stunning ability to magically write anything about anything is what helped AI break through the tech innovation noise so dramatically last year. However, once the original euphoria wore off, there was some key learning about important risks, like:

Accuracy. AI outputs are based on what’s predicted, not necessarily what’s real.
Security. ChatGPT is open-source software and uploaded information can become publicly available.
Personality. AI is discoverable and it’s possible for our most important customers to discover that our ‘authentic voice’ is artificial.

All of which leaves AI in an RFP writing pickle. Given that what’s written may not be accurate, secure or authentic, most organizations are not, cannot and will not rush into turning their RFP writing needs over to AI. It’s simply too risky.

Looking Ahead at the Impact of AI on RFP Writing

RFP writers and writing suppliers will continue to seek and find ways to use AI to improve response quality and reduce writing time and resources. It’s not like AI will be banned from the RFP writing world – it is WRITING after all.

RFP issuers are pursuing similar goals and seeking ways to integrate AI into their RFP and overall purchasing process. The large purchasers (like governments) will continue to bulk up on tools and information that will help them eek the most value out of their supplier relations.

Therefore, at some point down the road, perhaps it’s possible to foresee a completely automated, AI-driven buyer/supplier process. However, that point seems a few years off. Both the buyer and the supplier sides currently lack the sophistication necessary (some have it, most don’t).

Until then, AI will continue as a writing tool evolution and not a writing revolution.