Assorted Insights on AI

I’ve spent my mornings this week attending a virtual conference on Artificial Intelligence (AI) and its use/potential use by proposal professionals. Attendees were global and across industries. Proposal formats included everything from questionnaires to open responses.

Now, I know that lots of you have had a deep dive into artificial intelligence technology and some of you have barely gotten your toes wet. I thought I had a good working knowledge of AI and implications for proposal world. Now I know more. No judgement here, just things I learned, realized, or had clarified. Here goes!

1. No magic wand. Artificial Intelligence is not the panacea sent to cure all RFP team ills. It is not omniscient. It is not all-knowing. Is it impressive? Yes. Is it going to do some heavy lifting for RFP writers? Yes. But its immediate use is specific to first drafts, document summaries and editorial reviews.

2. Words have power. The AI being used by proposal software vendors is based on Large Language Models (LLMs). Predictive text is their game. Particularly for teams looking for AI to also tackle numerical data retrieval and input in their proposal responses, this is important to understand (Asset Management, I am looking at you). I know that other AI capabilities can do a wide range of analytical tasks, but most established software vendors for RFP work are rolling out capabilities for first draft generation, RFP requirements compliance and editorial consistency.

3. Structured vs. Unstructured Content. Structured content is what legacy RFP teams are familiar with - your knowledge library, assigned SMEs, content reviews, etc. Unstructured content is your intranet (or the broad internet, which is crazy business for RFP teams). Unstructured content relies on having that content being accurate and current outside of the RFP Team’s process controls. I don’t know about your organization, but I did a search for a specific policy at my old (very large) firm and came up with documents dating back to 2009 on its intranet.

   Depending on your firm’s oversight requirements, things like content ownership, periodic review and approval trails, and version controls may not be visible to an RFP writer as the content goes into proposal.

4. “Garbage In, Garbage Out” will still be the bane of an RFP team’s existence. I happen to believe that bringing AI online in an organization will exacerbate this principle as SMEs mistakenly assume that proposal software will access only correct, up-to-date, and approved source materials for content generation. This is especially true when SMEs work on updating other marketing materials.

   Many business partners that I worked with assumed that material changes in presentations, pitch decks and one-pagers get channeled to the RFP Team for content updates. One of my RFP team has reported directly to the CMO remained unaware of significant organizational changes and updates in marketing materials. Does anyone talk to the RFP team?

5. Content is King. AI will not replace or make up for a failing content management ecosystem. Teams must have a set process for creating content, reviewing content, and revising content. This process must account for net-new content and for content that is revised in the proposal draft process.

   Depending on your industry and organization, you might need to factor in regulatory compliance officer reviews and legal reviews in addition to the original content contributor acting as an ongoing content owner.

   Teams that have weak content maintenance - no published content process, no SLAs around content creation/maintenance, no consistency of team expectations and actual follow-through - will continue to sacrifice quality content at the altar of “too busy” and “don’t have time.” AI sourcing bad, outdated content will only highlight this organizational dysfunction. Address it now.

6. “That’s the answer the system gave me.” You’ve all had this writer on your team. The value of AI content generation is the speed to market of a first draft. There is danger in quickness versus quality.

   It assumes that an RFP writer will read the AI generated response and is knowledgeable enough to assess where the first draft might not be accurate. Your process may parse out the whole draft to SMEs for review and not rely on a writer to identify areas that need particular attention. Either way, now the review process may get bogged down in rework - the devil to process efficiency.

   Luckily, this was a topic of discussion on a couple of panels where there is assurance that future AI development is looking to address response evaluation in addition to quick generation. Whew!

My big takeaway? RFP teams need someone on their team to jump on Artificial Intelligence science, build a working knowledge. You’ll need to understand implications of bringing a proposal software vendor capability online, with the cybersecurity hoops that it might entail. You’ll have to make some decisions about what your AI will access and where it will get its information. You’ll need to wisely implement AI capabilities as technology tools that support your RFP writers.

You will also have to manage your leadership’s expectations around what new AI capabilities will accomplish. Consensus at this conference (and others that I have attended this year) was that AI absolutely will not replace the human element in the RFP/DDQ/proposal process. It may be good, but it’s not that good.

AI is a white hot topic. It holds great promise for improving strategic proposal team outcomes. Everyone is looking to do more with less, to do better than the rest of the marketplace. But, AI does not stand for “Achieve the Impossible.” Get out in front of that while you can!