AI Removed the Cap on Bad Proposals. Who’s Checking Yours?
A manufacturing executive fired a 20-year vendor this week. Big deal, right?
Not over price. Not over service. Over a proposal AI wrote that promised something the vendor couldn’t actually build.
Again, not necessarily a big deal, except AI had added 1 additional line item this vendor missed; the client then expected it, and they couldn’t deliver on it.
That story belongs to Lou Leuzzi, from his latest piece in The Mentored Edge, and trust me, it is worth 5 minutes.
Business development and sales organizations are producing proposals faster now than at any point in my 25 years in this channel.
What happens next is not new. It is the oldest complaint in professional services consulting. Sales promises something, the deal closes, and then the delivery team is left holding the bag and picking up the pieces with a client who feels bait-and-switched on what they were promised.
Every consulting firm I have worked with has some version of that story.
That problem was always capped by human bandwidth. One person could only put together so many proposals in a week, which quietly limited how often it could happen.
AI just removed the cap. Same sales team, several times the output, and in most firms the same number of people checking it, which is very often nobody in particular.
Most sales leaders believe they “assume” someone downstream will catch whatever is off. That is a lot of assuming for a document that becomes binding the moment a customer signs it.
If your organization is using AI to build proposals, and I would guess most of you are, then somebody specific needs to own checking them before they go out the door. A named person, or a small team, with that as a real responsibility rather than something squeezed in around a quota.
And Lou’s advice on who that should be is pretty interesting and probably the best approach. Do not give it to your best closer, but instead, give it to whoever has the least to gain from the deal closing at all.
One last thing, and I went and verified this myself rather than take it on faith.
In January 2026, the Insurance Services Office issued 3 new generative AI exclusions for commercial general liability.
Chubb, Travelers and Berkshire Hathaway have since had AI exclusions approved by state regulators, who cleared more than 80% of the requests they saw. Berkley and others are pushing the same language into D&O, E&O and fiduciary lines.
The courts have been just as blunt. An Arizona appeals court held that citing fabricated AI-generated cases is sanctionable no matter what you intended.
So “the AI wrote it” is not a defense, and your policy may not cover the loss either I’m afraid.
Who owns checking the proposals leaving your organization this week, and can you name them right now?
Not over price. Not over service. Over a proposal AI wrote that promised something the vendor couldn’t actually build.
Again, not necessarily a big deal, except AI had added 1 additional line item this vendor missed; the client then expected it, and they couldn’t deliver on it.
That story belongs to Lou Leuzzi, from his latest piece in The Mentored Edge, and trust me, it is worth 5 minutes.
Business development and sales organizations are producing proposals faster now than at any point in my 25 years in this channel.
What happens next is not new. It is the oldest complaint in professional services consulting. Sales promises something, the deal closes, and then the delivery team is left holding the bag and picking up the pieces with a client who feels bait-and-switched on what they were promised.
Every consulting firm I have worked with has some version of that story.
That problem was always capped by human bandwidth. One person could only put together so many proposals in a week, which quietly limited how often it could happen.
AI just removed the cap. Same sales team, several times the output, and in most firms the same number of people checking it, which is very often nobody in particular.
Most sales leaders believe they “assume” someone downstream will catch whatever is off. That is a lot of assuming for a document that becomes binding the moment a customer signs it.
If your organization is using AI to build proposals, and I would guess most of you are, then somebody specific needs to own checking them before they go out the door. A named person, or a small team, with that as a real responsibility rather than something squeezed in around a quota.
And Lou’s advice on who that should be is pretty interesting and probably the best approach. Do not give it to your best closer, but instead, give it to whoever has the least to gain from the deal closing at all.
One last thing, and I went and verified this myself rather than take it on faith.
In January 2026, the Insurance Services Office issued 3 new generative AI exclusions for commercial general liability.
Chubb, Travelers and Berkshire Hathaway have since had AI exclusions approved by state regulators, who cleared more than 80% of the requests they saw. Berkley and others are pushing the same language into D&O, E&O and fiduciary lines.
The courts have been just as blunt. An Arizona appeals court held that citing fabricated AI-generated cases is sanctionable no matter what you intended.
So “the AI wrote it” is not a defense, and your policy may not cover the loss either I’m afraid.
Who owns checking the proposals leaving your organization this week, and can you name them right now?
Source: Lou Leuzzi, “What happens when everyone’s fast,” The Mentored Edge, 13 August 2026. [ADD SUBSTACK LINK]
hashtag#MicrosoftDynamics hashtag#NetSuite hashtag#ProfessionalServices hashtag#AIGovernance hashtag#SalesLeadership