Use ChatGPT to Combat Bad Specifications?

One of the things that frustrates me to no end about our industry are bad specifications.

If you want to skip the story and dive right to the end – my ask today is that you comment below on what you would want an automated tool to check for when it reviews a set of specifications?

In other words, what issues have you found in specifications in the past that you would want an ideal tool to check for?

I’M GUILTY, TOO
Before I dive deeper and sound preachy, I have two disclaimers:

1. I am not above making poor specifications and I’m confident that I’ve issued bad specifications as a consultant – so I’m absolutely not above the violators I harp on here - and
2. If you are a consultant and care about the quality of your work output (true quality – as in doing quality work that helps foster a smooth project) – let me be clear that this is not directed at you. If you care to improve your work quality over time, you’re clearly not in the camp that I’m talking about today. The bad offenders, honestly, don't care - and it shows in their work.

WHAT MAKES A BAD SPECIFICATION?
What makes a bad fire protection specification?

The most dangerous is probably direction which would not meet code minimum. Ambiguity or conflicting information makes bidding difficult. Mandating things which don’t exist for the rest of the industry (such as velocity limitations in hydraulic calculations) can be unnerving and increase cost unnecessarily. Some of the most obvious parts of a bad specification are mandates for products or manufacturers that no longer exist.

The goal of a good specification is the same as the plans – clear, unambiguous communication of what is included and not included in a scope of work.

LITTLE RECOURSE
After a project is awarded, a contractor naturally has very little leverage to change the scope of work.
Perhaps there are cost-savings options that may be asked of a contractor.

Perhaps there’s a change in the project that opens up opportunities to revisit early design decisions.

But essentially, after contract award, there’s not a whole lot of leverage against complying with a bad set of specifications.

How do we address bad sets of bid documents in our industry?

If it’s life threatening and/or egregious, perhaps we could turn people into the governing boards. But how often is that done? How useful is it to permanently burn a bridge for reporting someone that may not even have any consequence?

The answer from those I speak with is almost never.

Consultants who don’t care about fire protection continue to issue plans and specifications, mostly the same as they always have, with little concern or incentive to change.

OUR INITIATIVE
Part of creating the community here is recognizing that uplifting everyone makes our industry better.

More knowledgeable contractors mean better detailed design and installations.

More knowledgeable plan review and inspectors means better policing and better final results across the board.

More knowledgeable consultants means that projects flow smoother, owners get what they need, and projects are more timely and on-budget.

Part of our responsibility here is to uplift the industry by sharing best practices and making helpful information & tools available that help us all do work better.

We have the educational piece (MeyerFire University), we have shorthand tools and cheatsheets. I write posts here. I have ideas in the works on helping improve access to basic, quality sets of specifications.

But what about now - as in today?

What is the best possible way to actually address a bad set of specifications that will get in the way of a smooth project?
 
PRE-BID RFIs
In my opinion, the most underutilized and best way to help foster a smooth project is challenging the scope before bid with a pre-bid RFI.

Pre-Bid RFIs (Request for Information) is a documented way to ask questions about the scope of a project before it is bid.
These can give an opportunity for a consultant to check their work, check their assumptions, give an opportunity to make a change if necessary, or give a chance to clarify an aspect of the scope.

Consultants can choose to play ball – help clarify the job on what should and shouldn’t be included. They can make changes if necessary, and allow bidders to bid apples-to-apples.

Contractors can also choose not to play – perhaps double down on the (incorrect) mantra of “this is the contractor’s responsibility to determine”, or something similar.

In either case, whether answered or not, Pre-Bid RFIs give the bidders either the information they seek or have greater permission (leverage?) to do as they see fit regarding the scope of the project.
 
SO MORE WORK FOR ME, JOE?
Crafting a good pre-bid RFI historically isn’t the easiest thing, though.

First – the writer has to digest enough of the project to write something coherent and competent – meaning they need to spend time looking through everything.

Second – pre-bid RFIs can sometimes have the presumption that a contractor is causing issues before they’re even on the job. This all comes down to the tone, silly as it might sound. If the pre-bid RFI is accusatory, that’s one thing. But if it’s written to help streamline a smooth project for everyone – then that’s a win for everyone.

Third – and perhaps the reason that pre-bid RFIs don’t happen as often as they should, is simply time. Bid days are time crunches. There’s a lot on the line. Going out of your way to clarify a project when you’re already on a time crunch can be tough.

This is the piece I’d like to help solve, and I think we can with some of your input.
 
THE CONCEPT
What if we had an automated tool that read a set of specifications and generated a helpful, appropriate, Pre-Bid RFI for your project?

While you’re reviewing the specifications and putting together your estimate, you do a 3-step copy and paste into ChatGPT (or something similar) that checks a whole host of specification issues and writes a Pre-Bid RFI for you?

You could have the time savings (huge), but also have AI do the work for checking for the 30 or 50 or 80 things that have been issues in the past – all stemming from specifications.

How convenient would that be?
 
If we could take the onus off of reporting bad players to state boards and instead focused on finding clean, appropriate, and easy ways to help make a project smoother for everyone – without adding any time burden – well that would be nothing short of awesome.
 
What I want to do from here is write a prompt and a step-by-step that I can share back with you all, that incorporates your list of grievances.

Essentially – everyone then has access to an easy way to gut-check specifications and get a custom-written Pre-Bid RFI out of it.

I need your input though to make it as useful for you as possible:
 
WHAT DO YOU NEED FROM ME JOE?
What I would love your input on is your answer to the following:

What have you seen in a specification that was clearly wrong which negatively impacted your project?

What have you found in a specification that makes bidding difficult, isn’t code compliant, or hurts the project?

I’m looking to create a list of checks that AI can do, for you, when it only has access to a project’s specification.

Comment below and let me know your thoughts – and in the next few weeks I’ll test and share a prompt and provide instructions back with you on how to use it. ​