General FAQs
How accurate is it?
BidLevel is highly accurate, but 100% isn’t the goal. It’s designed to assist, not replace, the CA/QS
Can I trust it?
In terms of accuracy, it’s best used when you trust but verify (as with all A.I, shouldn’t trust it blindly) - Use the Side x Side view to help verify the extraction
In terms of security, yes - data is secure and handled under strict isolation and privacy standards (see Security White Paper & A.I Policy).
Can I invite my colleagues?
Yes - within the app. Just Open the Side Nav and you can find a button to Invite your Team down the bottom.
Whilst you can invite other people into BidLevel, we don't currently have a concept of "teams" or "Projects" or sharing of packages yet.
What’s in the roadmap of improvements?
Short term roadmap:
Plug prices for excluded items (highest / lowest / average)
One click RFI clarification email based on assumptions and exclusions
Exceptions Report/mark up → showing what hasn't been picked up in the comparison, but is included on the quote
Improved handling of large / complex packages
Can I plug numbers (eg: highest, lowest, avg)?
Not yet - planned for an upcoming release.
Can BidLevel create an clarifications list?
Not currently - planned for an upcoming release.
What about the words/context around the numbers (which are critical)?
BidLevel analyses both numbers and descriptions.
Flags where specs or wording differ (via Notes)
Still need to read quote?
Yes - BidLevel has been designed to assist and accelerate the CA/QSs but doesn’t replace their judgment (and isn’t intended to).
Use the Side x Side feature by clicking "Open bid review".
Will this replace my job?
No - it’s built to make CA/QSs faster and more productive, not redundant. Your insight, value engineering and analysis is still essential.
Why OpenAI? Aren't there better models?
The team tested multiple AI models (and continues to test them).
OpenAI currently provides the best balance of performance and reliability.
However, true accuracy comes from updating logic and prompting the AI to handle the domain specific scenarios vs which model it is.
If you're using OpenAI can't I just put my quotes in ChatGPT and have it do this for me?
No - BidLevel is trained specifically for construction quote comparison, achieving ~90% accuracy and source traceability.
The BidLevel team tracks failures, experiments, and updates logic to handle quote-specific complexities.
Generic ChatGPT can’t reliably handle multi-quote comparisons and doesn’t preserve data security.
Do you use my data/usage to improve the product?
Refer to our A.I Policy.
We already use ProcurePro’s set Pricing Comparison, how does this fit in?
Use BidLevel for a fast, first-pass comparison of received quotes.
Review extractions side-by-side in BidLevel to get familiar before exporting.
Export results to Excel, then copy the data into ProcurePro’s Excel Comparison
Products are separate for now but will integrate over time, while still working as standalone tools.
How much has this been tested?
Live with 600+ users
Extensively tested internally with construction professionals.
Continuously improving prompts to capture construction-specific edge cases.