Hello HN,
I’m a solo developer building tools for real estate workflows. I built OfferGridAI after watching listing agents repeatedly struggle with the same problem during hot markets.
When a property gets multiple offers, each offer usually comes in as a 10–20 page PDF. Under tight time pressure, agents have to manually dig through each document and rebuild a spreadsheet to compare things like price, net to seller, contingencies, financing, closing timeline, escalation clauses, etc. It’s not conceptually hard, but it’s stressful, time-consuming, and easy to miss details buried deep in the PDFs.
I wanted a way to make that moment less chaotic.
The idea:
Upload multiple offer PDFs → extract the key terms → generate a clean, side-by-side comparison grid that’s easy to walk through with a seller.
Instead of just dumping text, the tool normalizes the information into comparable fields (price vs net, contingencies, financing strength, days to close) and adds a short summary highlighting tradeoffs (e.g. highest price vs highest certainty to close).
What it focuses on:
Structured extraction of common purchase-agreement terms
Normalizing offers so sellers can compare apples to apples
Producing a seller-ready grid rather than raw AI output
What it intentionally does not do:
Make decisions for agents or sellers
Replace professional judgment
Integrate with MLS or transaction management systems (at least for now)
The goal is to be a fast decision-support tool for a very specific, high-pressure moment.
I’m early and still refining the scope, especially around:
Which fields matter most in practice
How to communicate “risk” without over-claiming
How tolerant users are of “best effort” extraction vs perfection
I’d love feedback from anyone who’s worked with complex PDFs, document comparison, or decision-support tools under time pressure, or from anyone who’s built vertical SaaS in heavily regulated industries.
Happy to answer questions and learn from the community.
For covering the risk of mistakes I suggest considering ways of "visually quoting" the documents.
If the summary says "closing timeline: X" but there's an icon I can click that pops open an overlay with a visual cropped screenshot of that part of the original PDF - maybe even with a red circle around that detail - I can trust those summaries a whole lot more.
Gemini 2.5 has image bounding box and masking features that can help with this (sadly missing from Gemini 3.)
Because it’s just using structured response so it should be doable with Gemini 3 ? (We are using Gemini 3 for some docs processing and its visual understanding is just incredible)
I've never owned a home and would like to try to buy one in the next year or two. There doesn't seem to be much in the way of API's/software tools that let you analyze historical data and prices of listings in specific areas.
How can I get my hands on the right information to make sure I don't get ripped off?
> I assume this means the reviews under "Trusted by Listing Agents" are AI generated as well?
There's a question mark here, but this is not a question, it's an uncharitable statement. You know what happens when you assume.
Try something like "if you have no paying customers, how do you have reviews?" Even asking the question helps you think about the perceived conundrum, because your next thought might be "are paying customers necessary for reviews?" (no), which would probably lead you to solve your mystery yourself.
When a property gets multiple offers, each offer usually comes in as a 10–20 page PDF. Under tight time pressure, agents have to manually dig through each document and rebuild a spreadsheet to compare things like price, net to seller, contingencies, financing, closing timeline, escalation clauses, etc. It’s not conceptually hard, but it’s stressful, time-consuming, and easy to miss details buried deep in the PDFs.
I wanted a way to make that moment less chaotic.
The idea: Upload multiple offer PDFs → extract the key terms → generate a clean, side-by-side comparison grid that’s easy to walk through with a seller.
Instead of just dumping text, the tool normalizes the information into comparable fields (price vs net, contingencies, financing strength, days to close) and adds a short summary highlighting tradeoffs (e.g. highest price vs highest certainty to close).
What it focuses on:
Structured extraction of common purchase-agreement terms
Normalizing offers so sellers can compare apples to apples
Surfacing risk factors (financing type, contingencies, timeline)
Producing a seller-ready grid rather than raw AI output
What it intentionally does not do:
Make decisions for agents or sellers
Replace professional judgment
Integrate with MLS or transaction management systems (at least for now)
The goal is to be a fast decision-support tool for a very specific, high-pressure moment.
I’m early and still refining the scope, especially around:
Which fields matter most in practice
How to communicate “risk” without over-claiming
How tolerant users are of “best effort” extraction vs perfection
I’d love feedback from anyone who’s worked with complex PDFs, document comparison, or decision-support tools under time pressure, or from anyone who’s built vertical SaaS in heavily regulated industries.
Happy to answer questions and learn from the community.
If the summary says "closing timeline: X" but there's an icon I can click that pops open an overlay with a visual cropped screenshot of that part of the original PDF - maybe even with a red circle around that detail - I can trust those summaries a whole lot more.
Gemini 2.5 has image bounding box and masking features that can help with this (sadly missing from Gemini 3.)
Quick question are you talking about this feature?
https://docs.cloud.google.com/vertex-ai/generative-ai/docs/b...
Because it’s just using structured response so it should be doable with Gemini 3 ? (We are using Gemini 3 for some docs processing and its visual understanding is just incredible)
But the bounding box stuff might work well enough in Gemini 3 to handle this case as well.
Thank you for the feedback.
I've never owned a home and would like to try to buy one in the next year or two. There doesn't seem to be much in the way of API's/software tools that let you analyze historical data and prices of listings in specific areas.
How can I get my hands on the right information to make sure I don't get ripped off?
When sold out vacation home, we had multiple offers, but I seem to recall the offer letters being 1 pagers. Does offer letter length vary by region?
The rest of the document has been a semi-standard contract (used by the real estate agent associations).
2026: "This app could have been a prompt"
There's a question mark here, but this is not a question, it's an uncharitable statement. You know what happens when you assume.
Try something like "if you have no paying customers, how do you have reviews?" Even asking the question helps you think about the perceived conundrum, because your next thought might be "are paying customers necessary for reviews?" (no), which would probably lead you to solve your mystery yourself.