Images carry the condition story
Roofs, rooms, finishes, damage, deferred maintenance, and renovation quality are visible signals. The hard part is making them consistent enough to use.
Applied AI for property intelligence
HomeSignal
We are building software that helps people make better sense of property imagery, documents, and workflow context, turning heterogeneous inputs into structured signals people can utilize.
Core
Property imagery, document context, and structured signals
Stage
Core system in active development, shaped by domain feedback
Input
Built with feedback from technical peers and domain experts
Core thesis
Property work still depends on information that is visual, fragmented, and hard to structure.
Mortgage and real estate decisions often rely on photos, inspection notes, appraisals, listing descriptions, county records, and PDFs that were never designed to become software-readable signals. HomeSignal is focused on that translation layer.
Documents carry the workflow context
Property records, loan artifacts, appraisals, inspections, and listing content describe the same asset in different formats and at different levels of precision.
Structured signals unlock better tools
Once property information is extracted, normalized, and connected to evidence, it can support better review, comparison, and decision workflows.
What we're building
A focused technical core before a broad product surface.
The work is intentionally focused on the core first: understanding property images and documents well enough to extract useful features with confidence.
Ingest heterogeneous inputs
Property photos, document snippets, descriptions, and contextual records come in with different quality, formats, and assumptions.
Extract visible features
The system identifies property features, condition clues, and evidence that can be tied back to the source image or document.
Normalize the signal
Extracted details become structured data that can be compared, reviewed, filtered, and improved over time.
Shape useful workflows
The intelligence layer can support lender review, agent workflows, property analysis, and other expert-facing surfaces.
Why it matters
The value is not the AI demo. It is the evidence-backed property signal.
The important idea is that property workflows improve when visual and document-heavy inputs become structured evidence. The technology should be understandable enough to trust and practical enough to fit expert review.
For lenders
Collateral review depends on property condition, comparable context, and exception handling. Those inputs are often buried in photos, PDFs, and notes.
For agents
Listing strategy, buyer confidence, and deal risk all improve when visible property issues are surfaced earlier and tied to clear evidence.
For builders and domain experts
The interesting question is where the intelligence layer belongs, what it should extract, and how experts should review and trust it.
Follow the build
Building in conversation with the people who understand the problem.
- Short notes on what we are learning about property images, documents, and workflow design.
- Questions for people who know lending, appraisal, underwriting, real estate, computer vision, and applied AI.
- Progress updates that stay honest about what is working, what is uncertain, and what we are refining next.
Team
Three experienced engineers building a focused applied AI project.
The credibility belongs on the team page, but the short version is simple: this is a hands-on technical team building the core system and using domain feedback to shape the product around it.
Craig Brown
Co-Founder
Paul Huber
Co-Founder
Samuel Fleming
Co-Founder