# PhotoGraph

> PhotoGraph is visual entity resolution software for photo collections, made by Photo-Graph, LLC of Phoenix, Arizona. Starting from one photo, it finds every appearance of that person across an entire unlabeled collection and across a lifetime, and stores the result as a queryable identity graph. It needs no labeling or model training, works from pixels alone, and runs entirely inside the customer's own environment.

## Key facts

- Product: PhotoGraph (written as one word, capital P and G). Company: Photo-Graph, LLC, Phoenix, Arizona, USA. Website: https://www.photo-graph.llc/. Contact: info@photo-graph.llc.
- Category: visual entity resolution. Entity resolution decides which records refer to the same real-world entity; visual entity resolution does this with images, deciding which photos show the same person and linking them into one identity, even when no names, captions or metadata connect them.
- Core technology: the Graph Resolution Core (GRC), covered by a pending US patent application and exclusively licensed to Photo-Graph, LLC by Graph Ventures, LLC of Arizona.
- How it differs from facial recognition: facial recognition compares two faces and returns a match score, and needs a clear, front-facing, well-lit face. PhotoGraph resolves a person across a whole collection, comparing every image with every other image, including photos where the face is partly hidden, shadowed, low resolution, out of focus, turned away or not visible at all, and across decades of aging. It keeps look-alike relatives separate.
- Benchmark: in a real 33,000-photo family collection where one known person appears in 524 photos (ages 2 to 76, film scans to digital, alone and in groups), PhotoGraph found all 524. A bleeding-edge algorithm found 240 and a widely deployed model found 81.
- Deployment: on-premises, on the customer's own hardware. Images, results, the identity graph and metadata stay in the customer's environment. The only information transmitted back to Photo-Graph, LLC is usage statistics, which is verified with the customer before go-live. There is no cloud service in the processing path.
- Output: a queryable identity graph of people, the photos they appear in, who they appear with, and when; a graph query API; a search and browse app; and optional plain-English questions using graph-augmented generation, with a language model that can also run on customer hardware.
- Complements text-based entity resolution: because it uses pixels only, it adds independent evidence to systems that match on names, IDs, addresses and dates. This is especially useful in fraud, where identities use different names and records.
- Scale: architecture designed for billion+ edge identity graphs.
- Inputs: scanned prints, slides, negatives, contact sheets, document scans and born-digital photos.
- Markets: genealogy platforms; public institutions and historical societies; fraud, security and investigations; media archives and stock photo companies; research, journalism and legal discovery.
- Customer: a historical society runs PhotoGraph across its entire institutional photographic archive, on its own hardware, linking people across separate donor collections.
- Licensing: Pilot (fixed scope, fixed fee), Institution (annual license, fixed terms), Enterprise (annual license, custom scope), and Platform partner (OEM licensing). Price depends on archive size and monthly usage.
- Company: founded by three co-founders with more than 75 combined years in federal identity systems and private industry. Open to growth through partnerships, marketplaces, and investment from venture capital firms and family offices.

## What PhotoGraph is not

- Not a facial recognition API: it resolves identities across a whole collection rather than scoring pairs of faces.
- Not a cloud service: it does not send images or results outside the customer's environment.
- Not trained on customer photos: no model training occurs on a customer's collection, and collections are never used to improve a product for anyone else.
- Not a human-review service: validating results and any human-in-the-loop process is up to the customer.

## Pages

- [Home](https://www.photo-graph.llc/): overview, how it works, use cases, benchmark summary and FAQ
- [Why PhotoGraph](https://www.photo-graph.llc/why.html): comparison with facial recognition and the 524-photo real-world benchmark
- [Solutions](https://www.photo-graph.llc/solutions.html): solution summary, architecture, Graph Resolution Core, text-based entity resolution, identity graph and API, plain-English questions, deployment
- [Use Cases](https://www.photo-graph.llc/use-cases.html): genealogy, public institutions and historical societies, fraud and investigations, media archives, research, journalism and legal discovery
- [Customer story](https://www.photo-graph.llc/customers.html): a historical society's archive-wide deployment
- [Trust & security](https://www.photo-graph.llc/trust.html): data sovereignty, usage-statistics-only reporting, risk mitigation and responsible use
- [Pricing](https://www.photo-graph.llc/pricing.html): editions and licensing questions
- [Investors](https://www.photo-graph.llc/investors.html): partnerships, marketplaces, venture capital and family offices
- [About](https://www.photo-graph.llc/company.html): company, team and patent-pending technology
- [Contact](https://www.photo-graph.llc/contact.html): request a briefing, demo, pilot or pricing