Privacy-first identity verification for US businesses
ShieldID checks the driver's license or passport, the live face and that a real person is present, and shows you the evidence behind every verdict.
Stage 1 of 6 · Document
Read edge to edge, not photographed. Front and back, read on the applicant's phone. Barcode and MRZ checked.
Reading the government photo ID.
Seven minutes
Avery opens a link on their own phone: the front and back of the ID, a live face, a short guided turn. Then the business sees what ShieldID found, twelve verdicts each with its evidence, and makes the call. No slides, and no step skipped.
Press play to see a verification run, from the link to the record.
Every business that meets its customers at a distance has the same problem. A tenant application. A new borrower. A seller signing up. You need to know who they are.
And you may never meet them. So you ask for a photo of the ID. Here is one. The licence is real. It just isn't hers. So you ask for a live selfie. Here is one. It is a video, played to the camera.
And this one was never a person at all. A generated face, built to pass a photo check. And the document itself can be edited: a photo swapped, a date of birth changed, a barcode that no longer says what the print says.
Most checks cannot tell. So businesses do one of two things. They trust too easily. Or they keep a copy of every ID and every face they were ever sent, which is a risk of its own.
Trust too easily, or store too much. Is Avery really Avery? Someone on the other end of this link needs to know. Avery opens it on their own phone. At the bottom: verified by ShieldID.
No account, no app to install. The front of the licence, then the back. Now Avery's face, and a short guided challenge: turn slowly to the left, and back. Done. Avery is verified, and the screen says the ID photos are deleted after seven days, by default.
Simple for Avery. Harder for the business, which may never meet Avery. That is the gap. ShieldID closes that gap. Privacy-first, built in Canada. Identity verification that shows its work.
Most providers hand you one score, and keep the images for years. ShieldID hands you twelve verdicts, each with its evidence, and deletes the images after seven days by default.
Most keep your applicants' data outside Canada. ShieldID processes and stores ID data in Canada. Most are priced for millions of checks a year. ShieldID is sized for teams that need a few dozen a month.
So what just happened? Six stages. One record. One product doing the work of many. Capture: a link or a QR code, both sides of the ID, and the face. ShieldID extracts what Avery's licence says, front and back,
barcode included, then compares it all, field by field. Sources that disagree are flagged, never averaged. Next, a live check that a real person is present right now,
not a photo, a screen or a mask. The industry calls it liveness. Turning is not enough: the head has to come back to centre, which a flat photo cannot do. Then face match: is this the same person as on the document?
One face, one portrait, never a search through other faces. Pass or fail, against a decision line the company sets. Three more checks look at the selfie: for a printed photo or a screen, a synthetic face, and a mask.
The document gets a tamper signal, advisory and not yet calibrated. Each signal is reported on its own, and no plan or setting turns one off. When checks fail, the rows say so.
Here the advisory tamper signal is flagged, and the face and the name do not match. You make the call. Last, the record: twelve per-field verdicts, written to the business account and delivered by a signed webhook.
The images are purged on a timer. The result stays. And Avery is back where they started. Now look underneath. Document capture frames and crops the ID on the phone.
A link, a QR code or a message starts the session. Crop and field models find the document and each printed field. A document reading engine reads the printed side.
Barcode reading decodes the back of the licence. Machine readable zone reading checks a passport's check digits. Cross-source checking compares every source against every other.
Liveness asks a real person to blink, turn and nod. Face match compares the selfie to the document portrait. Anti-spoof, deepfake and mask detection check the selfie three ways.
A document tamper signal is reported for review: advisory, not yet calibrated. Watchlist screening, an add-on, checks public sanctions lists. Secure image storage encrypts the images and purges them after seven days by default.
Secrets and encryption protect personal data and sign every webhook. A model inference runtime runs the detection and matching models. Sign-in and API keys control access to the service.
Hosting and TLS keep the service in Canada and every connection encrypted. API docs, in English and French, show developers how to call it. ShieldID is built from seventy-three components wired into the live service.
They come from forty-eight outside projects, vendors and data services. Grouped, that is eighteen kinds of capability a business would otherwise have to source separately.
Identity verification takes many components from many different sources. And ShieldID delivers them as one product and one integration, privacy first and hosted in Canada.
On the other end of the link: you. Every field read from the ID, with its source, and a verdict for each check. ShieldID reports what it found. You make the call.
Everything the decision needs. Nothing it doesn't. Avery's ID photos, selfie included, are deleted automatically after seven days by default. The record of the check stays.
Hosted in Canada: Canadian data, hosted in Canada. Designed to align with PIPEDA and provincial privacy law. Built for the businesses the big vendors ignore. Avery could be a tenant, a client, a condo owner, a borrower, a seller or a signer.
And ShieldID already runs inside Crown Stack's own products: RentKey, Crown Ledger and Trust Ledger. For the team building it: one REST API. Create a session. Send the link, or a QR code.
Receive a signed webhook, and read the result. Sandbox sessions are never billed. Go live in an afternoon. Start free in sandbox, or book a demo. Avery's part was simple.
The call is yours. ShieldID shows its work. shieldid.ca. Book your twenty minute demo today.
What each side sees
The applicant finishes on their own phone, under your brand. You get every field that was read, where it came from, and a verdict for each check.
How it works
Scroll through a verification. The phone shows what the applicant sees at each stage; the text says what ShieldID does behind it.
The applicant opens a link or scans a QR code and photographs the front and back of a government ID, then their face.
The printed side, the PDF417 barcode and the passport MRZ are read and checksummed, then compared field by field. Sources that disagree are flagged, never averaged.
A short guided challenge on the phone, tracked on 478 facial landmarks over at least ten frames, proves a real person is present rather than a photo or a screen.
The live face is matched to the document portrait by a model built for phone photos, and reported against a fixed decision line rather than a bare score.
Anti-spoof, deepfake, mask and document-tamper checks run on every verification and are reported on their own. You see verdicts and evidence, then you make the call.
Twelve per-field verdicts are written to your account and delivered by a signed webhook. The images are purged on a timer; the result stays. Ledger entries for every verification are coming to ShieldID accounts.
What you get
One platform for reading documents, matching faces, proving liveness and keeping the record, with the API and the portal to run it.
Activity Ledger
Entries can be added, never edited or deleted, not even by us. When someone asks who saw an applicant's ID, you answer with the record. Select an entry to see what it holds.
16 September 2026
req_b2d0e51f7aThis entry can’t be edited or deleted. The network and device detail behind it needs a written reason, and opening it writes an entry of its own.
The sample above is fictional. The ledger records sign-ins, document views and downloads, API-key activity and exports for Trust Ledger accounts today. Ledger entries for ShieldID API accounts, image-view events, decision events and purge receipts are coming soon.
Who it's for
Professional verification, priced and sized for teams that need it a few dozen times a month, not a few million.
Verify the applicant before the keys change hands.
Learn how : Rental & property managementClient identification you can stand behind in a file review.
Learn how : Legal & law firmsOwner and resident verification for boards and managers.
Learn how : HOAs & condo associationsCatch borrowed and synthetic identities without a manual review queue.
Learn how : Fintech & lendingTrusted sellers and workers without slowing sign-up.
Learn how : Marketplaces & gig platformsA signature is only as good as the identity behind it.
Learn how : E-signatureDevelopers
One REST API. Create a session, send the applicant a link or a QR code, and receive a signed webhook when the result is in. Sandbox sessions are never billed.
# Create a verification session (sandbox keys are never billed)
curl -X POST https://api.shieldid.ca/v1/sessions \
-H "X-Api-Key: vk_sand_your_key" \
-H "Content-Type: application/json" \
-d '{
"return_url": "https://yourapp.example/verified",
"webhook_url": "https://yourapp.example/hooks/shieldid"
}'
# → { "session_id": "…", "verification_url": "https://…/v/…", "environment": "sandbox" }How we compare
Categories, not names. The enterprise column describes what we found across thirty vendor sites in September 2026; your mileage will vary by vendor.
| Capability | ShieldID | Typical enterprise IDV | Manual checks |
|---|---|---|---|
| US and Canadian IDs read front and back, barcode and MRZ cross-checked | Yes | Usually | Eyeballed |
| Face match with an active liveness challenge | Yes | Usually | No |
| Anti-spoof, deepfake and mask checks on every verification | Yes | Often tiered | No |
| Per-field verdicts instead of one score | Yes | Rarely | n/a |
| Append-only Activity Ledger you can search | Coming soon for API accounts | Not offered to the customer | A filing cabinet |
| Runs on the vendor's own servers, not a public cloud | Yes | Rarely stated | n/a |
| ID images purged automatically on a short timer | Yes | Rarely stated | Never, in practice |
| Sandbox that is never billed | Yes | Sometimes | n/a |
| Pricing published, in US dollars | Coming soon | 18 of 29 publish none | Staff time |
| Minimum monthly commitment | None planned | Common, $49 to $500 | n/a |
Configured values read from the code. No accuracy percentages and no customer counts until they are measured and real.
Questions
The questions buyers ask before they trust us with an applicant's ID.
Get started
No sales call to see the product. Create a sandbox session, run a verification on your own phone, and read the result verdict by verdict.
Sandbox sessions are never billed; sandbox keys are issued on qualifying plans. Published pricing is coming soon.