One record supported by an entire ecosystem
Waypoint’s AI runs on Claude Opus 5.5. It summarises documents and reads scanned pages, reads intake reports, triages new claims, screens for fraud, reads underwriting submissions, completes forms, writes the long summary of a whole-claim print, and answers questions through Remy and the Ask Claude buttons.
Claim data reaches the AI only when four locks are all open: Waypoint’s platform lock, your organization’s signed AI terms recorded by Waypoint, the AI entitlement, and your organization’s own AI switch, which starts off. Every AI decision that drives an action is logged.
The in-house assistant: the expert in the corner
Every department has that one person who knows the system inside and out. Remy answers the “how do I…” questions about the portal a person is standing on — where a feature lives, what a field means, how a workflow runs — with numbered steps for a process.
Each Remy is scoped to its own portal, so the answer an adjuster gets is about the portal they are actually in — not a general manual that hedges across all eight.
A loss that arrives as an email, ready to work
The reviewer sees the extracted fields, the validation gaps, your fraud-rule results with their evidence and any duplicate hold. A draft is never a claim until a person promotes it — see how that lands in the Waypoint Center’s intake doors.
Trained on your book, not the industry average
Machine-learning scores are trained on your organization’s own claims, and only if your organization has said its legacy book may train its models. Every model is born illustrative and climbs one rung at a time — candidate, active, retired — and only active models serve scores.
Promotion to active takes a person’s reason and the typed model version. A score never writes to any claim field, and nothing retrains by itself.
A referral, never a determination
Fraud rules at first notice, a points-based question bank, cross-book party searches and an AI second opinion raise indicators for a person. Flags never hold a legitimate claim, and every final decision is a person’s.
Adjusters retain absolute ownership over strategy, negotiation and settlement approvals. A claim cannot close while a referral is open, which is where the Waypoint Center picks it up.
Spend is a number, not a surprise at renewal
AI use is metered call by call. With AI on, each seat brings a pooled monthly allowance shared across the whole organization, and each organization has an AI spend limit set at signing.
Every AI call writes a receipt with tokens in and out, cost and feature. Waypoint’s cost views label claims anonymously, so cost management is never a reason to read your data.
What the AI is not allowed to do
and see what you get
Every capability is bought, then switched on
Each AI feature passes the single door under its own entitlement, off until it is switched on for your organization — and nothing reaches the AI until all four locks are open.
Bring us a messy loss report
The best way to evaluate our platform is to test it against your most complex operational bottlenecks. Send a disorganized loss report — a forwarded email chain, a scanned page, a voicemail. We will run it through intake in front of you and show what is drafted, what is flagged, and what a person still has to decide.