Dograh

Insurance Claims

Claim calls that never leave your servers.

Open-source voice agents for claim status and intake, running inside infrastructure you already control.

Open source
Full source, auditable
Self-hosted
Your infrastructure
No per-minute fees
Infrastructure cost only
Bring your own models
LLM, speech-to-text, speech
70+ languages

01 / WHY DOGRAH

Built for claim data you cannot export

Claim audio stays put

Recordings and transcripts never leave infrastructure you control.

Verification before disclosure

Design the identity check that runs before any claim detail is read aloud.

Runs where you choose

Bare metal or private cloud, inside your existing boundary.

No per-minute fees

Pay for infrastructure, not for every call you make.

02 / VS HOSTED PLATFORMS

What changes when you host it yourself

CapabilityDograhHosted platforms
Claim audio stays on your servers✓✕
Full source code, auditable✓✕
Bring your own model stack✓✕
Transcript access without a vendor✓✕
Per-minute call fees✕✓

03 / HOW IT RUNS

Four decisions, then it is yours

  1. 01

    Deploy inside your boundary

    Self-host the platform on infrastructure you already run.

  2. 02

    Connect the claims system

    Read claim status over your own network rather than a third-party hop.

  3. 03

    Put verification first

    Confirm the caller against the file before anything is disclosed.

  4. 04

    Keep the transcripts

    Review, tag and retain calls under your own retention policy.

04 / RETENTION

A status call produces a recording and a transcript, and both can outlive the claim itself. When the storage is yours, you decide how long each one lives, and you can show that the deletion actually happened rather than asking a vendor to confirm it.

Questions

Can claim call audio stay on our own servers?+

Yes. Dograh is self-hosted, so recordings and transcripts remain inside infrastructure you control.

Does this cover claim status calls and intake?+

Yes. The same platform handles inbound status enquiries and first notice of loss intake.

Do we pay per minute?+

No. You pay for your own infrastructure. There are no per-minute platform fees.

Can we use our own models?+

Yes. Bring your own language, speech-to-text and speech models, or run open models you host.

How does the agent verify a caller?+

You design the check. Confirm details already held in the claim file rather than asking callers to recite anything sensitive.

Can we audit what the agent did?+

Yes. You hold the transcripts and the source code, so the call and the logic are both reviewable.

Run claim calls on your own stack

See how a self-hosted voice agent handles claim status without exporting a single recording.