Live · Amanda
Meet her

Meet Amanda.
Your AI benefits counselor.

Four hires you no longer need to make. Renewals in ninety seconds. Enrollment in plain English. The 10pm phone call, answered. The renewal letter, queued for your sign-off.

How she helps

Four jobs you don’t have to hire for.

Amanda runs the analyst, the integration team, the OE counselor, and the inbox, all in one pane, on the rails the rest of fintech runs on.

01 · RENEWAL ANALYST

Renewals in 90 seconds, not three weeks.

$0Ka year, off your payroll

Drop in three carrier proposals. Amanda parses each PDF, runs the cohort math against your active enrollment, ranks them, and writes the recommendation letter ready to send to leadership.

02 · EDI / HRIS SPECIALIST

Carrier integrations live in an afternoon.

$0Ka year, off your payroll

Workday, Gusto, Rippling, BambooHR, Paychex on the HRIS side. Aetna, Anthem, Cigna, Lincoln, Unum, Aflac on the carrier side. One API for the first, EDI through Stedi for the second.

03 · OE BENEFITS COUNSELOR

Every employee gets a counselor, 24/7.

$0Ka year, off your payroll

Eight minutes per employee. Zero HR tickets. Plan recommendations grounded in real spending and family situation, not generic education. In every language the workforce speaks.

04 · MEMBER SERVICES

She picks up. She drafts. She logs it.

$0Ka year, off your payroll

Year-round calls and emails. Amanda picks up, drafts replies in your voice, and queues every response for approval. Coverage starts the day of birth, every time.

Try Amanda

Ask her anything. In real time.

Amanda is built on Anthropic’s Claude with strict no-training and zero-retention settings. Your test conversation never trains a third-party model.

Want to feel what an enrollment meeting actually sounds like? Ask her about HSAs, hospital indemnity, or how a PPO compares to an HDHP. She’ll respond in plain English. If you ask about regulated topics, she escalates to a licensed human counselor.

Open the full sandbox
Amanda
Replies in seconds · Online now
Plan choice
Ask Amanda about a plan...
How she’s trained

A proprietary ontology, grounded in real benefits data.

Amanda is not a general-purpose chatbot wearing a benefits costume. She runs on a proprietary data ontology that ingests your plan documents, carrier rate sheets, broker strategy, and HRIS census, and gets sharper with every employer onboarded.

23
Plan types
Medical, dental, vision, life, HSA, FSA, pet, legal, LTD, STD, critical illness, hospital indemnity, and more
38
Carriers
Aetna, Anthem, BCBS, Cigna, Humana, UHC, Lincoln, Unum, MetLife, Aflac, Guardian, Allstate, more
15+
Languages
Native voice in Spanish, Mandarin, Tagalog, Vietnamese, Haitian Creole, Arabic, Portuguese, more
100%
Of plan documents
SBCs, certificates, master contracts, rate sheets, carrier portals, ingested per group, refreshed each renewal
How she’s compliant

Carrier-grade trust, on day one.

The trust posture every benadmin, carrier, and broker compliance team requires before they will integrate. We did the work so you don’t have to redo it.

HIPAA & BAA-ready

Business Associate Agreement available on day one. PHI is encrypted at rest with AES-256 and in transit with TLS 1.2 or higher. Access is least-privilege and centrally audited.

SOC 2 Type II

Independently assessed annually. Reports available under NDA. HITRUST certification on the 2027 roadmap.

No training on your data

Customer data and PHI sent through Amanda never train Anthropic’s general models. Zero-retention enterprise settings are enforced contractually and verified in our subprocessor agreements.

SSO & SAML, broker-strategy guardrails

Single sign-on with Okta, Google, Microsoft Entra. Amanda always reflects the broker’s plan strategy and never recommends against your book. Guardrails are version-controlled per client.

Fully auditable, every interaction

Immutable append-only event log. Every voice call, chat message, plan recommendation, and carrier filing is logged, transcribed, and searchable. Cryptographically signed and SOC 2-aligned.

Human-in-the-loop where required

Renewal letters, EOI submissions, and regulated communications are queued for human approval by default. You decide which workflows are fully autonomous and which require sign-off.

Fully auditable

Every interaction logged, transcribed, searchable.

If you can’t produce a record of what Amanda told an employee in March, you don’t have a compliance program. Keel keeps a complete event log of every Amanda interaction, every plan recommendation, every carrier filing.

  • Immutable append-only event log with cryptographic signatures
  • Full voice and chat transcripts, searchable in plain text
  • Per-recommendation reasoning trace, including the data Amanda used to decide
  • SOC 2-aligned access controls on the log itself
  • Exportable to CSV, JSON, or a compliance vendor of your choice
Read our privacy practices
11:42:01
[INBOUND] Voice call from +1 415 555 0179 routed to Amanda
11:42:02
[IDENT] Caller identified as [email protected] via phone match
11:42:04
[CONTEXT] Loaded thread #t_9F4K2 (12 prior interactions)
11:42:06
[QUERY] “Why is the Silver PPO the best fit for me?”
11:42:07
[REASON] Pulled census: 1 covered life. Pulled prefs: low_variance=true. Computed best-fit PPO > HDHP by $1,820.
11:42:09
[RESPONSE] Amanda delivers plan rationale + HSA tax math. DELIVERED
11:42:31
[INTENT] Detected scenario request: “add dependent”
11:42:33
[ACTION] Queued: model_family_coverage(marcus.hayes, dep=mother)
11:42:34
[POLICY] Broker strategy guardrail check: PASS
11:42:35
[SIGN] Event signed sha256:7b9c2f4a... SEALED
11:42:37
[INBOUND] Caller follow-up: “What about dental?”
Ready to see her work?

Amanda is hiring herself out to the next broker.