Agents
AI personas that do the work
Chat, quote, follow up, research. Each agent’s identity and rules are markdown you edit in seconds.
Agentic AI for insurance
Run your agency on a bench of AI agents.
Ambrose is the operating system where insurance agencies build AI agents and teams — to quote, follow up, research, and automate the busywork. Their behavior is plain markdown. Their data is yours. HIPAA-safe by default.
Runs on Claude via Amazon Bedrock or your own key · No credit card to start
Built for the way agencies actually work
MedicareACA / HealthLife & Final ExpenseICHRAP&C
What’s inside
One platform, from persona to production.
Everything an agency needs to build, connect, and run AI — without stitching tools together.
Teams
Orchestrate many agents
Route work across specialists; publish a team as one endpoint with its own webhook and tools.
War Room
Your executive bench
Ask a roster of exec personas — strategy, marketing, compliance — and get answers grounded in real data.
The Brain
33 federal + health data sources
Live CMS, ACA, Medicare, Census, CDC and more — plus an in-house warehouse of ~17M rows of CMS data.
Skills
Reusable expertise
Blog writers, website builders, quoting engines, dashboards, re-shop autopilot — attach one and an agent is an expert on day one.
Routines
Set it and it runs
Schedule any agent or team on a recurring run — renewals sweeps, daily digests, re-shop and follow-up campaigns — hands off.
Integrations
Meet your stack
Add your own MCP servers, connect up to 10 GoHighLevel accounts, plus Slack, Discord, Twilio, HealthSherpa, VAPI, Retell, and any REST API.
Call intelligence
Analyze every call
Record calls to your vault, auto-transcribe them, and let an agent summarize, extract next steps, and score the conversation.
Websites
Launch sites fast
Start from the Website Template Library, then let skills build compliant, search-ready agency sites and funnels.
Conversations
Every run, in one place
Chat, routine, and webhook threads — continue them, and answer an agent when it needs your call.
How it works
Live in an afternoon.
- 01
Create an agent or team
Describe who it is and what it does. Edit its identity, personality, and rules as plain markdown — or let the built-in assistant build it for you.
- 02
Give it tools & data
Attach skills, connect your CRM and calendars, and switch on The Brain for live CMS/ACA/Medicare data. Everything scoped to your agency.
- 03
Put it to work
Chat with it, publish it as a webhook or Slack bot, or schedule it as a routine. It asks for your approval before anything sensitive.
Security & compliance
The whole platform is HIPAA-safe.
Not a setting you enable — the foundation everything runs on. Protected health information passes through a PHI Rail that redacts and pseudonymizes before anything reaches a non-BAA destination, and each agency is isolated so its keys, data, calls, and conversations never cross the tenant boundary.
- PHI Rail: layered redaction before non-BAA egress
- Per-agency isolation enforced at every layer
- Bedrock or your own Anthropic key — you hold the keys
- Full audit trail of AI usage and staged actions
PHI Railredact → pseudonymize → route
Isolationagency_id on every read/write
KeysBedrock · or your Anthropic key
Auditevery action logged & approvable
Questions
Answers, up front.
What is Ambrose?
Ambrose is an agentic AI operating system built for insurance agencies. You create AI agents and teams, give them tools and your book of business, and put them to work — answering questions, running outreach, quoting, and automating the busywork. Their behavior is defined in plain markdown and updates instantly.
Is it HIPAA-safe?
Yes — the entire platform is HIPAA-safe, not just one feature. Every agent, team, routine, integration, and call transcript runs on the same rails: protected health information passes through a PHI Rail that redacts and pseudonymizes before anything reaches a non-BAA destination, and each agency’s data stays isolated to that agency.
Which AI models does it use?
Ambrose runs on Claude via Amazon Bedrock by default, or your own Anthropic key. You are billed for what you use, with clear per-message pricing.
Do I need to write code?
No. An agent’s identity, personality, and rules are markdown files you edit in the browser — change a file and behavior updates instantly. The in-app assistant can even build and configure agents for you.
How do agents connect to my other tools?
Add your own MCP servers, connect up to 10 GoHighLevel accounts, and wire in Slack, Discord, Twilio, HealthSherpa, VAPI, Retell, or any REST API. You can also publish any agent or team as an MCP server, webhook, OpenAI-compatible endpoint, or Slack bot.
Can Ambrose analyze my calls?
Yes. Record calls to your agency vault and Ambrose transcribes them automatically; then point an agent at the transcript to summarize the call, pull out next steps and objections, score it, and write notes back to your CRM.
What can the Brain do?
The Brain is one place your agents reach 33 live data sources and ~198 tools — CMS, the ACA marketplace, Medicare, Census, CDC, FDA, FEMA, SAM.gov and more — plus an in-house warehouse of ~17M rows of CMS data. It powers cited blog posts, real quotes, and market research, and looks anything missing up online rather than guessing.
Can my whole team use it?
Yes. Ambrose is multi-tenant with per-user roles, shared skills, conversations, and a credit ledger — everything scoped to your agency.
Put your agency on autopilot.
Create your first agent in minutes. No card required.