Skip to content
AskFlorence
Main Navigation ArchitectureFlorence AIAgentsMembersAgent PlatformValidationInfrastructure

Appearance

Sidebar Navigation

Overview

Home

Glossary

System Architecture

Consumer & Agent Flow

Florence AI

Overview

Principles

Runtime

Tool surface

Adding a tool

Tool registry

Knowledge: SBC scenarios & CSR

Voice

Evals & observability

Provider risk & portability

Outage playbook

Roadmap

Build plan

Agents

Overview

Workflows & pain points

Members

Overview

Medicaid coverage gap

Carriers

Overview

Marketplaces

Overview

Agency

Overview

Regulations

Overview

Agent Platform

Overview

Auth Architecture

MongoDB Permissioning

Compliance Model

Data Models

Data Sources

Overview

CMS Marketplace API

CMS dependency map

PUF Data

State Subsidies

SBE Ingestion Playbook

SBE State Watchouts + Decisions

CA Phase C/D Playbook

NY Phase C/D Playbook

Validation

Overview

Methodology

APTC Formula

California 2026

New York 2026

CAPS Formula

Scenario Results

Infrastructure

Account Inventory

AWS Setup Runbook

AWS Organizations

CloudTrail

GuardDuty

Security Hub

Config

CloudFront + WAFv2

Data sources & ingest

Phase 4 DNS

Change Log

Vulnerability Management

MongoDB Setup

Access Control

Data Classification

Documentation Hosting

Post-deploy Smoke

Development

Preflight (local CI mirror)

Testing strategy

Compliance

Overview (auditor entry point)

SOC 2 Control Mapping

HIPAA Control Mapping

CMS EDE Appendix A Mapping

Risk Assessment

Encryption Policy

Data Retention Policy

Privacy Impact Assessment

Consent Capture & Versioning

Incident Response Plan

Access Control Policy

Marketing vs. Portal Analytics

Vendor / Subprocessor Register

Dependency Vulnerability Policy

BAA / Compliance Evidence

Compliance-Automation Integration

Compliance-Automation Vendor Evaluation

Penetration Test Reports

Architecture

Portal entry handoff

Mobile app strategy

Deferred architecture decisions

Session cookie architecture

Share flows

Decisions (ADRs)

Index

0001 — Atlas project isolation

0002 — Append-only audit log

0003 — Narrow-scoped Mongo users

0004 — Cross-cluster Atlas PrivateLink

0005 — Delayed-job architecture

0006 — Mongo user simplification

0007 — Terraform owns ECS task def

0008 — E2E testing strategy

0009 — Self-hosted analytics + observability (superseded)

0010 — PostHog HIPAA Cloud (supersedes 0009)

Runbooks

Security Incident Response

Break-Glass Root Login

Onboard Team Member

Offboard Team Member

Atlas user provisioning

Deploy via Terraform (ENG-277)

Rollback via Terraform (ENG-277)

S3 data bucket migration (planned Phase 11)

Access Reviews

2026-Q2 Review

Session log

Index

2026-04-23 — Phase 10 DNS cutover

2026-04-22 — Phase 8 prod AWS mirror

2026-04-22 — Phase 7 Atlas VPC peering

2026-04-22 — Phase 6 CloudFront + WAF

2026-04-21 — Phase 5 staging go-live

2026-04-17 — Atlas staging

Briefs

Index

Member portal plan (ENG-187)

2026-04-16/17 handoff

2026-04-17 Atlas handoff

System briefing (2026-04-17)

Creative AdBundance proposal brief

Creative AdBundance analytics brief

ElevenLabs RN integration research

Policies

Overview

On this page

Agent Workflows and Pain Points ​

For: Taha + Ian Source: GRO-84 Last updated: 2026-05-11 Status: Working draft (v3). Refines with James's Agent Discovery Form and key agent partner calls.


TL;DR ​

Today an agent touches every member's workflow about 8 times across the lead → enrollment → coverage → renewal arc. With AskFlorence they touch it once — review + submit — and earn ongoing PMPM passively after that. The five pain points below are confirmed by James (call 2026-05-05) plus public coverage of the 2024 CMS crackdown and HealthSherpa's product moves. Everything beyond those five stays speculative pending the Agent Discovery Form.

The diagrams below are the punch: sequence diagrams show who's doing the work at each step. In today's row, the agent appears at nearly every interaction. In AskFlorence's row, the agent appears exactly once.


Pain points — confirmed ​

#Pain pointConfirmed byWhy it hurts
1Doctor + drug data doesn't persist across interactionsJames (2026-05-05); HealthSherpa shipped filter persistence in 2024 in response to agent demandAgents re-collect the same info every cycle. Re-typing into HealthSherpa + cross-checking carrier portals is a per-member tax
2Renewals + plan-elimination outreach is largely manualJames (2026-05-05)Disproportionate time sink concentrated in OE. Agent manually cross-references which plans got eliminated → builds outreach list → makes ~hundreds of calls
3Form re-entry for life changes + SEP updatesJames (2026-05-05)Member moves, has a baby, changes job — agent re-types overlapping info into the same forms. No portable state
4No member self-serviceHealthSherpa help center explicitly states clients can't self-service through the platformEvery routine question (deductible reset, find a provider, ID card replacement) routes through the agent. Volume eats the agent's bandwidth
5Agent-of-record switching + three-way call friction2024 CMS crackdown coverage + James (2026-05-05)Unauthorized AoR switches happen; CMS-mandated three-way calls add friction even on legitimate switches. Trust-damaging on both sides

Today's workflow — agent as the load-bearing actor ​

Touchpoint count: 8 agent interactions spread across the year. Carrier portals are the worst per-member tax — 4+ different UIs with inconsistent data quality.


With AskFlorence — agent appears once, earns passively ​

Touchpoint count: 1 agent interaction. AI does the data work. The member portal does the questions + life-change work. AskFlorence does the renewals work. The agent's economics shift from billable-hours-per-member to passive PMPM.


Touchpoint reduction — side-by-side ​

StageToday (agent touches)AskFlorence (agent touches)
Lead arrives——
Info collection✓ Agent (phone)— (AI handles intake)
Data entry to platform✓ Agent (HealthSherpa)— (AI handles)
Doctor + drug cross-check✓ Agent (4+ carrier portals)— (AI handles via CMS API + cross-cluster reference)
Plan walk-through✓ Agent + member— (AI-guided selection)
Application submit✓ Agent✓ Agent review + submit
Coverage card waiting——
Routine questions during year✓ Agent— (member portal)
Life events + SEPs✓ Agent— (member portal)
OE plan-elimination outreach✓ Agent (manual list build)— (automated)
Renewal call✓ Agent— (auto-renewal flow)
Total touchpoints~81

The agent's role compresses from "do all the work" to "be the licensed signatory for the moment of submission." Everything else is automated or member-self-served.


Visual summary ​

Legend:

ColorMeaning
🟨 SandAgent does the work
🟦 BlueMember does the work
🟩 GreenAI / AskFlorence does the work
🟥 RedWaiting (no one's working)
🟢 Dark greenAgent earns passively, zero work

The punch: 8 sand boxes today vs 1 with AskFlorence. The agent's only touch is review + submit. Everything else either moves to AI (intake, plan prep), member self-service (questions, life events), or AskFlorence (renewals).


Gaps for James + key agent partners to fill ​

Categories, not questions. The Agent Discovery Form pulls the specifics.

  • Real time spent per stage above (the workflow times are placeholders)
  • Where the actual bottleneck sits (we suspect lookup + renewals, need a ranking)
  • What CRM + admin tools sit alongside HealthSherpa day-to-day
  • How agents handle SEP and life-change updates today (what triggers the re-entry — member call? carrier alert? CMS notice?)
  • Which carriers have the worst data quality on doctor + drug coverage
  • Real-world retention numbers per agent book of business
  • Pain points we haven't named here

Cross-references ​

  • Agent Platform overview — auth, MongoDB permissioning, compliance model
  • Consumer & Agent Flow — full end-to-end member journey from quote → enrollment
  • System Architecture — where the agent portal sits in the stack
  • Linear: GRO-84 — origin spec for this doc
Pager
Previous pageOverview
Next pageOverview

AskFlorence Internal Documentation. Not for public distribution.

AskFlorence

Internal Documentation

Access restricted. Not for public distribution.