Website Update Brief — daisyai.ai

Date: 2026-02-16 Purpose: Hand this to the agent editing the www repo. Contains all strategic context, specific feedback, and copy direction for updating every page.

---

Strategic Context (Why We're Updating)

DaisyAI has converged on a clear go-to-market over the past 6 weeks through investor calls (7 completed), advisor conversations (Dan Wilson — ex-CEO of Moxe, deep health IT), and landing our first client engagement (Premera Blue Cross — FDE consulting for authorization AI).

The positioning is sharpening from:

"AI Engineering for Healthcare" (broad, could be anyone)

To:

AI-powered authorization and clinical review for health plans, outsourcers, and risk-bearing entities

The delivery model is FDE (Forward Deployed Engineering):

• Embed with clinical ops teams
• Build production AI on their infrastructure
• Expand across their authorization/UM workflow stack

The buyer is narrowing to:

1. Health plans (regional Blues, national payers)
2. UM/UR outsourcing companies
3. Risk-bearing entities / TPAs / healthcare services orgs

The specific workflows we help with:

• Prior authorization (auto-auth expansion)
• Appeals processing (payer-side defense against AI-generated provider appeals)
• Concurrent review automation
• Clinical chart summarization for nurse workflows
• UM flag extraction and case prioritization

---

Current Website Problems (from analysis of daisyai.ai)

1. Hero / headline is too generic

Current: "We embed with your team to design, build, and deploy AI in production." Problem: Could be any AI consulting firm. A UM director at a health plan wouldn't see themselves in this.

2. No authorization/UM language above the fold

The words "authorization," "prior authorization," "utilization management," "clinical review," and "appeals" are not prominently featured. These are the exact terms buyers search for and care about.

3. Target market list is too broad

Current targets listed: Regional health systems, Payers, UM/UR outsourcing companies, Physician groups, Medical billing and workers' comp providers, Behavioral health organizations Problem: 7 categories dilutes the message. Should lead with 3.

4. Problem framing is generic AI adoption, not UM/authorization pain

The 7 challenges (evaluation paralysis, governance gaps, talent shortage, etc.) are real but generic. They apply to any healthcare AI company, not specifically to authorization/clinical review.

5. No specific outcomes or proof points

Current stats are generic industry stats. Missing UM-specific outcomes.

6. "Wrapper on LLM" risk in written materials

An investor (Tau Ventures) explicitly said written materials trigger "wrapper on LLM" fear that the verbal pitch does NOT trigger. The website has this same issue — it reads as "we help you use AI" rather than "we ARE the authorization infrastructure."

---

Positioning Direction — What Every Page Should Convey

Core identity

DaisyAI is the embedded AI engineering partner for health plans and risk-bearing organizations doing authorization, utilization management, and clinical review workflows.

One-liner options (pick the best fit per context)

• "AI-powered authorization for health plans"
• "The AI layer for clinical authorization and review"
• "Embedded AI engineering for utilization management"
• "We help health plans automate authorization, accelerate reviews, and scale nurse capacity — without adding headcount"

Value prop (3 components)

1. Reduce review time — AI-driven chart summarization and evidence extraction cuts manual review from 20+ minutes to under 10
2. Scale authorization capacity — auto-auth expansion, appeals processing, concurrent review automation
3. Deploy in production, not just pilots — embedded in your systems, your data, your workflows. 95% of healthcare AI pilots fail to reach production. We ship.

Who we serve (narrow list)

• Health plans — regional Blues, national payers, Medicaid managed care
• UM/UR outsourcers — companies running clinical review on behalf of payers
• Risk-bearing entities — TPAs, IPAs, healthcare services organizations managing authorization workflows

What we do NOT lead with anymore

• Physician groups (not our current buyer)
• Medical billing (adjacent but not the wedge)
• Workers' comp (niche, distracting)
• Behavioral health (can follow later, not now)
• Generic "AI engineering" language without UM/authorization specificity

---

Page-by-Page Edit Guide

Homepage (/)

Hero section:

• Change headline to something authorization/UM-specific
• Suggested: "AI-Powered Authorization for Health Plans" or "Embedded AI for Clinical Authorization and Review"
• Subhead should name the pain: "Health plans are drowning in manual chart review, authorization backlogs, and AI-generated appeals. We embed with your clinical ops team to automate what matters."
• CTA stays: "Book a Discovery Call"

Problem section (the 7 challenges):

• Keep the structure but rewrite through UM/authorization lens
• Replace generic AI adoption challenges with UM-specific pain:
  1. Nurses spending 60%+ of time on administrative review tasks
  2. Authorization backlogs creating patient harm and premium escalation risk
  3. Providers 1 year ahead using AI to generate appeals — payers need to respond
  4. Auto-auth covering only a fraction of eligible cases
  5. 7-14 day manual processing times for authorization requests
  6. Clinical staff overwhelmed by caseload with fixed headcount (keep this one — it's already good)
  7. Legacy systems (HL7, ADT feeds, multiple EHRs) creating integration complexity (keep — already good)

Target market section:

• Narrow to 3: Health Plans, UM/UR Outsourcers, Risk-Bearing Entities / Healthcare Services
• For each, one line of specificity:
  • Health Plans: "Regional and national payers looking to scale authorization capacity, expand auto-auth, and defend against AI-generated appeals"
  • UM/UR Outsourcers: "Clinical review organizations that need to process more cases with the same headcount"
  • Risk-Bearing Entities: "TPAs, IPAs, and healthcare services organizations managing authorization workflows at scale"

3-phase engagement model (Embed → Build → Scale):

• Keep this structure — it correctly describes FDE
• But rewrite the descriptions with authorization-specific language:
  • Phase 1: "Shadow nurse workflows, map authorization decision paths, understand your criteria systems (InterQual, MCG), identify highest-impact automation opportunities"
  • Phase 2: "Write production code in your environment, integrate with your EHR feeds (HL7, ADT, CCD), deploy AI-assisted review workflows, ship iteratively with clinical team feedback"
  • Phase 3: "Measure outcomes (review time, auto-auth rates, appeals throughput), expand into adjacent workflows, build internal AI fluency, transition from project to partnership"

Stats / social proof section:

• Replace generic stats with UM-specific ones:
  • "50%+ reduction in nurse review time" (from Premera scope targets)
  • "25-50% of manual authorization work eliminated through auto-auth" (from Premera existing results)
  • "95% of healthcare AI pilots never reach production. We ship." (from internal research)
  • Optional: "Providers are 1 year ahead using AI for appeals. Health plans need to catch up."

Platform reference:

• Keep the mention of app.daisyai.ai and HIPAA compliance
• Add specificity: "HIPAA-compliant reasoning layer for medical record analysis, HL7 connectivity, ADT feed integration, and structured clinical extraction"

Services Page (/services)

• Lead with the specific services in authorization/UM language:

  1. Prior Authorization Automation — expand auto-auth coverage from a handful of conditions to hundreds of medical policies
  2. Appeals Processing — AI-assisted defense against AI-generated provider appeals at scale
  3. Clinical Chart Summarization — structured summaries mapped to your review templates, extracting UM-relevant flags
  4. Concurrent Review Automation — real-time monitoring of admissions with ADT feed integration
  5. UM Analytics & Decision Support — throughput dashboards, case prioritization, quality metrics

• For each service, describe:

  • The problem it solves (in buyer language)
  • How we deliver it (FDE model — embedded, production code, their infrastructure)
  • What outcomes to expect

Product Page (/product)

• Position the product as the AI layer underneath the FDE work, not a standalone SaaS tool
• Key capabilities to highlight:
  • Medical record analysis and structured extraction
  • HL7 v2 / ADT / ORU / CCD ingestion and parsing
  • Clinical chart summarization mapped to payer templates
  • UM flag extraction (start of service timing, critical diagnoses, queue priorities)
  • Confidence scoring (auto-approve vs. escalate to human — no automated denials)
  • Full audit trail and traceability (Phoenix-style observability)
  • HIPAA-compliant architecture
• Frame as: "The AI reasoning layer that powers our FDE engagements — built from real clinical operations, not from a lab"

About Page (/about)

• Keep "healthcare operators building AI for healthcare"
• Emphasize the intersection: "We've spent 2+ years embedded in clinical operations, understanding how nurses actually review cases, how authorization decisions get made, and where AI can safely augment — not replace — clinical judgment"
• The "no automated denials" principle should be prominent — this builds trust with payer buyers who are terrified of regulatory risk
• Team section: emphasize the clinical + technical + operational mix

---

Copy Snippets Ready to Use

Headlines (pick per page/section)

• "AI-Powered Authorization for Health Plans"
• "The AI Layer for Clinical Authorization and Review"
• "Embedded AI Engineering for Utilization Management"
• "Scale Authorization Capacity Without Adding Headcount"
• "From Manual Review to AI-Assisted Authorization"

Subheadlines

• "We embed with your clinical ops team to automate authorization, accelerate reviews, and defend against AI-generated appeals."
• "Health plans are drowning in manual chart review. We help them ship AI that actually works in production."
• "Prior auth. Appeals. Concurrent review. We build the AI that powers your nurse workflows."

Pain points (one-liners for bullets or cards)

• "Nurses spending 60%+ of their time on administrative tasks instead of clinical judgment"
• "Authorization backlogs measured in days, not hours"
• "Providers using AI to generate appeals faster than your team can review them"
• "Auto-auth covering 6 conditions when it could cover hundreds"
• "Manual chart review breaking at scale — you can't hire your way out of this"
• "95% of healthcare AI pilots die before production. Yours doesn't have to."
• "ADT feeds updating every 15 minutes but clinical workflows still running on faxes and PDFs"

Outcomes (for proof/stats section)

• "50%+ reduction in nurse review time"
• "25-50% of authorization work automated end-to-end"
• "Review time from 20+ minutes to under 10"
• "Production-grade AI deployed in weeks, not years"
• "Full traceability — every AI output tied to source clinical documentation"

Trust builders

• "No automated denials. Ever. AI approves, humans decide everything else."
• "We work in your environment, with your data, under your security controls."
• "HIPAA-compliant. Evidence-based criteria alignment. Full audit trail."
• "Built by healthcare operators who understand InterQual, MCG, and the reality of UM workflows."

FDE model description (for services or about page)

• "We don't sell software and walk away. We embed with your team — shadowing nurse workflows, mapping authorization decision paths, writing production code in your environment, and shipping AI that your clinical staff actually trusts and uses."

---

What NOT to Change

• The "Book a Discovery Call" CTA — keep it
• The HIPAA compliance messaging — keep it
• The 3-phase engagement model structure (Embed → Build → Scale) — keep, just rewrite descriptions
• The platform reference (app.daisyai.ai) — keep
• The dark/professional visual aesthetic — keep
• Contact: thomas@daisyai.ai — keep
• Social links (Twitter, LinkedIn, Substack) — keep

---

Tone Guide

• Lead with the business problem, not the technology. "Authorization backlogs" not "LLM-powered analysis."
• Be specific to UM/authorization. Every section should pass the test: "Could a generic AI company say this?" If yes, rewrite.
• Credibility through specificity. Name the data formats (HL7, ADT, CCD). Name the criteria systems (InterQual, MCG). Name the workflows (prior auth, appeals, concurrent review). This signals domain depth.
• No "AI wrapper" language. Don't lead with model names or AI capabilities. Lead with what the buyer gets (faster reviews, more auto-auths, appeals processed at scale).
• Humble confidence. "We ship production AI in healthcare" — not "we revolutionize healthcare with AI."

---

Investor Context (in case the about/press page mentions fundraising)

• Raising a seed round ($2M target)
• Revenue model: SaaS fee 5-10% + implementation fee (rate card, positioned as "discount to Palantir") + maintenance/licensing ~50-60% of contract value
• Active pipeline: 12 investors, Sorensen Capital and Refract VC warmest
• Key proof point: Premera Blue Cross engagement (FDE consulting, auto-auth + appeals)
• What resonates with investors: founder grit, payer-first strategy, UM as wedge, FDE delivery model
• What to avoid: "wrapper on LLM" framing, generic "AI for healthcare" positioning

---

Summary of Changes by Priority