Financial Planning Report

SGA Platform Cost Comparison

Three build models compared: Traditional 20-person team, AI-First team (5-6 people), and Super Architects + AI Workforce (2-3 people). For the full architecture and roadmap, see the Sprint Portfolio Consolidation.

Baseline
Traditional Team (20 engineers)
$7.0M
2-Year Total Cost
$4.3M
Year 1
$2.7M
Year 2
20+
People
$27K
Per Practice
Model A
AI-First Team (5-6 people)
$2.3M
2-Year Total Cost
$1.5M
Year 1
$900K
Year 2
5-6
People
$8.7K
Per Practice
Model B — Recommended
Super Architects + AI Workforce (2-3 people)
$1.7M
2-Year Total Cost
$1.05M
Year 1
$640K
Year 2
2-3
People
$6.5K
Per Practice
Model B saves $5.3M over 2 years vs. traditional build
75%
Cost Reduction
90%
Fewer People
24/7
Agent Ops
Cost Models
Total Cost by Model
Year 1 (build + 3mo ops) and Year 2 (full operations) stacked

2-Year Total Cost

All-in cost including team, vendors, infrastructure, one-time

Cost Per Practice (2-Year)

Total investment divided by 260 practices
Cash Flow
Monthly Burn Rate Over Time
How cash flows month-by-month across the 12-month Year 1 period

Monthly Spend — Model A vs Model B vs Traditional

Includes team costs, vendor/SaaS, AI tooling, and infrastructure
Cost Composition
Where the Money Goes
Cost composition by category for each model

Model A — Cost Breakdown

AI-First Team (Year 1)

Model B — Cost Breakdown

Super Architects + AI Workforce (Year 1)
Structural Shift
Human Cost vs. AI + Vendor Cost
The structural shift from labor-heavy to infrastructure-heavy

Cost Structure Comparison

Human labor as a percentage of total cost drops from 75% (traditional) to 35% (Model B)
Operations
Ongoing Operations Cost
Monthly steady-state cost once the platform is live (Month 10+)

Monthly Ops Cost Comparison

What it costs to keep the lights on each month

Ops Model — Agent Coverage

Model B: 90% of incidents handled by AI agents autonomously
Vendor Analysis
Top Vendor Costs at Scale
Monthly recurring costs at 260 practices (same across all models)
Vendor Monthly (Scale) Year 1 Total % of Vendor Spend Risk
AWS Infrastructure$9,340$98,00023%Medium
NexHealth$7,700$85,50020%High
Twilio HIPAA$7,360$73,40017%Medium
Swell + Cinnamon$5,200$45,40011%High
Anthropic Claude API$3,750$31,2007%Medium
Dental Intel$2,000$18,0004%High
Datadog HIPAA$1,800$17,2004%Medium
All Other Vendors$6,700$59,30014%Low
Side-by-Side
Full Model Comparison
Side-by-side across all dimensions
Dimension Traditional Model A Model B
Year 1 Cost$4.3M$1.5M$1.05M
Year 2 Cost$2.7M$900K$640K
2-Year Total$7.0M$2.3M$1.7M
Human Team (Build)20+ people5-6 people2-3 architects
Human Team (Ops)7 FTE3.5 FTE1 retainer + SGA IT
Peak Monthly Burn$463K$160K$129K
Steady-State Ops/Mo$213K$88K$50K
Cost Per Practice$27,000$8,750$6,500
Ops CoverageBusiness hrs + on-callBusiness hrs + on-call24/7 AI agent ops
SGA Internal TrainingNot includedNot includedBuilt into engagement
Key Person RiskLowModerateHigher (mitigated)
Knowledge RetentionIn people's headsIn people + docsIn agent skill library
Tech Stack Integration
Leveraging SGA's Existing Systems
SGA already operates 17 production systems across 260 practices. Rather than replace everything, we evaluated each tool against our architecture to find where existing systems accelerate the launch — and where building native is critical for the patient experience. The architecture and roadmap remain unchanged; these are adapter-level integration decisions.
Bridge Tool = Existing vendor system we use temporarily at launch
Native = Our custom-built system designed for the long term
Integrate = Keep permanently — we own the data layer, they render it
Patient-Facing Touchpoints
Build native. Our UX, our voice, our intelligence. Existing tools are transitional bridges with hard migration dates.
Internal / HQ Tools
Integrate where adoption is high. Power BI at 96% adoption means zero training friction for executives.
Data Sources
Integrate whatever gives the best signal. Our data layer means we can swap a vendor later without disrupting anything.
Launch Acceleration
6-9 wks
Faster by Using Bridges
Vendor Cost Reduction
-$11K
Monthly at Scale
Practices Protected
200+
No Retraining at Launch
Architecture Changes
Zero
Structural Modifications
Current Footprint
SGA's Current Technology Footprint
Adoption rates across 260 practices — high adoption tools represent integration opportunities, not replacement targets

Current Tool Adoption by Practice Penetration

Color indicates our integration strategy: green = integrate, blue = bridge then migrate, gray = no overlap
Integration Strategy
Integration Strategy by Tool
How each existing tool maps to our architecture — and what happens to it over time
Existing Tool Adoption Our Component Strategy Rationale
Microsoft Power BI 96% Executive Dashboard Integrate Leadership already lives in Power BI. ClickHouse feeds it — we own the data, they keep their tool.
Modento 79% Communications Service Bridge → Migrate Use for confirmations at launch to avoid retraining 205 practices. Build native. Migrate by Tier 2.
TrueLark 61% Concierge Agent Bridge → Replace Our Concierge deploys to non-TrueLark practices first. Replace at contract renewal — can't have two voices.
Patient Prism 85% H-Score Phone Component Evaluate vs. Cinnamon Backend data source — choose on data quality, not adoption convenience. 85% coverage is a bonus, not the reason.
Peer Logic 78% Phone System Integration Integrate (Priority 1) Our architecture listed RingCentral/8x8 — but SGA actually uses Peer Logic. Build this connector first.
Dental Intel 95% Morning Huddle OS Integrate Already planned. 95% adoption confirms it's the right call.
CallRail / Invoca TBD Funnel Data Lake Integrate Call attribution data feeds our funnel analytics. Connector in Tier 2.
OursPrivacy / Freshpaint TBD Compliance & Audit Layer Integrate Purpose-built PHI detection beats custom regex. Reduces HIPAA audit risk.
Dentrix / Eaglesoft / OpenDental 96% PMS Integration (NexHealth) Already Planned NexHealth wraps all three PMS systems. No change needed.
GoHighLevel / HubSpot / Salesforce TBD Automation Engine Sprint 0 Discovery If SGA has deployed a CRM, integrate it rather than build standalone lead tracking.
Migration Path
Patient Experience Migration Timeline
Patient-facing tools transition from "bridge" to "native" — ensuring UX quality improves, never degrades

Bridge → Native Migration Path

Existing tools (Modento, TrueLark) serve as launch bridges while we build and validate the native experience. Migration happens when our tool is proven better.
Strategic Rationale
Why "Bridge Then Build" — Not "Integrate Forever"
The UX case for building native patient-facing systems, even when existing tools technically work

⚠ Brand Voice Fragmentation

If 160 practices use TrueLark (generic chatbot) and 100 use our Concierge (brand-aware AI), patients get two completely different experiences. For a DSO building a unified brand, that's a problem. TrueLark must be a bridge, not permanent.

⚠ UX Ceiling Risk

Modento's confirmation sequences can't access a patient's funnel stage, H-Score, or content history. Our native Communications Service can. Delegating to Modento permanently means the patient experience hits a ceiling we don't control.

✓ Deployment Risk Reduction

The bridge strategy lets us launch without retraining 200+ practices on Day 1. Practices keep their current confirmation and chat workflows while we validate our native tools on a smaller cohort. Proven beats promised.

✓ Data Advantage Retained

Even during the bridge phase, all Modento/TrueLark events flow into our Event Bus. We capture the data for funnel analytics and AI training from Day 1 — the migration to native tools gets smarter over time.

✓ Power BI — The Exception

Unlike patient-facing tools, Power BI is an internal analytics tool for HQ executives. We own the data layer (ClickHouse), Power BI is just the rendering choice. If SGA ever wants to switch, the data model doesn't change. This is permanent integration done right.

✓ Contract-Aligned Migration

TrueLark has an enterprise agreement with a renewal date. Our migration timeline should align with contract expiry — avoiding early termination fees while ensuring we're ready to replace it when the window opens.

Financial Impact
Financial Impact of Stack Integration
Net cost savings from leveraging existing tools during bridge phase + permanent Power BI integration

Monthly Vendor Savings at Scale

Savings from reduced Twilio volume, Cinnamon elimination (if Patient Prism chosen), and ClickHouse downsizing

Build Time Reallocation

Weeks saved by using bridge tools, reallocated to differentiated features
Validation
Sprint 0 Validation Requirements
Three items must be validated before build begins — these are hard blockers
# Validation Owner Status If It Fails
1 Power BI DirectQuery → ClickHouse
Test ODBC connector, confirm latency < 5s for executive queries		 Platform Architect Blocker Revert to custom React dashboards (add 3-4 weeks to WS1-B)
2 Modento API Coverage
Confirm webhook availability, confirmation status events, intake form data access		 Integration Architect Blocker Build Twilio-native confirmations from Day 1 (add 2-3 weeks to WS1-A)
3 SGA Power BI Environment
Confirm licensing tier, gateway infrastructure, data refresh capacity for real-time		 Platform Architect Blocker May need Power BI Premium or Fabric — adds $5K/month or forces custom dashboards
4 Patient Prism vs. Cinnamon API Comparison
Compare data granularity, call scoring depth, real-time availability		 Integration Architect Important Default to Cinnamon as originally planned. Patient Prism is upside, not requirement.
5 TrueLark Contract Renewal Date
Identify enterprise agreement expiry to align Concierge migration timeline		 Product Manager Important Concierge deploys to non-TrueLark practices regardless. Migration timeline may shift.
6 Peer Logic API Capabilities
Confirm call records, missed call events, recording access via API		 Integration Architect Nice to Have Fall back to other phone system connectors (RingCentral/8x8/Weave).
3-workstream structure (Practice Platform, Content Engine, Culture OS)
12-layer shared infrastructure design
Event-driven architecture (Kafka)
ClickHouse analytical backend
Custom React for practice-level interfaces
Staffing model and cost recommendations
Milestone sequence and tier structure
36-week build timeline (potentially accelerated)
Bottom line: All integration recommendations are adapter-level additions within existing services. The platform is designed for the future — existing tools accelerate the launch, they don't constrain the architecture.
Action Items
Key Decisions & Next Steps
Items requiring action before budget finalization

⚠ NexHealth Pricing

Largest vendor uncertainty. Enterprise DSO pricing for 260+ practices could range $5K-$15K/month. Get a formal quote immediately — this swings Year 1 by $60-120K.

⚠ Existing SGA Contracts

Swell, Cinnamon, Dental Intel, iSolved are likely already contracted by SGA. Auditing existing contracts could eliminate $30-60K/year in vendor costs.

⚠ Dental Intel API

No public API exists. Integration feasibility and pricing need direct vendor confirmation. May require a partnership agreement.

✓ 2-Week Spike Test

Before committing to Model B, run a 2-week proof-of-concept: one architect + AI agents build 1-2 services to validate velocity assumptions.

✓ Architect Selection

Model B lives or dies on 2-3 exceptional architects. Identify candidates with HIPAA, Kafka, AWS, and AI agent orchestration experience.

⚠ Datadog vs CloudWatch

$15-18K/year decision. Datadog provides better observability but adds cost and requires its own BAA. CloudWatch is included with AWS.