FOR INVESTORS

Capital efficiency, operator DNA, paying healthcare pilot

SVTech is a vertical PaaS company with an unusual shape: a twenty-year commercial operator, self-funded at pre-seed, with a paying healthcare pilot, two shared-engine flagships running on Mono-PaaS, two newer expansion proof surfaces, a validated MAIA safety classifier, and a scoped Layer-6 fine-tune roadmap that builds a defensible clinical-analytics moat from accumulating signal.

Where we stand today

May 2026. No outside capital. Paying healthcare pilot. Two shared-engine flagships in active beta, plus BizFrameHub and Colombian Spanish Academy as public expansion proofs.

Stage
Self-funded pre-seed

Founder equity, no outside capital. Entity: SVTech Consulting Services LLC, formed December 14, 2022 - three-plus years of operating history predating the AI pivot.

First revenue
First paying Stage 1 pilot · Sept 2025

First paying Stage 1 pilot from a nurse-practitioner-led psychiatric practice - the same practice grown from zero to four practitioners on organic search before becoming the DWA pilot customer. Stage 2 licensee revenue follows once the engagement model graduates from custom-vertical work to platform-reviewed manifests.

Product footprint
2 flagships + 2 expansion proofs

DWA (HIPAA-aware, PHI) and GTM-OS (non-PHI) both run on the Mono-PaaS engine today - same identity, tenancy, AI orchestration, adapters, manifests. BizFrameHub is launching today as a 57-course AI Business Academy for SMBs and professionals. Colombian Spanish Academy is live as a rapid localized language-learning proof with 20 courses and a Bogotá guide module.

MARKET TAILWIND

The wedge is not the whole education market. It is AI-coached vertical academies that can ship now.

HolonIQ projects education approaches $10T by 2030, and WEF/IMF data points to a real reskilling cycle. SVTech should not frame this as a generic TAM grab. The investor-relevant wedge is narrower and sharper: regulated, professional, and operator communities that need branded AI learning products live in days, not after a custom platform build.

That is where Mono-PaaS matters: your branded academy, live in days, with an AI coach that knows your curriculum, and a safety layer you can dial up to clinical grade.

THE MOAT STACK

Five defenses that compound

Each layer is independently valuable; together they're the reason a competitor can't replicate this in a quarter.

1
Safety layer

MAIA - continuously-retrained distress classifier

Sentinet/suicidality (ELECTRA-base, ~110M params, CC0 baseline) running as a CPU sidecar at 22.3 ms/sample. Validated F1 0.93 / crisis recall 1.0. Classifier stays text-free; retrain pipeline promotes a successor only when it measurably beats the baseline. Compliance and clinical trust enforced at the engine level via failSafe="closed" routing.

2
Content layer

Evidence-based lesson library

922 lessons across 43 DWA courses + 974 localized English-Spanish lesson files across 49 GTM-OS courses, plus BizFrameHub's 57-course AI Business Academy and Colombian Spanish Academy's 20-course localized curriculum. Each DWA therapeutic lesson takes 15-20 hours to author at clinical quality. The broader portfolio proves both depth and repeatable packaging.

3
Intelligence layer (GPU-bound)

Adaptive learning · risk stratification · content-outcome correlation

Each requires longitudinal data that only emerges from real users over time. The AI layer page has the full roadmap - ClinicalBERT, RoBERTa, BKT, IRT, XGBoost, pgvector.

4
Distribution layer

Practice licensing - aligned incentives

Licensed partners do the marketing to their own patient base; revenue share aligns incentives. Custom vertical → licensed deployment → (eventually) Studio self-serve. See the engagement model.

5
Research layer

Publishable clinical outcomes

Longitudinal data produces publishable research directly. Academic credibility compounds into practice trust; practice trust compounds into distribution.

Architecture discipline as a capital-efficiency signal

One of the fastest ways to read a technical team is what they've chosen not to build. Every item on the SVTech kill list - no microservices, no Kubernetes, no custom AI gateway, no in-house vector DB, no custom auth, no multi-region by default - reduces surface area, operational cost, and hiring burden.

16 ADRs accepted, 0 rejected

Every non-obvious decision is documented with context, alternatives, and consequences. docs/adrs/ is a read-the-repo-to-understand-the-company artifact.

Admission rule on new engines

No package lands in the shared platform without ≥2 vertical consumers or a third committed use case. Prevents premature abstractions; keeps core surface area honest.

RLS regression gate · ships with v2

v1 enforces tenant scoping in application code with audited query paths. The tenantLeakHarness CI gate lands alongside the v2 RLS rollout, turning tenant isolation failures into build breaks rather than incidents.

Scripted provisioning, no click-ops

Six idempotent provision-01..06 scripts replay the full infrastructure onto an empty Dokploy. Sees-the-reality reproducibility.

Deployment model & cost posture

Modular monolith. Root-server deployed. Redundant and predictable.

SoloFrame is being built as a modular monolith that runs on a single node today and scales to a multi-node deployment under Dokploy + Traefik + Docker Swarm - without a code fork or a rewrite. Stateless app replicas, shared state in Postgres / Redis / object storage, multi-replica-safe background jobs. Multi-node brings redundancy as well as scale: a failed node doesn't take the platform down, and rolling deploys happen without maintenance windows. Premium isolation is a topology choice, not a branch.

Deployment is to VPS and root-server infrastructure, not Vercel-style managed-PaaS hosts. That matters for AI-heavy workloads: per-invocation pricing converts engagement into variable cost. Root-server capacity is predictable; scale and redundancy come from additive nodes, not a vendor's usage meter. Platform architecture ->

USE OF CAPITAL

Priced against what it actually unlocks

We're avoiding the "big raise" posture. Capital maps 1:1 to specific GPU-bound ML workloads, named clinical partnerships, and the next tranche of evidence.

GPU-bound ML

ClinicalBERT fine-tune · RoBERTa sentiment · BKT/IRT · XGBoost risk scoring · pgvector embedding · retraining buffer. Each a scoped, independently deployable system with a clear metric.

Clinical validation partnerships

IRB-compliant head-to-head studies with named practices and research partners. Outcome: publishable distress-detection performance + content-outcome correlation data.

Content authorship

Scaling the evidence-based lesson library across additional clinical tracks, with clinical-reviewer honoraria and authoring infrastructure.

Sales motion

Moving from founder-led licensing to a small, targeted BD effort against psychiatric networks and behavioral-health benefits providers. Operator DNA means we know what we're not hiring yet.

New · Fourth production vertical shipped

60-Day Founder - a new vertical on the same engine.

The portfolio is no longer three verticals. 60-Day Founder is the fourth production vertical on the shared engine, shipped as content rather than a fork - the strongest signal of the company claim: SVTech can repeatedly turn one platform grammar into new domain products for new buyers, at a velocity funded teams struggle to match.

257 lessons · 32 courses · 6 tracksAI build coach8 artifact workshopsclassifier: noop · non-PHI
See the 60-Day Founder vertical →