QuassLabs
QuassLabs is the AI lab and product studio of Taylor Quass — a cloud architect with 23 years building software and half a decade directing AWS infrastructure. We design AI-native products and the secure, cost-aware systems that run them, taking ideas from a rough brief all the way to production.
A working library of internal skills, distilled into the capabilities we bring to a build.
Automated PR-style code review with security and dependency scanning, integrated with CI gates.
Caught a CVE-flagged transitive dependency before merge in a recent sprint; flagged a model-config mismatch a bulk-edit had introduced.
Routing and coordination across local + cloud AI models for cost-aware task delegation.
Routes deterministic tasks to local models; reserves cloud reasoning for ambiguous architecture and judgment calls. Typical cost reduction: ~70%.
Consent-gated penetration testing, OWASP audits, and SBOM-based supply-chain analysis.
Pre-launch security gate produces CLEAR / CONDITIONAL / BLOCK verdicts with CVSS-scored findings; ships an SBOM in CycloneDX + SPDX formats for compliance audits.
Voice cloning, text-to-speech, audio segmentation, and podcast-style composition.
Generates synchronized audio + text reading mode for ebooks; produces brand-voiced narration on demand.
Valuation modeling, GTM planning, and engagement-scoped cost estimation.
Produces 5-year DCF + cap-table + 3-exit waterfall; emits Series A pitch financials with brand-styled PPTX in under an hour.
AWS account bootstrap with budget caps, IAM boundaries, and credential management.
Day-zero account setup produces budget alerts + least-privilege IAM + structured logging in under 30 minutes.
ADB-driven Android automation with vision-model UI state reading.
Wireless ADB loop drives device through full user flows; vision model reads UI state; ADB executes input commands at ~15× the cost reduction vs cloud-first.
Persistent context preservation across long-running AI sessions and engagements.
Per-project memory hydration on session start; cross-session context survives compaction; engagements pick up months later without re-onboarding.
Story-to-Jira-to-GitHub-to-release traceability with worklog automation.
Bi-directional sync between scope storyboard and Jira; daily worklog suggestions from GitHub commits; release-readiness audits across 6 sources.
Token compression, local-first inference routing, and per-session cost ledger.
Asymmetric session compression (AI turns compressed; user turns preserved verbatim) cuts transcript size 40-60%; local inference handles 70%+ of routine tasks at near-zero marginal cost.
Load testing, SLO/p99 tracking, and anomaly detection across deployed services.
k6 + Vegeta SLO gating with 5 profiles (smoke/load/stress/spike/soak); unified dashboard pulls CloudWatch + Prometheus + N8N + load-test results.
Figma QA, design-token systems, and competitive design review with developer-ready specs.
Light/dark token systems with FOUC prevention; competitive design review pulls 3-5 competitor references and emits prioritized improvement suggestions.
Products we've shipped or are shipping — each backed by a case study.
Live Automated worklog + email triage for engineering teams
Live Estimation as a service — from rough brief to proposal in minutes
Live White-label publishing platform + indie-author publishing house
Live Scripture video access in thousands of languages — stream or offline
Live Tooling for drywall contractors — estimating and job workflow
Live Preference-based decision engine — date ideas, games, movie nights
Coming soon Coming soon
Coming soon Coming soon
Scoping, architecture, AI integration, or a full product — start a conversation.
Start a conversation