State-of-art assessments, build-vs-buy analysis, and platform buildability studies that ground Assembly's engineering decisions in the current state of the industry.
All 15 technical modules rated on build complexity, established precedence, and innovation risk — classified into Commodity / Established / Cutting Edge / Bleeding Edge tiers, with closest existing solutions, gap-to-build estimates, and a strategic roadmap by module risk.
Read analysis →A buildability classification of the app types commonly generated on Lovable-style platforms, evaluated against a self-contained Supabase + Next.js + Stripe stack. Interactive matrix: fully / partially / not-fully buildable, with feature and external-module tagging.
Open report →Multi-human + multi-agent real-time voice orchestration (M4) does not exist as a product or open-source project. It is entirely original engineering — 4–6 months dedicated effort, started on Day 1.
Voice (sub-100ms STT/TTS), code generation, auth, and observability are fully solved in 2026. Deepgram + Cartesia + Claude API + Clerk + PostHog: configure and move on.
The M12 product data model is the prerequisite for everything. A 2–3 week schema design sprint comes first — AgentDecision attribution and the FeatureGraph are the hardest design problems.