BoxSight LLC is a USA-based technology architecture firm. Nine capabilities on one backend stack. Each entry below states its status explicitly — running in production, pre-launch, or research.
Live Intelligence Products
Receipt digitization and expense intelligence. Images are sent to Gemini for OCR and field extraction; structured outputs — merchants, line items, categories, spending patterns — land in a searchable 88-table schema. Features a gamified "Financial City" visualization.
AI/ML research synthesis engine. Ingests 30+ curated sources daily at 03:00 UTC and generates structured briefings via vector search and LLM inference. Running continuously since initial deployment.
Academic collaboration platform for sustainability research. Aggregates 20+ university-affiliated feeds plus curated RSS sources, with AI-generated weekly video briefings, trend detection, researcher profiles, and threaded discussions. Seeded with 20 universities including Lund University.
Tap-first nearby-place finder where travel time replaces distance as the search dimension. Two taps — category plus time radius — return matching places with AI-generated result labels, 22 pre-built situation bundles, and multi-stop trip chaining. Currently running on iOS Simulator pre-launch.
Infrastructure & Frameworks
Two-primitive specification language (Entity + Transition) with a 6-gate LLM pipeline and 5-model consensus mechanism. Moves from natural-language intent to deployed application across six target stacks. Used to build six applications with zero human coding after intent submission. ADL-UI — a browser-based spec session that generates a preliminary build plan — is coming to adl.boxsight.ai.
Read the ADL Whitepaper →Multi-model consensus engine using the Model Context Protocol. Queries 5 LLMs simultaneously with Jaccard similarity clustering to surface disagreement and reduce single-model failure modes. Published on GitHub and npm.
A methodology for auditing the rules that hold the internet together — IETF RFCs, W3C specs, OAuth/WebAuthn/JOSE flows — by reading what they MUST/SHOULD/MAY require and then proving where real-world implementations diverge. HUNTER's first novel CVE shipped in 2026: CVE-2026-48522 — a previously-unreported divergence across the JOSE ecosystem (Python/Java/Node) — found by following the same workflow it applies to every protocol.
Now spanning eleven phase milestones across JWT/JOSE, X.509, OAuth 2.0, WebAuthn, CBOR/COSE, DANE (TLSA), TLS 1.3 trust-anchor handling, SAML, and others. Each milestone runs five gates — RFC grounding, probe authoring, library matrix, oracle determinism, and confound audit — before any divergence claim is made, so findings are reproducible by anyone with the spec text and the probe corpus.
Read the HUNTER Capability Brief →Local-first network security monitoring. Distributed agents running local LLMs (Ollama) produce plain-language threat detection and network health reports. No telemetry leaves the perimeter.
Pre-flashed USB for zero-configuration network OS deployment. Plug in, pick an OS, walk away. Installs a PXE boot control center alongside any existing OS with cloud licensing, auto-updates, and telemetry. Supports Linux, macOS, and Windows targets via iPXE, GRUB2, and WinPE. 88 requirements delivered across POSIX Shell, Node.js, Python, and PowerShell implementations.
Shared Intelligence Fabric
Nine capabilities. One backend stack. Every BoxSight production system runs on PostgreSQL + pgvector for structured and vector data, FastAPI for async APIs, and Redis for caching and real-time state. SpendCity, LIITOS-AI, LIITOS, and Topos share a single production server with centralized authentication, unified deployment scripts, and a common admin console pattern.
The multi-model consensus engine (cross-review-mcp) runs across the ecosystem — validating LIITOS-AI synthesis output, gating ADL pipeline stages, and driving automated security analysis in HomeSentinel. The same Gemini integration that powers SpendCity's receipt OCR also generates Topos result labels and LIITOS weekly briefings.
This is deliberate architecture, not accidental reuse. A capability built for one product — vector search, LLM reasoning, on-device inference — deploys to the next without re-engineering. LINETBOOT-USB extends this thinking to bare-metal infrastructure, applying the same spec-driven development methodology to hardware provisioning that ADL applies to software generation.