PRODUCTS AND IP · PATTERNS AND ACCELERATORS
Reusable patterns behind practical AI systems.
Aurora AI is an advisory agency backed by accelerators and repeatable patterns — we do not start from a blank page. What you see here is the architecture and components we use for client projects — from agent management and an AI gateway, through proposal automation and knowledge systems, to operational workflows.
PRODUCTS AND IP
Ten patterns we already have.
Each entry is a working component, an internal accelerator or a ready reference project — used in real client work, not sold "off the shelf".
AgentHub
A production agent-management dashboard with role-dependent views (CEO, CFO, CMO, COO and operations teams) — built from ready templates. Vendor-independent event normalization: Anthropic, OpenAI, Gemini, local LLMs and tools connected via MCP. Built and used inside Aurora; deployed with clients.
- Deployed· production — client deployments
AgentDash
A multi-tenant dashboard layer with tenant and admin panels, role-based access control, visibility of agents, tasks and logs, and live data over WebSocket. Production stack: Node.js / Fastify / PostgreSQL / JWT — deployed and running.
- Deployed· production stack — running today
FluxGate
An AI gateway for governed deployment: model routing, semantic cache, multi-tenant API keys with per-team budgets, billing and SLA visibility, an audit trail, SSO/OIDC, MCP-compatible agents. Deployed with enterprise clients.
- Deployed· production — enterprise deployments
AI Proposal Agent
A proper RFP agent — it knows your products, pricing, history, security policy and acceptance criteria. It cuts the go/no-go decision time from 8 h to 15 min; cost lower by ~90 %. Not a chat subscription — a system with the company’s knowledge, an audit history and ROI measurement.
- Deployed· production — with metrics
Knowledge Capture + RAG
An interviewing agent that leads through domains and turns tacit knowledge into structured Markdown chunks with YAML metadata and a confidence score. Multi-dimensional search (domain × topic × type × confidence × tags) feeds RFPs, internal assistants and a reusable sales corpus.
- Deployed· built — internal system
Email Agent (M365)
A project email-intelligence agent — triage, classification by project, a daily digest. Designed with human approval gates for every outgoing action: an observer and an assistant, not an autonomous actor.
- Deployed· built — Microsoft 365
Product / media metadata automation
Enrichment of product images for e-commerce — SEO-grade names and alt texts from raw photography, so that listings are searchable from day one.
- Deployed· built — operational tool
Mission Control
Aurora AI’s internal command center — a web app the team starts its working day with. It reads the company knowledge base in read-only mode and assembles a morning view from it: project status, today’s priorities and current tasks in one place. In the version being developed it orchestrates AI agents that delegate tasks, prepare draft documents — emails, presentations, notes — and summarize project health. The boundary is hard: the tool never changes the knowledge base silently — the only path to synchronization is a deliberate refresh, and every agent work product passes through a review queue before anything takes effect.
- Internal accelerator· v0 working, v1 in active development
Token Dashboard
A local, private analytics tool for working with agentic coding in Claude Code, which Aurora Labs uses internally. It reads local session transcripts, builds a local cache and serves an interface on localhost — cost per prompt, tool and file heatmaps, subagent attribution, cache analytics, project comparisons and rule-based hints. It runs fully locally: zero telemetry, no external calls. The tool was built on an open-source (MIT) base that Aurora hardened and extended for its own needs; an OSS release is in preparation.
- Internal accelerator· developer tool used inside Aurora Labs
AI platform for large-format printing
At the core of the platform is an AI image lab: the client generates artwork from a description, upscales its resolution to print scale and colorizes the material, and if they have their own file — they upload it along the same path. The platform assembles the generated or uploaded source into a product ready for large-format printing, from meter-scale artwork to wall format. Premium positioning: the image lab is the value here that a typical poster shop does not offer, not an add-on to the catalog. We built and deployed this platform as a working online application — it runs in production on the client side, and we keep developing the codebase it stands on.
- Deployed· running in production at the client; Aurora’s own codebase developed further
HOW WE USE THEM
A pattern as the foundation of a project, never instead of advisory.
We use a pattern as the foundation of a specific client project — so we do not start from a blank page when we know how to solve a similar problem. We never sell a license or a component detached from the advisory context: without process mapping, data classification, permission boundaries and success measures, even the best accelerator quickly becomes technical debt. Aurora AI is an advisory agency — products and IP exist to shorten the path to a working workflow, not to replace the conversation about the process.
LET'S START
Use a pattern, not a blank page.
Bring one process, one investment decision or a bottleneck — we will show which of these patterns gives the fastest path to a working solution in your context.