Every card comes from InnerChispa’s canonical registry. When a new project is approved, MCP exports the content and Astro updates this page.
Source updated: Oct 21, 2026, 01:26 PMPersistent cognitive memory layer with hierarchical retrieval: episodic logs in MongoDB and vector semantic graphs in Qdrant with Cognee (Winner: Best Use of Cognee in San Francisco).
Live multi-vendor network topology mapper and hardware anomaly correlation engine integrating Grandstream, Ruijie, and UniFi telemetry.
Local inference acceleration stack on AMD Radeon RX 9700 silicon using ROCm 10 and vLLM, providing private 30B LLM generation.
Contextual smart environment interface bridging Alexa, Home Assistant, and InnerOS reasoning for hands-free voice operations.
published End-to-end B2B quoting pipeline that ingests unstructured customer inquiries, verifies margins and fiscal withholdings, and requests owner approval before issuing invoices.
published Auditable technology credit and funding tracker managing infrastructure grants, compute allocations, resource verification, and compliance documentation for AI research.
published The canonical Model Context Protocol (MCP) router connecting organizational memory, MongoDB collections, Contífico invoicing, Evolution WhatsApp, and supervised agent pools across sovereign nodes.
published Offline-first intelligent quoting platform running on local Ollama models. It transcribes audio notes, validates Ecuadorian tax requirements, and prepares auditable B2B service proposals with zero cloud API dependencies.
Core sovereign multi-agent platform orchestrating local AMD inference, MongoDB memory, MCP tooling, and field operations.
Evidence-first infrastructure diagnostic agent built at AWS Builder Loft SF, combining live read-only telemetry and verified citations.
High-speed SRE reasoning and network security mesh powered by Nebius Cloud and NVIDIA TensorRT, correlating firewall events and preventing outages.
Audited evidence ledger tracking external cloud compute grants, sponsor credits, and compute consumption records.
Remote perimeter security and gateway monitor bridging edge video perception, automated intrusion logging, and verified alerts.
Read-only live infrastructure investigation copilot built at AWS Builder Loft SF, combining You.com live research with approval-gated actions.
An experimental experience connecting authorized memory, context, responsible pattern discovery, wellbeing, and human agency—support, never automated diagnosis.
High-throughput agent-to-agent society running local Qwen 2.5 Coder on AMD ROCm silicon, debating infrastructure decisions with full evidence provenance.
PC Doctor provides computer diagnostics, repair, maintenance, networking and IT support in Ecuador, while InnerChispa builds supervised AI systems, automation and private operational memory for real business workflows.
Yes. The team can coordinate remote support when the issue can be diagnosed safely online, and on-site support when hardware, networks, cameras or physical infrastructure require field work.
The fastest path is WhatsApp at +593999059000 or email at [email protected], with a short description of the device, issue, location and urgency.
AI agents are used as supervised tools for intake, evidence organization, pattern detection, quoting assistance and operational follow-up; sensitive actions remain human-reviewed.