bizcity-twin-ai

The second brain concept

Bizcity Twin AI reframes WordPress as an operating platform for a business rather than a publishing tool. The README presents it as a second brain that the business itself trains with its internal knowledge, then uses to run CRM, content, marketing, customer care, and automation. The authors argue that general chat AI tools sit apart from an organization's data, processes, and working methods, and that internal knowledge is fragmented across records, documents, products, customers, and conversation history. Their approach turns WordPress into the AI operating platform by organizing documents and experience into a knowledge graph, then combining that graph with Graph RAG, memory, CRM, communication channels, and workflows. The practical goal stated in the README is to help businesses own their data and processes, cut repetitive work, improve customer care and marketing, and adopt AI at a cost that suits small and medium sized organizations. The project follows an open source, free core model, so a business can self host on its own WordPress stack and expand by degrees instead of buying many separate AI accounts. The tone of the document is a product thesis more than a technical manual, but it states the direction clearly. The free core model means a business can start on its own WordPress install and add capabilities over time, which the README presents as a lower risk path than adopting many disconnected paid AI services.

Graph RAG as the core engine

Graph RAG is described as the core engine and the shared knowledge foundation of the second brain. The README explains that the aim is not only to find a text fragment similar to a question, but to connect entities, relationships, events, provenance, and time across the business. Vector only retrieval is useful for similarity search, the document says, but it is weaker at multi hop reasoning, at telling apart entities that share a name, at tracking change over time, and at explaining where an answer came from. Graph RAG adds structure and provenance so that every module, from chat and CRM to automation, works from the same contextual knowledge base. The retrieval pipeline is laid out as six steps: structure the inputs into entities and relations, retrieve with a BM25 and dense fusion, expand by graph traversal on relations and time, rerank by relevance, recency, and provenance, compose an answer or workflow action, and record the outcome as memory, audit trail, and feedback. This pipeline is what the README says gives the system persistent business memory, grounded answers, decision support, and a path from understanding to action in daily operations. By recording the outcome as memory and audit trail, the pipeline also gives the system a feedback loop that the README says makes the second brain more useful as it is used.

Tooling, channels, and architecture

The integrated tooling table lists what the plugin ships. Graph RAG plus a KG Hub is the central reasoning core. Twin CRM handles leads and campaign funnels. A channel gateway connects Zalo OA, Facebook, Messenger, Zalo Bot, and webchat. A Bizcoach pro component analyzes people profiles and collaboration fit. Automation and scheduler modules handle workflow, reporting, and triggers. A content stack covers marketing content, documents, and image generation. The README also describes how businesses use Twin AI across CRM and customer care, marketing and content, internal operations such as task assignment and KPI summaries, and people management. The high level architecture diagram places the KG Hub at the center, feeding Twin CRM, the channel gateway, and automation, all sitting on the WordPress operating system that provides data, UI, and API. The project structure shows a core directory with knowledge, memory, intent, channel gateway, scheduler, and tools, plus modules such as twinchat, twinshell, and twinsearch, and plugins such as the CRM and Bizcoach. The document closes with a legal notice that trademarks Bizcity and Bizcoach are protected and that copyright is declared for the digital twin brain network model and its toolchain. The structure listing shows that the plugin is organized as core modules plus optional plugins, so a deployer can see which pieces provide the CRM, the coaching, and the content features before installing.

Editorial conclusion

The project is released under the GPL-2.0-or-later license and its most recent commit was on 2026-08-26.