Joist AI builds proposal automation software for AEC teams, with a tech stack built around Python, React, and multi-model LLM orchestration (OpenAI, Claude, Mistral, Llama via LangGraph and LlamaIndex). The company is actively shipping agentic AI applications and ML evaluation frameworks while adopting infrastructure-as-code tooling (Terraform, Pulumi) — indicating a shift toward production-grade AI deployment at scale. Engineering dominance in the hiring mix (5 of 14 roles) paired with internal pain points around "deploying bleeding-edge AI to production" and "reducing ticket-waiting" suggests the team is solving hard problems in generative model evaluation and operational efficiency.
Joist AI is a sales enablement and content management platform for architecture, engineering, and construction teams. The product helps proposal teams respond to RFQs and RFPs by retrieving institutional knowledge, generating compliant content, and catching quality issues — reducing both turnaround time and internal friction. Founded in 2022 and based in San Diego, the company operates at 11–50 employees with a senior-heavy engineering organization (10 senior-level roles across engineering, product, and support). Active development spans web application infrastructure, agentic AI for autonomous decisions, ML pipelines for production scalability, and observability tooling.
Joist AI integrates multiple large language models including OpenAI, Claude, Mistral, and Llama, orchestrated via LangGraph and LlamaIndex for retrieval-augmented generation and agentic workflows.
Front-end: React. Backend: Python with FastAPI. Cloud: AWS (Lambda, RDS, Step Functions). LLM/ML: PyTorch, TensorFlow, OpenAI, Claude, Mistral, Llama, LangGraph, LlamaIndex, Crew AI, FAISS, Pinecone, Weaviate. DevOps: GitHub, GitHub Actions, Terraform, Pulumi.
Other companies in the same industry, closest in size