Nexus AI Rolls Out Real-Time Collaborative Knowledge Graph for Enterprise Teams

Turning Team Chaos into Structured Intelligence

Most enterprises have a knowledge problem. Documents live in ten different tools. Decisions get made in Zoom calls and never get recorded. Institutional memory walks out the door every time someone switches jobs. Nexus AI, a four-year-old startup headquartered in Austin, Texas, thinks it has built the solution: a platform called CoGraph that continuously maps an organization's collective knowledge into a real-time, queryable graph — and hooks directly into the tools teams already use.

Launched publicly on September 15th, CoGraph connects to Google Workspace, Microsoft 365, Slack, Notion, Confluence, Jira, and GitHub via over 200 native integrations. As users work — drafting documents, sending messages, filing tickets — CoGraph's ingestion engine quietly parses, links, and structures the content into a knowledge graph with entities (people, projects, products, decisions), relationships, and confidence scores. Any employee can then query the graph in plain English: "What was decided about the Q3 pricing strategy?" or "Who owns the API integration roadmap?" The system returns sourced answers with citations and confidence ratings.

The underlying model is a fine-tuned variant of GPT-5, which Nexus AI trained on 2.8 trillion tokens of enterprise data under strict data-retention agreements that prevent any customer data from being used in future model training. The fine-tuning process took six weeks and cost approximately $4.1 million in compute, according to Nexus AI's head of research, Dr. Amara Osei.

Early enterprise customers include accounting giant Deloitte, which deployed CoGraph to 45,000 consultants across North America; logistics firm Flexport, which is using the platform to track cross-border regulatory knowledge; and healthcare network MedStar Health, which is piloting the system for clinical guidelines retrieval. Deloitte reported that new consultant onboarding time dropped by 34% in a controlled trial, since new hires could query institutional knowledge directly instead of scheduling dozens of orientation meetings.

Pricing follows a per-seat SaaS model: $28/user/month for teams under 500, with volume discounts available for larger deployments. Nexus AI also offers an on-premises deployment option for regulated industries at $42/user/month, with a minimum annual contract of $500,000. The company says it is already profitable on a unit economics basis, with a gross margin of 71% — unusually high for a young enterprise SaaS company.

Nexus AI raised $120 million in Series C funding in August 2027, led by Insight Partners, bringing total funding to $210 million. The company employs 340 people globally and expects to reach 500 employees by year-end, with a planned expansion into the Asia-Pacific market in early 2028.

The broader enterprise knowledge management market was valued at $38 billion in 2026. Nexus AI's approach — grounding LLMs in proprietary organizational data rather than generic web text — represents a deliberate pivot away from the "AI that answers everything" approach toward what the company calls "AI that knows what your company knows." It's a narrow but increasingly crowded space: competitors include Glean, constructed from web-scraped public data and launched its own enterprise knowledge graph in March 2027, and Notion AI Q&A, which bundles GPT-4o-powered search directly into the popular productivity suite.

What separates CoGraph, Nexus AI argues, is depth of integration and graph-native architecture. "Most competitors are just adding a chatbot to a search box," said CEO and co-founder James Kwan. "We're building a memory for the organization that gets smarter every day."

The platform is available now. Nexus AI is offering a 30-day free trial for teams of up to 25 users, with no credit card required.