A unified intelligence platform — conversational AI, autonomous SDLC pipelines, installable agent skills, knowledge & code graphs, and real-time team Spaces — built for the UBT engineering team.
Each product is independently accessible and purpose-built for a distinct engineering workflow.
Conversational AI assistant with real-time access to your codebase, documentation, Confluence pages, and Jira tickets. Ask anything about the UBT platform - the bot answers with context-grounded, source-cited responses.
AI-powered developer portal — run autonomous SDLC pipelines, collaborate in shared Spaces, install agent Skills & Kits, and tune the agent layer. You stay in control at every human-in-the-loop gate.
Central control panel for all indexed knowledge. Manage document ingestion, browse graph tools, configure data integrations, and access the MCP server endpoint - all from one hub.
Autonomous SDLC workflows built on LangGraph — each pauses at human-in-the-loop gates so you review, edit, and approve before anything ships.
Reads a Jira ticket → plans → writes code → opens a PR. Grounded in the code graph, with a quality-scan + approval gate before the PR.
BA-mode agent: analyses requirements and produces user stories with acceptance criteria, ready for the backlog.
Turns a ticket + Figma screens into a full suite — manual cases, Gherkin, QAlity+ CSV, and runnable Playwright specs with coverage analysis.
UX flow → screen briefs → interactive HTML prototype → QA review. All in-browser, no external design tools.
Runs generated Playwright specs against a live environment, self-heals failures, and reports results back to Jira.
Bring the same grounded workflows into Claude Code, GitHub Copilot, or claude.ai. Skills are portable agent capabilities; Kits bundle them; the MCP server feeds them live context.
Browse, install, rate, fork, and submit reusable AI skills and agents. A review queue keeps the org catalog high-quality.
22 skills covering the full STLC — triage, requirement analysis, test design, authoring & running automation, and reporting back to Jira / QAlity+.
Add Cortex as an MCP server in your IDE for per-user, source-cited access to the code graph, doc graph, and knowledge base tools.
Agents remember decisions and conventions across runs — managed per project and task type in the AI Engineer console.
Pick your target in the catalog, copy the one-line install, and every skill, prompt, and tool binding lands in your editor at once.
Spaces are context rooms where everything pinned grounds the AI — the agent knows your workspace before it starts.
Real-time team chat. Mention @cortex to invoke the assistant with the Space's full pinned context injected automatically.
Deliverables from AI tasks are versioned with line-by-line diffs and a comment → request-changes → approve review flow.
Pin knowledge articles, repos, and notes — every AI task launched from the Space inherits them as grounding.
Add colleagues as owner / editor / viewer. Roles gate who can approve artifacts and manage the Space.
Every component is containerised, observable, and designed for the internal UBT engineering environment.
