LUMEN — AI Assistant / Chat UI
Think out loud.
A premium AI assistant interface — the polish of the best chat products as a clean static front-end. It streams assistant replies token-by-token with rich markdown and copy-able code blocks, keeps a date-grouped searchable history, and ships a prompt library, a discover gallery and full settings. Dark-first with a ⌘K palette and one documented LLM seam.
Role · Design + front-end build (simulated LLM via one swappable seam) · Updated Jul 2026
- A streaming chat with markdown, copy-able code blocks and message actions (copy, regenerate, feedback)
- A date-grouped, searchable conversation history with rename/delete + eight seeded demo threads
- A Prompt Library (17 templates), a Discover gallery (8 personas) and full Settings — all wired
- A ⌘K command palette, no-flash dark/light theming and a Zustand + localStorage persistence layer
- Delivery
- 24–48h
- Licence
- One-time licence · full source
- Support
- 14 days setup support
- Updates
- Free updates included
Backend · Your LLM API
A complete chat UI you connect to any LLM via one function (streamCompletion). A worked Anthropic SDK example ships in the file; a static export needs a small proxy route for the API key (documented).
Delivered as full Next.js source. Implement streamCompletion() against your LLM (with a key-proxy route) to go live.
What makes it special
Simulated streaming, done right
Assistant replies reveal token-by-token with a caret and a brief 'thinking' beat from one async-generator seam; the reveal is deterministic and snaps to instant under reduced motion, and the chat log is an aria-live region.
A real conversation product
Eight seeded threads grouped by date with fuzzy search, inline rename, delete and new-chat. Rich markdown — headings, lists, tables, fenced code with per-block copy — plus per-message copy, regenerate and feedback.
One LLM seam, provider-agnostic
Every response comes from streamCompletion() — an async iterable of text chunks. Replace its body with a streaming SDK call (a worked Anthropic example is included) and nothing else in the app changes.
Library, Discover & Settings
A categorised, searchable prompt library that loads templates into the composer; a discover gallery of tailored assistants; and settings for model, a persisted system prompt and theme.
Accessible, near-zero CLS, static
Keyboard send, labelled controls, focus-visible, focus-trapped dialogs, role=switch and role=log, no external images, self-hosted fonts, no-flash dark/light — a fully static export.
What’s included
- A streaming chat with markdown, copy-able code blocks and message actions (copy, regenerate, feedback)
- A date-grouped, searchable conversation history with rename/delete + eight seeded demo threads
- A Prompt Library (17 templates), a Discover gallery (8 personas) and full Settings — all wired
- A ⌘K command palette, no-flash dark/light theming and a Zustand + localStorage persistence layer
- A single documented LLM seam (one function) with a worked Anthropic SDK example, plus full typed source
More Technology & SaaS templates
Explore more collections
Frequently asked
- What do I get when I buy LUMEN?
- A one-time licence with the full, editable source code, including a streaming chat with markdown, copy-able code blocks and message actions (copy, regenerate, feedback), a date-grouped, searchable conversation history with rename/delete + eight seeded demo threads, a prompt library (17 templates), a discover gallery (8 personas) and full settings — all wired. 14 days setup support and free updates included.
- What is LUMEN built with?
- It's a Next.js dashboard / admin, built with Next.js 15, TypeScript, Tailwind v4, Zustand, react-markdown. You receive the complete source to host and extend anywhere.
- Can YantraCore customise LUMEN for my brand?
- Yes — buy it as-is for $3,000, or have YantraCore tailor it to your brand, content and requirements. Start from this page or commission a custom build.
- How fast is delivery?
- Delivery is 24–48h after purchase.
Make LUMEN yours.
Buy it as-is, or have YantraCore tailor it to your brand and content.