SkyeChat is the Kaixu-powered conversational interface for operators who need more than a chatbot. Multi-model streaming, token-gated access, spend cap enforcement, conversation export, and PWA offline — chat infrastructure built for the way real teams work.
Requires Kaixu virtual key. Get yours →
You're running a real operation. Client deliverables, SOP drafting, research threads, billable work output — and you're doing it through a consumer chat interface that exposes your API key in the browser, has no spend controls, and loses your conversation the moment you close the tab.
Most browser AI tools pass your provider key client-side. One devtools tab and it's gone. SkyeChat routes through the Kaixu Gateway — your key never leaves the server.
Generic chat UIs have no billing awareness. SkyeChat enforces your Kaixu virtual key cap — 402 spend-limit responses surface in the UI, not as a surprise invoice.
Close the tab, lose the thread. SkyeChat persists conversation history to IndexedDB and lets you export any session as markdown, plain text, or structured JSON.
ChatGPT is GPT. Claude is Claude. SkyeChat lets you select the model per conversation — same interface, multiple providers, routed through the Kaixu multi-model gateway.
Every feature in SkyeChat exists because operators needed it — not because it looked good in a demo.
Responses stream token by token via Server-Sent Events — no polling, no waiting for a full response block. The UI reflects each delta as it arrives with smooth progressive rendering.
Pick the right model per conversation. Gemini 2.0 Flash for speed, others for depth — all routed through the Kaixu Gateway with unified billing and token tracking.
Your Kaixu virtual key is stored encrypted and never exposed in the browser. All gateway calls hit the server; the client only sees streaming deltas and status codes.
Your virtual key carries a monthly cap. SkyeChat surfaces HTTP 402 (cap reached) and 401 (invalid key) directly in the chat thread — no silent failures, no surprise overages.
Every session is persisted to IndexedDB. Open SkyeChat tomorrow and your threads are waiting. Search, revisit, and continue any prior conversation from the sidebar.
Export any conversation as markdown for docs, plain text for emails, or structured JSON for automation pipelines. One click from the thread toolbar.
Install SkyeChat to any home screen. Service worker caches the shell and all prior conversations. Draft messages offline — they queue and send when signal returns.
A persistent budget strip shows tokens used this session, monthly spend, and remaining cap — all pulled from the gateway meta event on each response. Always visible, never hidden.
Set a system prompt per conversation. Give the model a role, a client context, or an operator persona — scoped to that session, saved alongside the thread, exportable.
From zero to a working AI conversation thread in under two minutes.
Request a Kaixu virtual key through the SOL gateway portal. Your key carries a configurable monthly spend cap. It never goes client-side — SkyeChat stores and uses it server-side only.
Choose the AI model for your session from the model picker. Optionally set a system prompt to give the model a focused role — client persona, SOL operator, research assistant, whatever fits the task.
Every message routes through the Kaixu Gateway via SSE. Tokens stream into the thread progressively. The budget strip updates with every gateway meta event — token count and spend visible at all times.
Save the conversation as markdown and paste it into SkyeDocx. Export as JSON and feed it into a SkyeFlow automation. Or simply close the tab — the thread persists in IndexedDB waiting for your return.
| Feature | SkyeChat | ChatGPT / Claude | Generic AI Chat Wrapper |
|---|---|---|---|
| Server-Side Key Storage | ✓ Always | ✓ Their infra | ✗ Client-side |
| Spend Cap Enforcement | ✓ 402 in UI | ⚠ Hard limit only | ✗ None |
| Multi-Model Selection | ✓ Per session | ⚠ Paid plan only | ✗ Single model |
| Persistent Thread History | ✓ IndexedDB | ✓ Cloud-based | ✗ Session only |
| Export (MD / TXT / JSON) | ✓ All 3 | ⚠ Limited | ✗ None |
| PWA Offline Mode | ✓ Fully offline | ✗ Requires connection | ✗ None |
| SOL Ecosystem Integration | ✓ SkyeFlow, SkyeDocx | ✗ Siloed | ✗ Siloed |
SkyeChat is the conversation layer. The SOL ecosystem turns those conversations into documents, automations, and deployed code.
Every SkyeChat message routes through the Kaixu Gateway — multi-model routing, per-customer virtual keys, spend caps, and diagnostic events broadcast for full visibility.
Export a SkyeChat conversation as markdown and paste it into SkyeDocx. Proposals, SOPs, reports — the chat output becomes the document draft without re-typing a word.
Export structured JSON from any thread and feed it directly into SkyeFlow automations. The conversation becomes the trigger input, not a manual copy-paste task.
Use SkyeChat to discuss architecture and test prompts. Switch to SkyeKIDE to implement. Both tools share the same Kaixu key and gateway budget — one account, unified spend tracking.
Server-side keys. Spend cap enforcement. Real-time streaming. Persistent history. Multi-model selection. Export to any format. SkyeChat is the conversation tool that fits inside a real workflow — not alongside it.
Kaixu virtual key required. Talk to SOL →