Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.pacificalens.xyz/llms.txt

Use this file to discover all available pages before exploring further.

The AI Assistant is a conversational market intelligence tool embedded in the Analytics & AI page. It answers questions about funding rates, technical analysis, DeFi concepts, position sizing, and social trends — and every response is grounded in live Pacifica mark prices so you always get current data, not stale training-set figures. A connected wallet is required to use the assistant.

​
Dual-model routing

When you send a question, the assistant automatically routes it to the most appropriate model based on what you’re asking.
Elfa AI handles questions about social activity, trending tokens, Twitter/X buzz, whale narratives, and on-chain sentiment. It pulls live data from Elfa’s aggregation endpoints — trending tokens by social mention count, top tokens on Twitter/X, recent token news — and formats the results into a readable summary.If you ask the same social query within that window, you’ll receive the cached result instantly without consuming additional API quota.Elfa is triggered by keywords such as: trending, Twitter, social media, whale, smart money, sentiment, narrative, hype, news, viral.
Groq handles everything outside social queries: technical analysis, DeFi mechanics, risk and position sizing, funding rate explanations, open interest interpretation, liquidation math, and general market questions.Live Pacifica mark prices for BTC, ETH, SOL, BNB, XRP, DOGE, AVAX, LINK, OP, ARB, and other active markets are injected into the system prompt on every request. This ensures the model references real current prices rather than outdated training data.

​
Example questions

The following questions illustrate the range of topics the assistant handles well. Elfa routes are marked with (E); Groq routes are marked with (G).
  • “What’s the funding rate on SOL right now?” (G)
  • “Analyze BTC technically — key levels and trend structure” (G)
  • “Which tokens are whales accumulating?” (E)
  • “What are the top trending tokens on social media today?” (E)
  • “Is 2% risk per trade too much for a $1,000 account?” (G)
  • “Most talked-about coin on Twitter right now?” (E)
  • “Explain how liquidation cascades work in perpetuals” (G)
  • “Where is smart money flowing?” (E)
  • “What is the difference between isolated and cross margin?” (G)
  • “Which markets have the highest open interest on Pacifica?” (G)

​
Analytics & AI page

The AI Assistant is part of the broader Analytics & AI page, which you navigate using the left sidebar. The full section includes:
SectionWhat it shows
AI AssistantConversational market intelligence (this feature)
Funding Rates — Most ExtremeBar chart of the 10 most extreme funding rates across all 63 markets, color-coded by direction
All Markets Funding RateColor heatmap of funding rates for all 63 markets simultaneously — high-intensity cells display white text for readability
Long/Short RatioPer-market bias bars estimated from funding rate direction
OI DistributionDonut chart with total open interest in the center label, plus volume dominance bars
Market SignalsReal-time alerts for OI spikes (≥ 2% change) and funding anomalies
Liquidation MonitorFull liquidation section with grid, list, and feed views (see Liquidation Monitor)
Global NewsAggregated news feed with All / Crypto / Macro filter tabs
Economic CalendarForexFactory macroeconomic events for the current and next week
The AI Assistant requires a connected wallet. If you are browsing as a guest, the input field is locked and a Connect → prompt is shown in its place. Connect any supported Solana wallet (Phantom, Solflare, Backpack, or Privy embedded) to activate the assistant.
Responses are cached server-side to reduce latency and API usage. Social and sentiment queries via Elfa are cached for 15 minutes; market analysis and general queries via Groq are cached for 30 minutes. If market conditions change rapidly, you may want to rephrase your question slightly to bypass the cache and fetch a fresh response.