Provenance
Data provenance built for markets
Elfa brings the best of 24/7 monitoring, signal surfacing, and movement tracking to power agents with real context awareness, turning raw market data into structured, relationship-aware intelligence.
Monitoring 45,000 accounts...GET /v2/data/keyword-mentions { "keywords": "HYPE", "period": "24h" }Response 200
{ "data": [{ "summary": "Galaxy Digital OTC-transfers 1.3M HYPE…", "source": "x.com/lookonchain/status/…", "timestamp": "2026-09-25T06:14:02Z", "entities": ["HYPE", "Galaxy Digital"], "confidence": 0.94 }, {
BTC
ETH
Solana
Aave
Jupiter
Pumpfun
Hyperliquid
Uniswap
Apple
Tesla
Amazon
Microsoft
Google
Meta
Spotify
Netflix
Sources traced · last 24h
X58%
Telegram21%
YouTube6%
TikTok4%Monitoring global events
Ask
Auto
Events and entities linked in real-time, from the people & events that matter down to the smallest voices.
Full provenance.
From ingestion to verification
- <7 secs
Accounts, channels, and sources
- 300k+
Markets tracked across equities, commodities & crypto
- 100K+
in delivery of high impact actionability insights
- 15%
Retrieval. Our Retrieval System model monitors, dedupes and filters across multiple platforms in real-time.
Extraction. The harness and tool calls come wired into the full knowledge stack, running live the moment you connect.
Screener. An always-on monitoring engine, ensuring agents act only when it matters.
Insertion. Approved candidate gets added into the context graph alongside other canonical entities.
Linkage. Across entities, the context graph connects the dot and enriches the data with relationship resolution & events.
Handover. All relevant information and parties are handed over to agents for evaluation.
The difference between slop and relevant data.
Built for global markets
Sources are selected and weighted for what moves price: the accounts & channels that break news first. Your agent starts on finance ground, not the open web.
Relationships resolved
Events and entities are linked as they arrive, so a mention resolves to the token, the fund, and the counterparties behind it. Your agent inherits the connections instead of rebuilding them every query.
Production-grade trust
Every data point carries its origin and the second it landed, so you can replay any answer back to the moment it was true. That's what a risk review or a postmortem asks for.
More signal per token
Retrieval returns resolved context, not raw pages. You spend fewer tokens per query and leave your agent less room to guess where the data thins out.
what are crypto natives talking about most right now
Hyperliquid at all-time highs. HYPE is the dominant CT story. It hit an ATH of $97.81 on September 23 and is trading around $96. The fuel is a stack of headlines: Kraken's parent Payward planning US onchain perps, Lion Group disclosing a major HYPE treasury position, and BitMEX shutting down, which people expect to push traders toward venues like Hyperliquid. It's also shipping steadily, with portfolio margin, manual borrows, and trailing stops recently going live. The counter-narrative is regulatory: CME and ICE have reportedly warned the CFTC and Congress that Hyperliquid could enable manipulation and sanctions evasion.
Perp DEX airdrop season. Tied to the above, Variational's tokenomics drop yesterday has points farmers running valuation math, and the broader "which perp DEX is next" debate is live. The Robinhood engineers' insider trading case, where they allegedly traded Hyperliquid perps ahead of confidential listings, added a darker angle to the same conversation.
Zcash and the privacy trade. ZEC hitting new highs has sparked a real fight over whether $ZEC should be governed more like Bitcoin or like a company. Arthur Hayes has also been pushing the Zcash bull case as the privacy narrative heats up.
Macro and regulation hangover. Last week was rough on paper. The Fed hiked 25bps to 3.75% to 4%, its first hike since 2023, with 16 of 18 officials projecting at least one more this year. The CLARITY Act failed cloture 50 to 49, short of the 60 needed. What people keep pointing out is how little BTC cared: it dipped to around $75,000 and then rebounded above $81,000 by the weekend. "Bitcoin shrugged off two objectively bad headlines" is itself a talking point.
Tokenization and stablecoin rails. Quieter but important: the SEC granted five years of conditional relief for platforms trading tokenized US stocks, and Circle launched the Arc mainnet with USDC gas and support for 22 fiat stablecoins.
One caveat: this is stitched together from news and aggregator coverage, which lags what's actually moving in Telegram groups and group chats by a bit. If you want the real-time social signal, that's more of a job for sentiment tooling than a web search.
The loudest crypto-native conversation is security risk: Bitget’s reported $351.6M wallet breach and the KelpDAO–LayerZero lawsuit. In the market chatter, traders are also focused on tokenized finance, XPL’s unlock-driven volatility, and speculative Robinhood-chain memes.
Fast & accurate.
Elfa flagged risk to CRCL the moment OpenUSD was announced, before the move showed up in price.
1h
Detected before the price moved
Elfa connected Jensen Huang's optimism on memory to names beyond Nvidia, surfacing SK Hynix and TSMC as the real beneficiaries.
2 months
The connection stayed high-impact in price action
Elfa connected Ottawa's C$400M to Trail's germanium output, reading a metals headline as a chip supply chain story.
43s
From detection to verification
Elfa traced the Lilly–Nvidia lab past both names to the AI drug discovery layer, surfacing RXRX and SDGR as the next to re-rate.
19s
From detection to verification
Elfa connected ANSEM's breakout to names beyond the token itself, surfacing JUP and JTO as the Solana ecosystem repriced.
1h
From detection to the price breakout
