Powering 200 Teams with Real-Time Intelligence: Elfa x Mantle AI Awakening Hackathon

Builders need more than an LLM — they need context

Hackathons built around AI agents run into the same wall almost immediately. A model can reason, plan, and write code, but it has no live view of what's actually happening in the market it's meant to operate in. Every team at Mantle’s AI Awakening Hackation faced the same early decision: to spend hackathon time building a social-listening and sentiment pipeline from scratch, or plug into an existing service.

Elfa’s sponsorship made that decision easy.

Instead of asking teams to wire together X scraping, Telegram monitoring, KOL scoring, and sentiment models under a 48-hour-adjacent deadline, Elfa gave every team direct API access to a social intelligence stack that was already live, scored, and mapped to real tokens.

What Elfa offered

Elfa provisioned free API credits to the first 200 teams that signed up for the hackathon, which ran from April 16 to June 16, 2026, with credits valid through June 30. Claiming credits was designed to be a seamless process, without the need for multiple layers of manual approvals:

1. Register at dev.elfa.ai and generate an API key
2. Upgrade to the Pay-As-You-Go tier — sponsored credits cover usage, so nothing is charged until a team's allotment is exhausted
3. Submit the hackathon claim form with team details
4. Credits load instantly onto the API key after verification

Under the time pressure of a hackathon, no team wants to wait on a manual approval cycle. We made sure that they could focus on building out their vision with minimal delays.

Mapping the API to the hackathon tracks

The hackathon organized submissions into five tracks, and Elfa's API and capabilities catered to each of the available tracks:

Under the hood, teams were pulling from the same endpoints Elfa's production partners rely on: top-mentions and keyword-mentions for social data, trending-tokens and trending-narratives for aggregation, smart-stats for account-level intelligence, and a six-mode /v2/chat endpoint (chat, macro, summary, tokenIntro, tokenAnalysis, accountAnalysis) for teams that wanted conversational or narrative output without building their own prompt layer.

The latest addition to the Elfa stack, Elfa Auto, launched just two days before the hackathon began. It takes a natural-language strategy from a builder and turns it into a structured monitoring or trading plan, delivered either via webhook or outbox polling. Elfa Auto also supports x402, an accountless, pay-per-call flow settled in USDC on Base, meaning that an agent could pay for its own data access without a human sitting in the loop managing API keys or invoices.

Battle-tested infrastructure and data

By the time Mantle launched the AI Awakening Hackathon, Elfa's data was already powering products with real users. CoinGecko surfaces Elfa’s insights across 10M+ monthly users, Mantle’s own Fluxion DEX embeds it directly in its trading terminal, and Virtuals' ACP agents call it in production.

In fact, just in the most recent hackathon Elfa co-sponsored with Pacifica, one of the winning entries, Clash of Perps — a Clash of Clans-style perps trading game, won the Social & Gamification track after integrating Elfa's API for trade suggestions, real-time insights, and social signals.

What's next

Since the hackathon, Elfa has launched Iris, an always-on intelligence stack that moves from batched updates to continuous, real-time monitoring across social sentiment, on-chain activity, and news events. Iris is connected through a knowledge graph that links related signals as they happen instead of surfacing them in isolation, tracing second-order effects before a downstream agent even starts reasoning.

For hackathon teams that kept building past the submission deadline, Iris is the natural next layer, the same real-time signal that powered their hackathon build, now with relationship-aware context attached.

Want to integrate Elfa into your app? Get started with the Elfa API or contact us.