Pear Protocol powers Agent Pear's triggers with Elfa

Pear Protocol is a pair trading platform with risk management built into the position. A trader opens a long position on one asset and a short position on another, held as a single position, with the outcome depending on how the two assets perform relative to each other. A long SOL, short ETH position pays out when SOL outperforms ETH, whether both assets rise or both fall. Institutional desks have traded this structure in equities for decades. Pear brought it on-chain, where traders have since run $1.82B in cumulative volume.

Why monitoring is the hard part of pair trading

A pair trade rests on a relationship between two assets holding, and that relationship can break for reasons specific to either side. A large holder unwinding one leg, or an account changing its stance in public, moves one side of the position without touching the other and invalidates the original thesis.

Tracking that means watching two assets and the narratives around both, in a market that runs continuously.

Agent Pear is Pear's answer. It works as a quant and narrative analyser, answering questions about assets and pairs, and runs automated triggers that notify users when defined conditions occur. The design problem behind a system like that is deciding what counts as a condition worth acting on. Price levels are simple to define, but complex conditions require more than price data, so Pear integrated Elfa’s Iris stack to complement their existing internal statistical library.

How Agent Pear integrates Elfa

The integration works through tool calls. A user asks Agent Pear about a specific token, the agent queries Elfa, and returns with sentiment scores, recent news, and the social context surrounding that asset.

Pear processes each user's history with Agent Pear, past conversations, questions asked, positions discussed, into a memory layer, and its own algorithm decides what each user's triggers should watch. Pear's internal statistical engine measures the market side of those conditions against live exchange data, and Elfa supplies the narrative side, the always-on watching of news, sentiment and social conversation that no price feed can express. The result is personalised alerts shaped by what each user has been researching and trading.

Elfa maintains proprietary context on every asset it tracks, drawing on its own data pipeline alongside external sources including CoinGecko. Elfa assembles that context before the query arrives, so response times hold steady whether a user asks about ETH or a fresh new token listed last week.

Multiple trigger types on one signal layer

Agent Pear's trigger system runs on Elfa data across multiple categories, applied against what each user is tracking.

Price triggers fire when an asset crosses a defined level, and technical triggers fire on indicator conditions. Sentiment triggers watch the sentiment around an asset and fire when it moves. Social triggers, read the conversations: a smart account with a track record making a call, mention volume climbing on an asset, a narrative forming around one leg of a pair.

Social and sentiment triggers involve identifying which accounts matter, verifying them, scoring their history, and detecting narrative formation as it happens.

Iris

We built Iris as an always-on intelligence stack. It monitors social sentiment, on-chain activity, and news events as they happen, connects related signals through a knowledge graph, and traces the second-order effects of an event before an agent queries it.

That is what Pear plugged into. Building sentiment monitoring, account scoring, and narrative detection in-house would have meant reconstructing years of data pipeline, and Iris gave them a framework to build on. Elfa triggers now run as a core part of Pear's agentic system, firing against each user's own thesis rather than a generic watchlist.

The two systems sit side by side. Agent Pear holds a knowledge base of what each user is pursuing, while Iris monitors what is happening in the market. Agent Pear asks whether an event affecting one asset should change a position held in two, and Iris supplies the connections that answer it.

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