Iris is a real-time intelligence stack from Elfa AI that gives AI trading agents live context on financial markets, starting with crypto. It combines social chatter, on-chain activity, and market data into a single signal layer that agents can query or subscribe to, so they can act on sentiment shifts, narrative moves, and early insights the moment they form rather than after the fact. The launch matters because Elfa AI is rolling out a platform that lets AI agents listen continuously to market conditions and execute transactions on Solana in real time, triggering automated actions the moment specified conditions are met , which pushes agent tooling past static prompts into event-driven execution.
Iris ships alongside Alpha, a consumer-facing agent that sits on top of the stack. Traders can ask Alpha open questions about what is happening in a token or narrative right now, or hand it a set of conditions and let it watch, flag, and trade on signals in the background. Developers get the same primitives through an API, so any agent framework can plug in for surveillance, entity linking (who is involved, how they are connected, what it led to), and execution triggers without building a data pipeline from scratch. Elfa indexes price and trading-related data, social chatter, news, and prediction markets continuously, and surfaces structured insights the moment something relevant happens.
The company is headquartered in Singapore and building a real-time context layer for agents , led by co-founder and CEO Tristan Teo. It has already been embedded across the crypto data stack, powering a share of CoinGecko's altcoin and macro insights and integrating with builders on Virtuals and Drift, and its backers include Amber, along with angels such as CoinGecko co-founder Bobby Ong and Etherscan founder Matthew Tan. Iris is the productized version of that infrastructure, aimed at teams who want their agents to trade on current reality instead of stale training data.
Comments (11)
Sign in to join the discussion.
P
Priya BalakrishnanJul 28, 2026
"Supercharge Agent Awareness" reads like a mindfulness app for chatbots. Try: "Give your agents a nervous system."
T
Tomasz KowalczykJul 28, 2026
The cut from black to the Iris logo at 0:07 is clean but that voiceover is doing a LOT of heavy lifting. Whoever wrote the script owes the editor a beer.
R
Ravi DeshmukhJul 28, 2026
Demo looks great but I need to know if that latency was real or a montage. "Real-time" is doing philosophical work in this launch copy.
A
Ade OkonkwoJul 28, 2026
Cool cool cool but where are the docs. Webhook payloads? Rate limits per condition? Can I subscribe to 10k triggers without getting throttled into oblivion?
H
Hiroshi N.Jul 28, 2026
We had something spiritually similar at my old shop in 2019, called it "signals fabric". Killed by reorg. Rooting for you to not get reorged.
M
Milos VranicJul 28, 2026
The real-time context market is a feature, not a company. Every agent framework will ship this by Q2 and eat your lunch.
B
Bjorn HaraldsenJul 28, 2026
Reminds me of what one of my portfolio teams is doing on the retrieval side, different wedge though. The trigger primitive is the interesting bet here.
F
Fatima ZahraJul 28, 2026
The "Iris" wordmark has beautiful spacing but that gradient on the eye motif is fighting the background. Two competing focal points on the hero shot.
L
Lena OseiJul 28, 2026
ok wait, "set conditions, get triggered, act" is actually a killer three-word product loop. that's the whole tweet right there.
S
Sunni AroJul 28, 2026
Third "context layer for agents" launch this month and the pace is accelerating. Feels like everyone read the same memo in October.
C
Camila RestrepoJul 28, 2026
Landing page is gorgeous but I clicked "Get Started" and got a waitlist. In 2024 that's a vibe killer, ship the sandbox.
Work With Us
Book an intro call to explore our suite of services. We might have a waitlist.
Elfa AI
Supercharge Agent Awareness
Supercharge Agent Awareness
Tags
TLVC Rating
Community Rating
No ratings yet
to rate this launch.
Iris is a real-time intelligence stack from Elfa AI that gives AI trading agents live context on financial markets, starting with crypto. It combines social chatter, on-chain activity, and market data into a single signal layer that agents can query or subscribe to, so they can act on sentiment shifts, narrative moves, and early insights the moment they form rather than after the fact. The launch matters because Elfa AI is rolling out a platform that lets AI agents listen continuously to market conditions and execute transactions on Solana in real time, triggering automated actions the moment specified conditions are met , which pushes agent tooling past static prompts into event-driven execution. Iris ships alongside Alpha, a consumer-facing agent that sits on top of the stack. Traders can ask Alpha open questions about what is happening in a token or narrative right now, or hand it a set of conditions and let it watch, flag, and trade on signals in the background. Developers get the same primitives through an API, so any agent framework can plug in for surveillance, entity linking (who is involved, how they are connected, what it led to), and execution triggers without building a data pipeline from scratch. Elfa indexes price and trading-related data, social chatter, news, and prediction markets continuously, and surfaces structured insights the moment something relevant happens. The company is headquartered in Singapore and building a real-time context layer for agents , led by co-founder and CEO Tristan Teo. It has already been embedded across the crypto data stack, powering a share of CoinGecko's altcoin and macro insights and integrating with builders on Virtuals and Drift, and its backers include Amber, along with angels such as CoinGecko co-founder Bobby Ong and Etherscan founder Matthew Tan. Iris is the productized version of that infrastructure, aimed at teams who want their agents to trade on current reality instead of stale training data.
Comments (11)
"Supercharge Agent Awareness" reads like a mindfulness app for chatbots. Try: "Give your agents a nervous system."
The cut from black to the Iris logo at 0:07 is clean but that voiceover is doing a LOT of heavy lifting. Whoever wrote the script owes the editor a beer.
Demo looks great but I need to know if that latency was real or a montage. "Real-time" is doing philosophical work in this launch copy.
Cool cool cool but where are the docs. Webhook payloads? Rate limits per condition? Can I subscribe to 10k triggers without getting throttled into oblivion?
We had something spiritually similar at my old shop in 2019, called it "signals fabric". Killed by reorg. Rooting for you to not get reorged.
The real-time context market is a feature, not a company. Every agent framework will ship this by Q2 and eat your lunch.
Reminds me of what one of my portfolio teams is doing on the retrieval side, different wedge though. The trigger primitive is the interesting bet here.
The "Iris" wordmark has beautiful spacing but that gradient on the eye motif is fighting the background. Two competing focal points on the hero shot.
ok wait, "set conditions, get triggered, act" is actually a killer three-word product loop. that's the whole tweet right there.
Third "context layer for agents" launch this month and the pace is accelerating. Feels like everyone read the same memo in October.
Landing page is gorgeous but I clicked "Get Started" and got a waitlist. In 2024 that's a vibe killer, ship the sandbox.