Token intelligence, on-chain forensics, prediction brokers, sentiment agents, and specialised inference providers. The category that supplies the rest of the agent economy with the signals it acts on.
An Intelligence & Analytics Agent is an autonomous software agent whose primary function is to analyse on-chain and off-chain data and produce signals, insights, predictions, or labels — without itself transacting financially. Its outputs are consumed by humans or by other agents that take action.
Most Financial Agents in the AHM Taxonomy depend on at least one I&A agent in their decision pipeline. The Olas Mech Marketplace pattern — where a trader agent commissions a prediction from a separate Mech service before placing a bet — is the cleanest worked example of this dependency. I&A is the supply side of the agent economy's decision loop.
In the Phase 2.2 classification run, Intelligence & Analytics Agents were anchored on a small set of on-chain contracts: Ritual Infernet + EIP712Coordinator (decentralised inference network), Olas Mech Marketplace (agent-to-agent prediction and inference services), and CARV (data layer for autonomous AI). An agent qualified as I&A if its transaction history showed direct interaction with one or more of these. The methodology is documented in the POC summary; subsequent classification phases are expected to expand the anchor set.
Narrative detection, KOL tracking, alpha signals. Often consumer-facing with token-gated terminal access. The aixbt-style category.
Entity attribution, fund-flow tracing, address labelling at scale. Underpins both human investigation and downstream agent decisions.
Paid inference services for other agents. The Olas Mech model: agents pay for predictions and use them as input to their own decisions.
Agents extracting sentiment from social platforms, on-chain discussion, and community signals. Bridges the off-chain narrative layer with on-chain decision-making.
Domain-specific ML inference offered as a service: probability estimation, classification, scoring. Adjacent to Mechs but operating across protocols rather than within one marketplace.
The agents below are well-known illustrations of the I&A category as defined above. They are not necessarily the agents identified in the Phase 2.2 classified sample — that classification was anchored on the on-chain contracts listed in Definition & scope. Olas Mech Marketplace is both an anchor contract and the infrastructure underneath one of the examples (Eolas_AI), which is a useful coincidence. The other examples illustrate adjacent patterns within the broader category.
Reportedly 99% autonomous narrative-detection agent on Virtuals, monitoring 400+ key opinion leaders and on-chain signals to surface alpha. Built an audience of 460k+ X followers and a token-gated terminal accessible to large holders. The canonical example of an I&A agent operating as a public-facing market intelligence service rather than as a backend feed.
Agentic on-chain intelligence platform built on a database of 500M+ labelled addresses. Exposes its data as both a traditional API and via x402 micropayments, and ships an MCP server for direct consumption by AI tools including Claude and Cursor. Representative of the multi-channel distribution pattern: the same intelligence layer served to humans, to AI co-pilots, and to autonomous agents.
CreatorBid agent built on Olas Mech infrastructure that delivers on-demand predictions via X mentions. Behind the user-facing chat is a fully agent-native flow: payment, task delegation, and prediction delivery all happen between agents on-chain. A worked example of how prediction brokering can wear a consumer-friendly skin while running on agent-to-agent rails underneath.
A class of agents rather than a single agent: the prediction brokers that sit on Olas's Mech Marketplace and answer probability requests from trader agents like Polystrat and Omenstrat. Bridges Intelligence & Analytics with Financial — the I&A agent never trades, but its output drives most trades downstream. Surfaces taxonomic dependency cleanly: Financial Agents depend on I&A Agents as a routine matter of operation.
Machine-learning infrastructure layer providing real-time predictive inference accessible to other agents. Distinct from Mechs in that it is protocol-level rather than marketplace-level — agents integrate Allora as a dependency rather than commissioning individual inferences via on-chain calls. Representative of the shift from per-request inference markets toward always-on inference infrastructure.
Decision support for autonomous trading. The bulk of I&A activity by transaction volume sits in the supply chain feeding Financial Agents. Trader agents call prediction brokers; portfolio agents call entity-labelling services; arbitrage agents call sentiment monitors. The Olas Mech Marketplace pattern — where this commissioning happens via on-chain payments and IPFS-stored deliverables — is the most legible operating model.
Public-facing market intelligence. Agents like aixbt have shown that an autonomous I&A operator can build and sustain a meaningful audience on its own, separate from its role as a backend signal source. The value capture is via token-gated access to deeper outputs rather than per-call payment. As of Q1 2026 this is one of the more commercially mature consumer-facing patterns in the agent economy.
Multi-channel distribution of intelligence. Established analytics providers (Nansen, Arkham) have begun shipping the same intelligence through three channels in parallel: traditional human-facing dashboards, AI-co-pilot integrations via MCP, and per-call agent-facing APIs over x402. This three-track distribution is becoming the default for any I&A operator with prior infrastructure and a credible data moat.
Composable inference. The longer-term shift is from per-request prediction markets toward always-on inference infrastructure that other agents integrate as a dependency. An I&A agent specialising in (say) liquidation-risk scoring becomes a building block for any Financial Agent that wants to size its positions defensively. Composability is the architectural commitment that will determine whether I&A scales as a category or stays a craft.
Composite 0–100 health score blending solvency, behavioural consistency, and operational stability. Calibrated for I&A baselines.
Full diagnostic report covering operational patterns, output cadence, and cross-referenced on-chain activity for I&A operators.
Smart money labels via Nansen, PnL enrichment, and consistency checks between published signals and observed flow.
Surface the downstream Financial Agents and upstream data sources that depend on this I&A agent's outputs.
@misc{ahm_taxonomy_intelligence_analytics_2026,
title = {Intelligence \& Analytics Agents --- AHM Taxonomy v1},
author = {{Agent Health Monitor}},
year = {2026},
month = {May},
url = {https://intelligence.agenthealthmonitor.xyz/taxonomy/intelligence-analytics},
note = {Accessed: 2026-05-07}
}
Classification methodology and category boundaries are documented in the AHM Taxonomy v1 POC summary.
Counts reflect the Phase 2.2 classification run (25 April 2026): a 1,000-agent random sample from Base mainnet wallets, of which 757 (75.7%) were classified across 6 anchored categories. Updated as new classification phases complete.