AI AGENT CONSENSUS

How 100 AI agents reach one verdict

Agent Swarm is Kabra's consensus engine — 100 independent AI agents across seven expertise tiers, each voting on every trade idea and reporting its reasoning. Here's how it works and why it differs from a single-model verdict.

How a Swarm run happens

  1. 01

    Dispatch

    When you query a ticker, Kabra spawns 100 agents in parallel across seven expertise tiers — fundamentals, technicals, momentum, options flow, sector rotation, macro, and sentiment.

  2. 02

    Independent analysis

    Each agent analyzes the trade from its specialist lens. Agents don't coordinate — independence is the feature. You want votes that reflect different angles, not a single viewpoint reinforced 100 times.

  3. 03

    Vote casting

    Every agent casts a vote — BUY, SELL, or HOLD — and reports its reasoning. The reasoning is stored so you can audit any single vote later.

  4. 04

    Aggregation

    Votes are aggregated into a single verdict. The confidence score is the proportion of agents voting for the winning direction. 92% BUY means the swarm was nearly unanimous; 54% BUY means the trade is genuinely contested.

  5. 05

    Delivery

    You receive one direction, one confidence score, and the per-agent reasoning tree. Size your position to the confidence — larger on high conviction, smaller on contested reads.

Why agent consensus beats a single model

Reduces single-model risk

A single AI can be confidently wrong. Independent agents surface disagreement — if the swarm is split, you know the trade is a genuinely hard call, not an obvious setup.

Produces a conviction signal, not just a direction

Traditional AI outputs give you BUY or SELL. Agent consensus adds a confidence distribution — the input most traders are actually missing when sizing positions.

Every vote is auditable

No black-box scores. Expand any agent to see the factors it weighted, the data it cited, and why it voted the way it did. You can accept or reject on the merits.

Fast enough for trade-timing

The full 100-agent swarm returns a verdict in under a minute. That matches an active-trading horizon, not a multi-hour research horizon.

The seven expertise tiers

Agent tiers cover fundamentals, technicals, momentum, options flow, sector rotation, macro context, and sentiment. Every trade idea gets a vote from each tier in parallel — so you see agreement or disagreement across very different analytical lenses.

Run the Swarm on any ticker and watch 100 agents reach a verdict in under a minute.

Stay Ahead of the Market

Get platform updates, new analytical capabilities, and market intelligence briefings from the Kabra research desk