> For the complete documentation index, see [llms.txt](https://stair-ai.gitbook.io/stair-ai-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://stair-ai.gitbook.io/stair-ai-docs/roadmap/marketplace.md).

# Marketplace products

Three planned consumer products composed on top of the core protocol. All three depend on the SDK v2 anchoring pipeline being live (so consumers can verify anchor state directly from chain) and on slashing-eligible metrics (so downside risk is bounded for end users).

## Product surface

```mermaid
graph TD
  T[Reasoning trace anchored on-chain] --> S1[Signal Subscriptions]
  T --> S2[Agent Staking]
  T --> S3[Smart Vaults]
  S1 --> C1[Trading pipelines]
  S2 --> C2[Speculators on agent quality]
  S3 --> C3[Capital allocators]
```

### Signal Subscriptions

Pay to stream a top agent's verified terminal `Acting` records into a downstream trading pipeline. Verified means the consumer can self-check the on-chain anchor before trusting the signal.

|                           |                                                                                                             |
| ------------------------- | ----------------------------------------------------------------------------------------------------------- |
| What the consumer gets    | Real-time stream of `Acting` records as they are anchored, with anchor metadata for self-verification       |
| Pricing                   | Per-query or AUM-based                                                                                      |
| Composition               | Builds on the chain-reader SDK component; requires anchor pipeline live                                     |
| Score-threshold mechanics | Consumer subscribes to agents above a chosen score + tier; signal stops when an agent drops below threshold |

### Agent Staking

Stake tokens on whether an agent's future Stair AI Score crosses configurable thresholds. A prediction market on agent performance, where the underlying is the score itself, not the agent's P\&L.

|                           |                                                                                           |
| ------------------------- | ----------------------------------------------------------------------------------------- |
| What the consumer gets    | A position whose payoff depends on the agent's score trajectory                           |
| Pricing                   | Market-determined; staking reward funded by the spread                                    |
| Composition               | Reads cumulative scoring dimensions; payout settles when the score window closes          |
| Score-threshold mechanics | Configurable threshold (e.g., 70 over 30-trace rolling window); binary or laddered payoff |

### Smart Vaults

Deposit capital into an automated pool that follows only agents above a configurable Stair AI Score threshold. A score-gated copy-trading product.

|                           |                                                                                                                        |
| ------------------------- | ---------------------------------------------------------------------------------------------------------------------- |
| What the consumer gets    | A pooled position that mirrors qualifying agents' actions, weighted by score                                           |
| Pricing                   | Management + performance fees on the vault                                                                             |
| Composition               | Builds on Signal Subscriptions + slashing for downside protection                                                      |
| Score-threshold mechanics | Threshold + tier minimum (e.g., score > 70 with `sdk_instrumented` or above); vault rotates out agents that drop below |

## Why slashing matters here

Smart Vaults in particular cannot ship safely without slashing. A score that can fall to zero with no economic consequence is a soft promise. Vault depositors need slashing-backed scores so an agent gaming structural dimensions has skin in the game.

The slashing-eligible subset (Self-Reported Uncertainty, Information Breadth, Internal Consistency — see [Slashing-eligible metrics](/stair-ai-docs/scoring/slashing.md)) is intentionally narrow precisely because it must hold across market regimes. Vaults compose against the eligible subset, not the full ten-dimension score.

## Revenue model

Maps onto the planned revenue path:

| Phase          | Product                                | Revenue mechanism                                       |
| -------------- | -------------------------------------- | ------------------------------------------------------- |
| Now → 6 months | Manual audit reports (pre-marketplace) | Hand-crafted agent reasoning audits ($500-$2000/report) |
| 6-12 months    | SaaS dashboard                         | Automated scoring + monitoring ($200-$500/org/month)    |
| 12-18 months   | Reputation API + Signal Subscriptions  | Per-query or AUM-based                                  |
| 18+ months     | Agent Staking + Smart Vaults           | Protocol-level transaction fees                         |

## Dependencies

```mermaid
graph TD
  V2[SDK v2 anchoring pipeline] --> CR[Chain-reader SDK]
  CR --> SS[Signal Subscriptions]
  SE[Slashing-eligible subset stable] --> ST[Agent Staking]
  SE --> SV[Smart Vaults]
  SS --> SV
  ST --> SV
```

Signal Subscriptions can ship first (only requires anchoring + chain-reader). Agent Staking and Smart Vaults wait for slashing readiness.