The Prompt-Based Strategy Builder is Treeova's primary user-facing interface for creating an agent. Instead of writing code, the user types a plain-English description — "alert me when SPY's IV Rank crosses 50 and dark-pool short ratio is below 0.40" — and the platform compiles that prompt into a runnable agent chain with explicit modality, tools, and risk gates.

    Technology

    Prompt-Based Strategy Builder

    The Prompt-Based Strategy Builder is Treeova's primary user-facing interface for creating an agent. Instead of writing code, the user types a plain-English description — "alert me when SPY's IV Rank crosses 50 and dark-pool short ratio is below 0.40" — and the platform compiles that prompt into a runnable agent chain with explicit modality, tools, and risk gates.

    Quick definition

    Treeova's no-code interface for creating AI trading agents using natural language descriptions. Users describe their strategy in plain English, and the platform generates an executable agent.

    From sentence to runnable chain

    The builder routes the prompt through Agent Chain Generation, which selects an archetype, picks compatible tools, and validates the result against modality rules. The user sees the compiled chain before deploying — every node, every tool, every gate is visible and editable. There is no black box step between intent and execution.

    What gets enforced for the user

    Symbol pinning, modality separation, position sizing caps, and per-tool budgets are all enforced by the platform, not by the prompt. If the user writes "alert me about SPY" and the model accidentally proposes a tool call for QQQ, the executor rejects the call — the strategy the user described is the strategy that runs.

    Iteration and observability

    Every run writes to the agent's audit ledger so users can see exactly what their prompt produced. If the agent behaves unexpectedly, the user can edit the prompt and the rebuild happens in the same view — no redeploy, no waiting on engineers. The Agent Health Pulse surfaces per-agent state continuously so failed builds and stalled runs are visible at a glance.

    Related terms