Policy Intelligence
Auto-Dynamic PIL
Premium adaptive policy assistance. Auto-Dynamic PIL can tune a bounded set of execution parameters for the next transaction in the same market: deterministically, transparently, and without touching your durable policy defaults.
What Auto-Dynamic PIL is
Auto-Dynamic PIL (Policy Intelligence Layer) is an opt-in premium feature that analyzes same-market signals and produces a bounded, next-transaction policy overlay. The adjustment is:
- Same-market scoped: signals and adjustments are specific to the market pair in the pending transaction.
- Next-transaction only: the overlay expires after one use. It is not persisted.
- Bounded and deterministic: every adjustment is capped by hard policy limits. The system does not guess freely.
- Explainable: each adjustment produces receipt evidence so you can inspect what was applied and why.
The three modes
Auto-Dynamic PIL has three operating modes selectable from the Policy Controls screen:
| Mode | Behavior |
|---|---|
| Off | PIL is fully disabled. No adaptive changes are computed or applied. Your durable policy defaults govern every transaction without modification. |
| Recommend | PIL computes bounded recommendations and surfaces them in the Policy Controls UI. Nothing is applied automatically. You review and accept or dismiss each suggestion. |
| Auto (bounded) | PIL computes and automatically applies the bounded next-transaction overlay without requiring manual approval. The overlay is still constrained by hard caps and expires after the next transaction in that market. |
What it can change (v1)
In v1, Auto-Dynamic PIL can adjust one field:
max_slippage_bpsthe maximum tolerated slippage for a DEX swap, in basis points. PIL may propose a tighter or wider bound based on same-market volatility signals.
The set of eligible fields expands over time as confidence and verifiability increase. Additional fields will be documented here as they become available.
What it does not do
Auto-Dynamic PIL has hard boundaries. It does not:
- Rewrite trust boundaries: it cannot expand your DEX allowlist, modify your allowed protocols, or change spend limits beyond the pre-configured range.
- Mutate allow or deny lists: token and program allow/deny lists are durable policy. PIL does not touch them.
- Override your explicit overrides : if you have manually set a value in Policy Controls, PIL will not silently supersede it beyond bounded deltas.
- Persist changes: no adaptive adjustment is written into your durable policy. Every overlay is ephemeral.
Durable policy vs adaptive overlay
Understanding this distinction is key to using Auto-Dynamic PIL safely.
Durable Policy
Your standing defaults and overrides. Configured in Policy Controls. These are the permanent rules that govern every transaction unless an adaptive overlay is active.
- Survives across all transactions
- Explicit operator and user intent
- What you set is what enforces
Adaptive Overlay
A bounded, next-transaction adjustment computed by PIL. Not written to durable policy. Expires after one use in the same market.
- Single-use, then discarded
- Bounded by hard caps
- Produces a receipt event for auditability
A pending overlay is an adaptive adjustment that has been computed and queued but not yet consumed. You can see pending overlays in the Policy Controls adaptive status grid. Once the next qualifying transaction executes, the overlay is applied and then discarded.
Safety rails
Every Auto-Dynamic PIL adjustment operates within multiple layers of safety constraints:
- Bounded delta size: each eligible field has a maximum allowed change magnitude. PIL cannot push a value beyond that delta, regardless of market signals.
- Hard caps: absolute policy ceilings enforced by ATF. An adaptive overlay can never result in a value that exceeds the configured hard cap for that field.
- Same-market scope: signals are scoped to the specific market pair. An adjustment for SOL/USDC does not influence ETH/USDC behavior.
- Next-transaction TTL (time-to-live) Overlays expire after one transaction. They are not cumulative and do not carry forward.
- Explainability and receipts: every applied overlay generates a receipt event. You can inspect the pre-overlay value, the applied adjustment, and the reason at any time.
How to use it safely
When to leave it Off
If you have strict compliance requirements, prefer full operator control, or are running in a context where every policy change needs human sign-off, leave PIL off. Your durable policy governs entirely.
When Recommend mode is best
Use Recommend when you want the benefit of market-signal intelligence but want to review every suggestion before it applies. Good for learning the system or for contexts where you are actively monitoring trade outcomes.
When Auto (bounded) makes sense
Use Auto (bounded) when you trust the bounded behavior, want frictionless next-transaction adaptation, and are comfortable with the hard caps in place. Suitable for automated agent workflows where real-time human review is not practical. Review your receipt events periodically to verify behavior.
How to verify what happened
Adaptive events are surfaced in two places:
- Policy Controls status grid: shows the latest adaptive event name and timestamp directly on the Policy Controls page.
- Receipt evidence: each applied overlay appends a structured event to the transaction receipt. The event includes the field adjusted, the pre-overlay value, the applied value, and the PIL signal reason.
If an adjustment was applied and you want to understand why, locate the corresponding receipt and inspect the adaptive_pil field, which contains applied_event and recommendation_event records for full auditability.
Availability
Auto-Dynamic PIL is available on Pro and Enterprise plans. Free-tier accounts can view the adaptive section and access this documentation, but the mode selectors will remain locked until upgrading.