Private Bundles vs the Public Mempool

Answer first: Public submission broadcasts a transaction through the network's normal propagation path. A private transaction endpoint limits where a signed transaction is initially sent. A bundle adds provider-specific ordering or atomic execution semantics. Private submission can reduce information exposure, but receipt, simulation, and submission do not make inclusion certain or a favorable result.

Four concepts that are often mixed together

The word "private" describes an initial submission path. It is not a universal statement about every participant, log, fallback, or later on-chain observation.

Public mempool submission

On Ethereum, a normal transaction can be propagated among peers and observed in pending-transaction feeds before it is included. This visibility lets wallets, explorers, builders, searchers, and other systems react to pending state.

Public submission is still appropriate for many ordinary transactions. The important question is whether the transaction contains information that creates adverse-selection or ordering risk before inclusion.

Private Ethereum transactions

Flashbots Protect is an example of a private Ethereum transaction endpoint. The user sends a signed transaction to the Protect endpoint instead of starting with a normal public broadcast.

This interface is not the same as a searcher bundle. Users should read the provider's current documentation for cancellation, fallback, privacy, refund, and status behavior. A wallet or application integration can also change how the route is used.

Flashbots bundles

Flashbots documents bundles as ordered groups of signed transactions submitted for a target block. A bundle can express execution relationships that a single transaction cannot.

Bundle validity and inclusion depend on the target state, transaction validity, builder access, competition, and the economic value of the candidate block. A successful simulation or endpoint response should not be presented as proof of final inclusion.

Jito bundles

Jito's bundle interface is designed for Solana and follows Solana's execution and leader model. Its documentation describes sequential and atomic processing within one slot. The API can return a bundle identifier after receiving a request, while separately warning that this does not guarantee processing or landing.

A material exception is documented in Jito's uncled-block guidance: transactions observed in a skipped or uncled block can be rebroadcast individually. Normal processing of those transactions does not preserve the bundle's atomicity or reversion protection. Review the provider's state-check safeguards rather than assuming bundle rules cover every rebroadcast path.

The similarity in names does not make a Jito bundle interchangeable with an Ethereum builder bundle.

Cost and failure language

Statements such as "failed bundles cost nothing" are too broad. A bundle that is never included may avoid the on-chain execution cost associated with an included transaction, but operators can still face provider charges, infrastructure costs, opportunity costs, tips or fees associated with a successful path, and losses from configuration or state assumptions.

The exact cost model must come from the provider and the final on-chain receipt, not a universal rule.

Where FRB fits

FRB Agent is a local execution assistant. It can coordinate user-configured simulation, limits, signing, and submission routes, but it does not operate the underlying builder, block engine, validator, or RPC network.

Use Private Bundle Workflows for the product boundary and Simulation Mode for the pre-execution review layer.

Decision checklist

1. Identify the network's real ordering and submission model.
2. Decide whether the workflow needs one private transaction or an ordered bundle.
3. Confirm provider privacy, fallback, cancellation, status, and retention behavior.
4. Simulate the same transaction shape and permissions intended for live use.
5. Set explicit fee, slippage, spending, and contract limits.
6. Treat acceptance as receipt, then verify the final chain result independently.

Methodology and limitations

This article uses Flashbots and Jito documentation for interface behavior. It does not compare measured inclusion rates, latency, profitability, or provider quality.

Next step: Review the Private Bundle Workflows documentation before selecting a submission path.