DreamerOS
Integrity Transparency

See what DreamerOS recorded.

DreamerOS records available integrity, engine-choice, and prompt-shaping events. The record you see depends on the path and the connected client.

This is what sets DreamerOS apart from raw AI. You can see the recorded checks and decisions that apply to your response instead of guessing what happened between question and answer.

Why this matters: when an AI answer turns out to be wrong, "the model said so" is not a real answer. Available records can show the engine, the prompt-shaping event, and the checks or verification fields that ran. You can use that record with a colleague, regulator, or your future self.

What gets logged

Here are the event types the audit record can show. Availability depends on tier and path.

Integrity Pre-checks

Before the engine responds

  • Intent and input checks
  • Available risk signals
  • Check status for that path
Checks after the answer

After the engine responds

  • Available output review results
  • Drift signals that the path recorded
  • Flagged follow-up information
Prompt Shaping

Prompt restructuring

  • Original prompt vs restructured prompt
  • Intent detection
  • Domain classification
Engine Choice

Model selection

  • Which engine was chosen
  • Why that engine was selected
  • Engine-choice confidence score
Verification

Claim checking

  • Temporal claims flagged
  • Live source lookups
  • Verification status per claim
Receipt Record

Append-only when created

When DreamerOS creates a receipt, database triggers reject changes or deletion. Other audit fields depend on the path that recorded them.

What the audit record shows

The app shows an audit record to every signed-in plan. The fields differ by plan and path. The complete customer walk on every tier remains In Beta 2.0. Holding back a flagged answer and removing personal data are plan extras: see what each plan includes.

Every plan

Checks, prompt and engine details

  • Every plan includes the pre-check before an answer
  • Every plan includes the checks after the answer
  • Available receipt and audit metadata
  • Pass or flag information from that path
  • Available prompt and engine-choice details
  • Available verification results
  • Fields vary by the path that produced the record
More detail

Fuller audit access

  • More audit fields, when the path records them
  • Available cost and engine-choice metadata
  • Available verification source details
  • Trace fields recorded by that path

Why this matters

Most AI tools are black boxes. You type something. You get a response. You have no idea what happened in between. DreamerOS makes it visible.

1

Was your prompt restructured? Where the path records it, see prompt-shaping information and why the system changed the request.

2

Did the integrity layer flag anything? Every answer gets a check before it and checks after it, when that path records them.

3

Which engine responded and why? Where an engine choice applies, the audit record can show the selected engine and recorded decision fields.

4

Was any claim verified against live sources? When verification runs, its recorded status can show whether a source check took place.

5

What confidence information is available? Review the confidence or flag fields that the path recorded for that response.

Integrity principles

These are examples of the principles used by available checks. The exact set depends on the path and tier.

Accuracy over completeness

Say less if it means saying it right. Never pad a response to seem thorough.

No false guarantees

If the system is not certain, it says so. No confident-sounding guesses.

Surface hidden assumptions

Detect what the user assumed and expose it before answering.

State uncertainty

When confidence is low, say so explicitly. Never hide doubt behind fluent prose.

Depth matching

Match the depth of the response to the depth of the question. Simple questions get simple answers.

Third Eye

Intent, Integrity, Intuition. Available paths can apply these three lenses before processing begins.

Input as raw signal

Input shaping can treat a request as raw signal to clarify, not as a finished instruction to follow blindly.

Showing seven examples. Check availability and recorded fields vary by path and tier.

Append-only guarantee

Your audit trail cannot be edited

Log entries can never be modified or deleted. This is an architectural constraint, not a policy.

Database triggers reject any UPDATE or DELETE on the audit log at the database layer. Once an integrity event is recorded, it exists permanently. These records hold no message text. If you delete your account, your memories, conversations and files are erased; the event records stay as a ledger with no text in them.

The rejection happens inside the database itself, not in application code - there is no code path that can modify a written row. Your audit trail cannot be rewritten, redacted, or erased.

See exactly what happened between your question and the answer

See what the path recorded. Start free. No credit card required.

Integrity you can see Try it now