Omni Intelligence

Omni Intelligence studies writing samples you choose to keep active, finds repeatable patterns, and turns that evidence into Writing Modes and practical guidance. Supported writing tools can use the current accepted guidance without silently replacing it when your sample library changes.

​

Open Omni Intelligence

Open the account menu and choose Omni Intelligence. From the Settings summary, choose Set up Omni Intelligence, Open Omni Intelligence, or View full report. The full Sample Drop, build, and accepted-report workspace opens at /intelligence inside Omni Writes.

​

Complete the three-step workflow

​

Add representative samples

Use examples that genuinely represent the writing you want Omni to follow. A mix of formats and situations is usually more useful than several near-identical posts. Links-only, hashtags-only, unfinished fragments, copied templates, or work that does not sound like you can weaken the result or fail validation. Valid samples autosave after a short pause. Before choosing Next, clear or complete every partially filled slot and wait until each valid sample is saved. If a save fails, your text stays in its slot. Use the sample’s Retry action, then wait until it is saved; Next remains unavailable while a valid sample is unsaved or saving. A too-short card or more than 10 active samples also keeps Next unavailable.

​

Manage saved samples

​

Follow the build safely

The Building screen reports real build progress rather than a decorative timer. If this is an update, your previous accepted Omni Intelligence remains active while the new build runs. A failed build or evidence conflict does not silently replace it. An initial failed build has no earlier accepted report to fall back to.

​

Read the accepted report

The shipped report is one accepted-summary surface rather than a set of technical analysis tabs. Open evidence, editing, and history from the report itself: use Preview samples for accepted excerpts, the guidance editor for do/avoid rules, Manage samples for the current library, and History for earlier accepted guidance.

• Your voice at a glance summarizes the first-pass findings in plain language.
• Voice Depth describes how much reliable accepted evidence Omni has: starting, developing, or rich.
• Insight cards explain recurring characteristics and where the guidance fits.
• Writing Modes represent distinct, natural facets of your writing. Open a mode when more than one is available.
• How Omni should write and What Omni should avoid show editable guidance for the selected mode.
• Preview samples show excerpts from the accepted snapshot, capped at 400 characters. Obvious email addresses, phone numbers, @handles, and links are replaced in the preview, but this display safeguard is not comprehensive anonymization.
• History lets you inspect earlier accepted snapshots and restore compatible guidance.
• Manage samples returns to Sample Drop without pretending that report guidance and sample evidence are the same thing.

​

Edit Writing Mode guidance

The shipped report lets you refine the selected Writing Mode’s ordered How Omni should write and What Omni should avoid rules. These instructions can steer supported Compose and Rewrite actions.

• Unsaved guidance stays in the local report draft across mode changes and in-page data refreshes. Save or discard it before reloading the browser or leaving the report.
• Omni warns before you leave with unsaved work and blocks actions such as returning to samples or restoring history until the draft is resolved.
• Guidance becomes read-only while an update is rebuilding the active Writing Modes.
• A save conflict from another tab preserves your local draft; refresh before trying to save again.
• After a rebuild, previous guidance is not copied into a new Writing Mode automatically. Choose Keep new guidance or Review previous guidance, then save deliberately.

​

Use History without changing the evidence

History keeps accepted snapshots rather than overwriting the current record in place. Restoring an earlier snapshot creates a new accepted restore point. History can restore saved names, descriptions, and rules for Writing Modes that are still active. It does not restore an old sample library, delete later history, or replace the current evidence. Guidance for a retired mode cannot be restored automatically.

​

Update after samples change

Adding, editing, excluding, restoring, or deleting a Writing Sample creates pending evidence. The current accepted report stays available and continues to guide supported features until you explicitly choose Update Omni Intelligence and the replacement succeeds. This separation prevents a half-finished sample edit or failed provider request from changing the guidance used by Write, Plan, Adapt, or Replies.

​

Understand report-build AI credits

A retry after a refunded failed run can create a new reservation. Use the existing Retry action rather than opening several new update attempts. See Billing and AI credits for current account allowances. The free first report build does not make later AI generation free. Each downstream action uses its own model-rated workload and credit receipt: Write in My Voice reserves one unit, 3 Variations reserves three units and settles only when three distinct usable results are returned, and Adapt by Network reserves one unit per requested destination with partial settlement or refund when only some destinations succeed. Plan and Replies follow their own workflow credit contracts.

​

Use accepted guidance across Omni Writes

Omni Intelligence is applied only when the supported workflow requests it and an accepted Writing Mode is available. Plan can choose a Writing Mode per request. Write in My Voice, 3 Variations, and Adapt by Network normally auto-match a mode to the current draft. Do not assume every AI action always uses Omni Intelligence: generic transformations can remain voice-neutral, and supported result proof can show that personalization was Applied, Unavailable, or Not requested.

​

Protect your data and review AI output

Building or updating Omni Intelligence analyzes the active samples included in that build and may process them with configured AI providers. Later writing requests generally send the draft or instruction, the selected or auto-matched Writing Mode, and bounded relevant excerpts from accepted Writing Samples to the selected provider or a configured fallback or repair provider—not every sample. The current Privacy Policy says Omni Writes does not train a general-purpose model on customer content. Account-specific suggestions can be derived from feedback and response signals, but proposed learning is not applied to the accepted Writing Mode until it is approved. Read the Privacy Policy and Terms of Service for the current provider, retention, deletion, and ownership terms.

​

Recover setup, build, or edit failures