Agent config
Workspace-level defaults for how myHERALD behaves. This is where you tell myHERALD things that apply across every draft: default tone, default posting cadence, topic rotation rules, languages you write in, off-limits topics, anything that should apply without being repeated in every brief.
You can edit these directly in the tab, or say them in chat (“Never repeat a topic within 30 days”) and myHERALD will save them here for you.
The workspace publication language is managed in Workspace settings. It is separate from your personal dashboard language under Account.
Scheduling
Set the timezone that defines calendar days, the maximum number of posts per platform per day, and the minimum gap between posts on the same platform. Defaults are one post per platform per day and a four-hour gap. You can override both values per platform, choose a Global English audience profile or one to three weighted audience regions, and add up to three preferred times for each platform.
Your preferred times are the highest-priority recommendation, while the capacity and gap rules still apply. If you have enough measured posts, myHERALD uses your own workspace performance next. Otherwise it uses a dated, linked Buffer benchmark for LinkedIn, X, Instagram, Facebook, Threads, or the matching YouTube format. Bluesky uses an explicitly labeled proxy combining the X and Threads patterns. Reddit uses audience-timezone overlap because no equivalent validated benchmark is registered.
Scheduling is managed only through this validated section. Older generic posting-time and timezone entries are hidden from Agent config and ignored, so two settings cannot contradict each other. You can also make the same validated changes in chat, for example: “Allow two LinkedIn posts per day” or “Target London and New York equally.”
Publication and export destinations such as WordPress, Ghost, Medium, Google Docs, and webhooks have no automatic platform-wide best time. Add a preferred time if you want myHERALD to schedule them independently. For a blog-led topic, myHERALD can instead derive an earlier blog time from the retained social recommendation. YouTube Shorts and long-form videos use separate evidence, and myHERALD does not guess when the format is unknown.
Default image model
Choose which engine myHERALD uses when it generates images for your content. The default is Gemini. If you have enabled an alternative provider (such as the OpenAI image model) via its sub-processor notice, you can select it here and it becomes the workspace default for all new image generation requests.
Brand logo badge
When a logo is set for your workspace (see Media settings), myHERALD can overlay it as a small badge on generated images before publishing. Controls in this section let you decide:
- Which content gets a badge. Choose All images, or scope it by tag (for example, only images tagged “Thought leadership”, or all images except those tagged “Meme”).
- Badge corner. Bottom-right, bottom-left, top-right, or top-left. A thumbnail preview updates as you change the position.
The badge is composited at export time, so changing the corner setting affects future exports only, not images already published.
Onboarding researcher model
During the onboarding research phase, myHERALD sends a researcher to learn about your company, audience, and competitive landscape. You can choose which model that researcher uses:
- Haiku (default): fastest and cheapest; good for most companies.
- Sonnet: more thorough; recommended if your domain is niche or technical.
- Opus: deepest research; use when nuance and completeness matter more than speed.
- Same as researcher: inherits whatever model is set as the workspace researcher default.
This selector applies only to the onboarding run. Re-running onboarding uses the model set here at that time.
Autopilot defaults
Planning cadence, platform targets, and content-type ratios that Autopilot uses when no campaign-level override is set. See Autopilot for a full walkthrough of how autonomous planning works.