Overview
Setup asks two things: which markets the agent works in, and which competitors ground its research. Both are editable afterwards.
The last step ends with Choose and start agent, which saves the configuration and starts the first run. Metadata and keywords always run; there is no permissions step.
Markets
Setting the agent up for an app starts with the markets it will work on. A market is one storefront and one language.
Filter the list with Search storefronts. Markets your listing already has appear under Localized, and the rest under Not localized; your listing's main market is marked Primary. Continue stays at the bottom of the setup pane while you scroll the list.
The agent can open App Store localizations your listing does not have yet and draft for them. The first time you pick one it asks you to confirm, because opening a localization is a change to your App Store listing rather than something Spek keeps to itself.
You can choose at most five markets. Each is researched independently, so five markets is five times the research and five times the review.
Competitors
Competitors ground the research. Your app has one shared tracked list; each market uses that storefront’s own keyword and ranking evidence for those apps. Search and suggestions reflect the market you are viewing.
You add competitors by searching the store inside the step. At least one is normally required: continuing without any is an explicit choice, and you can add competitors later from the Competitors page — the next run picks them up. Continue without competitors and Choose and start agent both sit in the bottom bar, which stays visible while you scroll.
Changing the setup later
Markets can be changed later from Generate suggestions or through Agent chat /reconfigure. Competitors can be edited from the Competitors page, and the next run picks up the change.
Resetting an agent clears its configuration along with the proposals and queued screenshot work that belonged to it, so you start clean rather than with cards referring to a setup that no longer exists.
Each metadata pass starts at your request. Spek shows a reminder after seven days rather than starting another pass automatically. See runs and what happens in one.
Common questions
Why does it insist on a competitor?
Competitors give your team useful market context, including for apps that are not on the store yet. You can continue without one during setup and add competitors later from the Competitors page; the next run picks them up.
Choose apps that address a similar user need in the markets you care about. A familiar brand in an unrelated category can make a comparison less useful. Review suggestions before adding them, and revisit the list when your product or target audience changes.
What does resetting the agent do?
It clears the configuration along with the proposals and queued screenshot work that belonged to it, so you start clean rather than with cards referring to a setup that no longer exists.
Reset sits in the agent header and is available to owners, admins and editors. It asks you to type reset to confirm and cannot be undone. Keyword research, competitors, Editor drafts and anything already published to the store are kept.
How should I choose competitors during setup?
Choose apps that meet a similar user need and are relevant to the market you are reviewing. A recognizable brand is not necessarily the most useful comparison if its audience and product are unrelated to yours. Inspect the suggested apps before adding them, and remember that your app's tracked list is shared while the observations remain specific to each storefront. You can revise that list later. The purpose is to give the review useful market context, rather than to collect the largest possible number of competitor names.
What should I check before selecting a new localization?
Confirm that your team can review the language and that the app can support the promise made to that audience. In the App Store workflow, opening a localization can create a record in App Store Connect, so read the confirmation associated with that action. A newly selected market needs its own research and review; it should not be treated as a copy of the primary locale. Check names, feature descriptions and local expectations before approving any wording prepared for the new audience.
Do I need to reset the Agent to adjust my research scope?
A routine change to markets or competitors should start with the relevant configuration controls, rather than with a reset. Generate suggestions lets you choose the markets for a pass, and the Competitors page manages the tracked comparison list. Reset has a broader effect on Agent configuration and proposals and asks for an explicit confirmation. Read that scope before using it. Keeping routine adjustments separate from a reset helps preserve useful review context while still letting the next run use the app and markets you intend.
How can a team prepare for the first multilingual run?
Agree on who will review each language and which product claims must remain consistent. Select a manageable set of markets, confirm the app's brand wording and identify relevant competitors. After the run, evaluate each locale in its own context before applying changes. Someone who understands the product may still need language review from a colleague. This preparation makes the output easier to act on and avoids treating a completed research run as proof that every generated phrase is ready for public release.