Frequently Asked Questions

General

Is it free to use?

Yes. Free credits (1 credit) are granted every week when you sign up. Credits expire 7 days after they are issued, so use them within the week. To use more, switch to a paid subscription.

Which LLMs are supported?

Anthropic Claude, OpenAI GPT, Google Gemini, MiniMax, and more. Internally we group models into three tiers — Frontier, Balanced, and Lite — and each agent can be set to whichever tier fits its job.

On paid plans (Starter and above), you can pick which specific model each tier maps to in LLM model settings for Balanced and Lite. Choosing the model for the Frontier tier requires a Pro plan or higher.

Can I use it on mobile?

There’s a native iOS app. On Android, use a mobile web browser. Push notifications are only available through the iOS app.

Agents

My agent stopped midway

The most common causes:

Awaiting approval: It may be waiting for your approval before running a risky tool. Check the approval button in the chat.
Awaiting your answer: It may have asked you a clarifying question and is waiting for your reply.
Rate limit: Too many LLM calls in a short window can pause it briefly. Try again shortly.
Out of credits: Check your remaining balance.

It keeps repeating the same action

If the agent calls the same tool over and over as if it can’t see its own results, message it in chat to add a stop condition to its behavior instructions.

Example: “Once you’ve delivered the result, stop working.”

Responses are too slow

Drop the model tier: Switch the agent’s model tier to Lite (or Balanced) in agent settings — cheaper and faster. Choose Frontier only when you need top-tier performance, as it costs more.
Enable only the tools you need: Disable tools the agent isn’t using.
Trim the behavior instructions: Long instructions consume capacity every turn — message the agent “shorten your behavior instructions”.

How do I receive results from autonomous runs (schedule or webhook)?

Tell the agent: “Notify me with the result after each autonomous run.” From then on it will send you a notification at the end of every automated run. To receive push notifications on iOS, make sure push notifications are enabled under Settings → Notifications.

Schedules

My schedule isn’t running

Confirm the agent is active (paused agents don’t run schedules)
Confirm the individual schedule itself is enabled in the agent’s settings
Check the agent’s Logs tab for run history

I didn’t get the result from a scheduled run

Tell the agent to report the result back from now on. For mobile alerts, push notifications must be enabled under Settings → Notifications.

Teams

How do agents talk to each other inside a team?

Mention another agent with @AgentName in team chat and the mentioned agent wakes up and responds automatically. When a team is created, every member is connected to every other member by default — so anyone can mention anyone and trigger an automatic run. You can remove specific connections in the team’s Graph tab to restrict certain mention paths (the chat message is still delivered, but no automatic run fires). The one exception is a thread reply: replying to an earlier mention wakes the original sender regardless of whether a connection exists.

Can a team-member agent wake up another member?

Yes. By default every member of a team is connected to every other member, so any agent can wake any other agent with an @mention. If you remove a specific connection in the team’s Graph tab, mentions across that path are recorded as chat text but no longer trigger a run. One exception: when an agent replies to a prior mention (a thread reply), it can always wake the original sender even without a direct connection.

Marketplace

My installed skill isn’t working

Verify all tools the skill requires are enabled
If the skill needs an external service (Notion, GitHub, etc.), open Marketplace → My Skills and complete the connection on that skill
Check the skill’s README on its detail page for prerequisites

Ask the agent in chat — “package this workflow as a skill” — and it packages the code and prepares a draft for your review. Once the draft is ready, a review screen opens where you can adjust details before publishing it as private (default) or to the public marketplace. See the skills guide for the full flow.

Notifications

I’m not receiving push notifications

Work through these steps in order:

Install the iOS app and sign in: Push notifications are only delivered through the iOS app. Download UpServe from the App Store and sign in at least once so your device is registered.
Enable push in the app: Go to Settings → Notifications and confirm the Push Notifications toggle is on.
Check device permissions: In your iOS Settings, make sure UpServe is allowed to send notifications.
Check Do Not Disturb: If the current time falls within your Do Not Disturb window in notification settings, alerts will be silenced.

Credits

When do free credits expire?

Free plan credits (1 credit per week) expire 7 days after they are issued. If you still have a balance when the next weekly grant arrives, the leftover amount expires at that point. Use your credits within the week to avoid losing them.

Security & Privacy

What happens to the data I enter?

See the AI Transparency policy . Your data is not used for LLM training by default.

Can the agent do something dangerous?

The system automatically evaluates every tool call before it runs. Actions with a high risk level — such as file deletion or external API calls that change remote data — trigger an approval request that pauses the agent until you respond. Human-in-the-Loop is on by default.

Still have questions?

Use the chat button at the bottom-right of the app
Browse the Human Feed for community answers