This page is a map of the administration interface inside your AiHummer
instance. Use it when you know what you want to do but do not yet know which
screen contains the action.
You need an account on the instance. Some screens are visible only when your
role has the corresponding permission. If a menu item is absent, ask a workspace
owner to check your role instead of trying to work around the restriction.
Work
Screen
Use it for
Successful result
Dashboard
See the health summary and shortcuts after sign-in.
The cards show current data and no action is marked as requiring attention.
Agents
Create an AI employee, choose its model and edit its behavior.
The agent is active and can start a test session.
Orchestration
Assemble a multi-agent flow visually.
The graph validates and can be assigned to an entry agent.
Sessions
Start and review conversations, files, cost and tool output.
A message receives a reply or the session history explains why it did not.
Einstein — memory & knowledge
Screen
Use it for
Successful result
Einstein (memory)
Review long-term information available to an agent.
The intended memory is active and unrelated data is not exposed.
Memory facts
Approve, correct or reject extracted facts.
The fact has the right status and wording.
Knowledge
Upload sources that agents may cite.
Processing finishes and the source becomes searchable.
Shared skills
Add reusable abilities to agents.
The skill is enabled for the intended agent.
Shared blocks
Reuse workspace resources instead of copying them between agents.
The selected members or agents can see the shared item.
Templates
Maintain reusable prompt fragments.
The saved prompt is available when configuring an agent.
Instructions
Store longer operating instructions.
The instruction is attached to the intended agent or workspace.
Channels & plugins
Screen
Use it for
Successful result
Channels
Connect Telegram and other supported customer channels and bind an agent.
The channel is active and a test message reaches the bound agent.
Bindings & access
Say which agent runs each channel and decide who the agent answers.
A channel message reaches the intended agent, and an unknown sender gets no reply.
Plugins
Install, update, configure or remove product extensions.
The plugin shows a healthy installed state.
Connections
Connect personal or workspace accounts for external services.
The connection status is active and the intended users can select it.
Webhooks
Send events to another system.
A test delivery is accepted by the destination.
Control
Screen
Use it for
Successful result
Notifications
Read events and choose delivery channels by category.
A test or new event arrives through the selected channel.
Approvals
Approve or deny sensitive actions requested by an agent.
The request leaves the pending list with your recorded decision.
Changes
Review changes waiting to be applied.
The change is approved and applied, or rejected with no effect.
Moderation
Review content stopped by safety rules.
The item has a final allowed or blocked decision.
Budget & usage
Set spending limits and see consumption.
The limit is saved and the remaining allowance is visible.
Audit
Review who changed or accessed protected settings.
The event can be traced to an actor and time.
Analytics
Compare usage, activity and cost over time.
The chosen period and filters show the expected workspace data.
Logs
Diagnose a recent problem from live operational messages.
The filtered entries identify the time and visible cause of the problem.
Status
Check whether required services and dependencies are available.
Required checks are healthy or identify the component that needs attention.
Management
Screen
Use it for
Successful result
Workspaces
Separate teams, agents and access boundaries.
A member sees the intended workspace and no unrelated workspace data.
Users & roles
Invite members and assign roles.
The member appears with the intended role and workspace access.
Secrets
Store tokens and passwords without displaying them again.
The secret is saved as a masked value and the intended feature can use it.
System & settings
Manage general, license, external-access, network and appearance settings.
The changed card shows its saved value.
Models
Connect model providers and choose available models.
The provider test succeeds and the model is selectable for an agent.
API keys
Create and revoke credentials for your own integrations.
The new key is copied once and works only for its intended scope.
Authentication
Configure sign-in and enterprise identity options.
A permitted test user can sign in and an unpermitted user cannot.
Crons
Run an agent action on a schedule.
The schedule is enabled and the next run time is visible.
Account
Review instance identity and account-related information.
The displayed account and license belong to this instance.
A safe first workflow
Open Status and confirm the instance is healthy.
Open Models, connect a provider and run its test.
Create an agent on Agents and attach only the skills it needs.
Start a test conversation on Sessions.
Connect a customer channel on Channels and bind that agent.
Watch Notifications, Analytics and Budget & usage during the first real
conversations.
If a screen does not work as expected
The menu item is missing: your role probably lacks access; ask an owner to
review Users & roles.
Save is disabled: complete the required fields and resolve the validation
message shown next to the field.
A test fails: keep the visible error text, check Status, then follow
the guide linked from the relevant row above.
A sensitive value looks different after saving: secrets are deliberately
masked. Enter a new value only when you intend to replace the old one.
This page is a map of the administration interface inside your AiHummer
instance. Use it when you know what you want to do but do not yet know which
screen contains the action.
You need an account on the instance. Some screens are visible only when your
role has the corresponding permission. If a menu item is absent, ask a workspace
owner to check your role instead of trying to work around the restriction.
## Work
| Screen | Use it for | Successful result |
|---|---|---|
| **Dashboard** | See the health summary and shortcuts after sign-in. | The cards show current data and no action is marked as requiring attention. |
| **Agents** | Create an AI employee, choose its model and edit its behavior. | The agent is active and can start a test session. |
| **Orchestration** | Assemble a multi-agent flow visually. | The graph validates and can be assigned to an entry agent. |
| **Sessions** | Start and review conversations, files, cost and tool output. | A message receives a reply or the session history explains why it did not. |
## Einstein — memory & knowledge
| Screen | Use it for | Successful result |
|---|---|---|
| **Einstein (memory)** | Review long-term information available to an agent. | The intended memory is active and unrelated data is not exposed. |
| **Memory facts** | Approve, correct or reject extracted facts. | The fact has the right status and wording. |
| **Knowledge** | Upload sources that agents may cite. | Processing finishes and the source becomes searchable. |
| **Shared skills** | Add reusable abilities to agents. | The skill is enabled for the intended agent. |
| **Shared blocks** | Reuse workspace resources instead of copying them between agents. | The selected members or agents can see the shared item. |
| **Templates** | Maintain reusable prompt fragments. | The saved prompt is available when configuring an agent. |
| **Instructions** | Store longer operating instructions. | The instruction is attached to the intended agent or workspace. |
## Channels & plugins
| Screen | Use it for | Successful result |
|---|---|---|
| **Channels** | Connect Telegram and other supported customer channels and bind an agent. | The channel is active and a test message reaches the bound agent. |
| **Bindings & access** | Say which agent runs each channel and decide who the agent answers. | A channel message reaches the intended agent, and an unknown sender gets no reply. |
| **Plugins** | Install, update, configure or remove product extensions. | The plugin shows a healthy installed state. |
| **Connections** | Connect personal or workspace accounts for external services. | The connection status is active and the intended users can select it. |
| **Webhooks** | Send events to another system. | A test delivery is accepted by the destination. |
## Control
| Screen | Use it for | Successful result |
|---|---|---|
| **Notifications** | Read events and choose delivery channels by category. | A test or new event arrives through the selected channel. |
| **Approvals** | Approve or deny sensitive actions requested by an agent. | The request leaves the pending list with your recorded decision. |
| **Changes** | Review changes waiting to be applied. | The change is approved and applied, or rejected with no effect. |
| **Moderation** | Review content stopped by safety rules. | The item has a final allowed or blocked decision. |
| **Budget & usage** | Set spending limits and see consumption. | The limit is saved and the remaining allowance is visible. |
| **Audit** | Review who changed or accessed protected settings. | The event can be traced to an actor and time. |
| **Analytics** | Compare usage, activity and cost over time. | The chosen period and filters show the expected workspace data. |
| **Logs** | Diagnose a recent problem from live operational messages. | The filtered entries identify the time and visible cause of the problem. |
| **Status** | Check whether required services and dependencies are available. | Required checks are healthy or identify the component that needs attention. |
## Management
| Screen | Use it for | Successful result |
|---|---|---|
| **Workspaces** | Separate teams, agents and access boundaries. | A member sees the intended workspace and no unrelated workspace data. |
| **Users & roles** | Invite members and assign roles. | The member appears with the intended role and workspace access. |
| **Secrets** | Store tokens and passwords without displaying them again. | The secret is saved as a masked value and the intended feature can use it. |
| **System & settings** | Manage general, license, external-access, network and appearance settings. | The changed card shows its saved value. |
| **Models** | Connect model providers and choose available models. | The provider test succeeds and the model is selectable for an agent. |
| **API keys** | Create and revoke credentials for your own integrations. | The new key is copied once and works only for its intended scope. |
| **Authentication** | Configure sign-in and enterprise identity options. | A permitted test user can sign in and an unpermitted user cannot. |
| **Crons** | Run an agent action on a schedule. | The schedule is enabled and the next run time is visible. |
| **Account** | Review instance identity and account-related information. | The displayed account and license belong to this instance. |
## A safe first workflow
1. Open **Status** and confirm the instance is healthy.
2. Open **Models**, connect a provider and run its test.
3. Create an agent on **Agents** and attach only the skills it needs.
4. Start a test conversation on **Sessions**.
5. Connect a customer channel on **Channels** and bind that agent.
6. Watch **Notifications**, **Analytics** and **Budget & usage** during the first real
conversations.
## If a screen does not work as expected
- **The menu item is missing:** your role probably lacks access; ask an owner to
review **Users & roles**.
- **Save is disabled:** complete the required fields and resolve the validation
message shown next to the field.
- **A test fails:** keep the visible error text, check **Status**, then follow
the guide linked from the relevant row above.
- **A sensitive value looks different after saving:** secrets are deliberately
masked. Enter a new value only when you intend to replace the old one.
## Next
- [Web UI overview](/en/v1.0/webui/overview) — layout, navigation and access.
- [Create your first agent](/en/v1.0/webui/agents) — a complete first setup.
- [Security screens](/en/v1.0/webui/security) — roles, secrets and approvals.