This section is where organisation admins decide which AI building blocks the company’s agents are allowed to use — the language models they run on, the tools they can call, and how they remember and compose work.
The page opens with five tabs: AI Models, Agents, Tools, Memory, and Skills (Preview).
AI Models
Decide which language models your agents may run on, and which is the fallback.
| Control / Button | What it does |
|---|
| Model card | One enabled model — provider, name, context window, costs, capability tags. |
| Default badge (crown) | Marks the model used when no specific model is chosen. |
| Suspended badge | Marks a model that has been turned off. |
| Set Default | Makes that model the organisation’s default. |
| Suspend | Turns a model off; requires typing a randomly generated safety phrase. |
| Restore | Turns a suspended model back on. |
| Show suspended ( N ) | Reveals suspended models. |
| Enable New Model | Opens the model catalog dialog. |
Enable AI Model dialog
| Control / Button | What it does |
|---|
| Search box | Filters by model name, provider, or model ID. |
| All providers | Multi-select filter. |
| All features | Multi-select filter for Tool Calling and/or Streaming. |
| Column headers | Click to sort. |
| Enable | Adds the model to your organisation’s enabled set. |
| Enabled tag | Marks already-enabled models. |
Agents
Browse and manage available agents and set departmental defaults per mode.
| Control / Button | What it does |
|---|
| Search agents | Filters by name, description, or department. |
| All departments | Multi-select filter. |
| Show suspended ( N ) | Include suspended agents. |
| Create Agent | Opens the dialog to set up a new agent. |
| Default Agents section | Per-department picker for each of the five modes. |
| Agent card | Name, departments, description, status dot, and mode chips. |
| Configure | Opens the agent’s detail page. |
| Duplicate | Tooltip “Duplicate (coming soon)”. |
| Suspend / Restore | Turns the agent off (typed confirmation) or back on. |
| Delete | Tooltip “Delete (coming soon)”. |
Tiers and status
- System tools — always on and locked.
- Default tools — pre-enabled; you can turn them off.
- Advanced tools — opt-in; off until you enable them.
- A tool may be marked Preview or Coming soon.
| Control / Button | What it does |
|---|
| Enabled tools summary (chips) | Live, category-grouped summary; click a chip’s remove control to drop that tool from the draft. |
| Tier / category / tool checkboxes | Tick or untick whole tiers, categories, or individual tools. |
| Unsaved-changes indicator | ”You have unsaved changes” or “All changes saved.” |
| Save | Applies changes. Removals show a confirmation dialog. |
Memory
Manage what your agents learn and remember over time. Powered by a temporal knowledge graph (“Zep Cloud”).
| Control / Button | What it does |
|---|
| Export audit | Downloads an audit export of memory activity. |
| Overview | A dashboard of memory health and activity. |
| Ascent Queue | Facts waiting to be promoted. |
| Policy Candidates | Proposed rules awaiting review. |
| Inspector | Search and inspect stored memory entries. |
| Audit Log | A record of memory changes and actions. |
| GDPR | Privacy controls for removing personal data. |
Skills (Preview)
Skills are multi-step tool sequences an agent can run as a single action.
This tab is a Preview showing sample data only — “Preview — Mock Data”. The on/off switches are disabled (display only).
Who can use it
All of AI Capabilities is an administrator area. Only organisation owners and admins can change anything here. 
