Asset Vision

Fetches the current live state of an Inspection directly from the AV Cloud Services API. Use this before assigning or scheduling an inspection to verify it has not changed since the last Data Services sync (which may be 15-30 minutes delayed). If the live data differs from what was retrieved via query_table, inform the user before proceeding. Note: field names in the live API response may differ from Data Services column names. Request format: - Provide arguments inside a top-level object named "req". - Example: { "req": { "field": "value" } } - Requests without the "req" wrapper are invalid.

Fetches the current live state of a Job directly from the AV Cloud Services API. Use this before assigning or scheduling a job to verify it has not changed since the last Data Services sync (which may be 15-30 minutes delayed). If the live data differs from what was retrieved via query_table, inform the user before proceeding. Note: field names in the live API response may differ from Data Services column names. Request format: - Provide arguments inside a top-level object named "req". - Example: { "req": { "field": "value" } } - Requests without the "req" wrapper are invalid.

Retrieves photos attached to Asset Vision records and returns them as image content the assistant can view and analyse. Photos are polymorphic: each is attached to a parent record identified by a parentType (Job, Inspection, Asset, Module, CapitalWork, or CapitalWorkTask) and that record's ID. Usage: first identify the relevant records with the other tools (e.g. Jobs matching a defect via InterventionName/ActivityName, or a specific Asset), then pass their IDs here to fetch the photos. - parentType: the kind of record the ids refer to. - ids: the parent record IDs whose photos you want. One call targets one parentType; to show photos for both an asset and its jobs, call twice. Returns each photo as embedded visual content the assistant can view and analyse, plus structured photo data (url, thumbnailUrl, parentId, caption, capturedDate) and a viewable link per photo in the text. Whether a client also DISPLAYS the image inline varies (some do, some don't), so always give the user the photo URL as a link so they can open it regardless. Only photos in contracts the user can access are returned. Because you can see the images, briefly describe what each depicts in field terms and offer a relevant follow-up — e.g. compare 'Job Before' vs 'Job After' to verify the work was completed, assess a defect or hazard's severity, or check the photo's overlay location/date against the record. Only describe what is actually visible in the images; never infer or invent photo contents. Request format: - Provide arguments inside a top-level object named "req". - Example: { "req": { "field": "value" } } - Requests without the "req" wrapper are invalid.

Returns column metadata for an allowed table in Asset Vision Data Services, including column name, type, and nullability. This endpoint is useful for discovering valid field names for filters, groupBy, and joins. Table selection is limited to the allowlisted tables exposed by the connector. Request format: - Provide arguments inside a top-level object named "req". - Example: { "req": { "field": "value" } } - Requests without the "req" wrapper are invalid.

Returns turn-by-turn driving directions between an origin and destination, with optional waypoints. This endpoint provides route summary information, distance, duration, and step-by-step driving directions. Required inputs: - Origin - Destination Optional inputs: - Waypoints Returns: - Route summary - Distance - Duration - Step-by-step directions

Returns the tables in Asset Vision Data Services that are allowed to be queried. Request format: - Provide arguments inside a top-level object named "req". - Example: { "req": { "field": "value" } } - Requests without the "req" wrapper are invalid.

Returns the list of active users who can be assigned jobs or inspections. Call this before assign_job or assign_inspection to discover valid usernames. Returns each user's username (email address, used as the assignedTo value) and display name. Present the display names to the user for selection, then use the corresponding username when assigning.

Returns the list of contracts the authenticated user has access to, as { id, name } pairs. Call this at the start of a conversation when the user mentions a contract by name (e.g. 'Moyne Council') so you can resolve the name to its numeric ContractId for subsequent filters. Sourced from the authoritative AV production database; closed and archived contracts are excluded. Call with no arguments.

Returns the supported join definitions and their relationships. Each result includes the join name, left table, right table, join condition, and join type.

Returns the most common Module.ModuleName values, also referred to as module types. This endpoint is useful for understanding the module names present in Asset Vision data. Optional filters: - Contract (exact match) - Region (exact match) Returns: - ModuleName + Count Request format: - Provide arguments inside a top-level object named "req". - Example: { "req": { "field": "value" } } - Requests without the "req" wrapper are invalid.

Returns the most common sub-register types. In Asset Vision, sub registers are Module records attached under another record via Module.ParentSourceTableName and Module.ParentSourceTableID. This endpoint returns the most common Module.ModuleName values for those sub-register modules. Optional filters: - ParentTable (Job/Inspection/CapitalWork/CapitalWorkTask/Module) - IncludeDeleted (default false) Returns: - ModuleName + Count Request format: - Provide arguments inside a top-level object named "req". - Example: { "req": { "field": "value" } } - Requests without the "req" wrapper are invalid.

Optimizes the driving order for a list of stops. Returns ordered stop IDs, per-leg distance and duration, and an overview polyline. This endpoint does not provide step-by-step driving directions. The ordered stops can be used as input to a separate directions endpoint when navigation steps are needed. Input fields: - origin: optional starting point - destination: optional ending point - returnToOrigin: if true, the route ends at the origin - stops: list of stops to visit, up to 20

Queries data across a supported pre-approved join in Asset Vision Data Services. Join selection is limited to the connector's supported join definitions. Field naming: - Left.<Field> refers to a field from the left side of the join. - Right.<Field> refers to a field from the right side of the join. - Raw SQL and table-name-prefixed field syntax are not supported by this endpoint. Special values: - '@me' (or 'me') may be used on supported user-related fields to refer to the current authenticated user. Examples: - where: { field: "Left.InspectionID", op: "isnotnull" } - selectRight: ["Name", "Code"] Request format: - Provide arguments inside a top-level object named "req". - Example: { "req": { "field": "value" } } - Requests without the "req" wrapper are invalid.

Queries a single allowed table using spatial filtering over SQL Server geography data. Spatial request shape: - Spatial.Near: returns rows within a radius in metres of a point - Spatial.WithinBbox: returns rows intersecting a bounding box - Spatial.Nearest: optional integer used with Spatial.Near to return the nearest N rows ordered by distance Important: - Nearest is not a standalone mode. Use Spatial.Near + Spatial.Nearest together. - DistanceMetres is available when Spatial.Near is supplied. Examples: - Near: { "spatial": { "near": { "lat": -38.0, "lon": 145.0, "metres": 500 } } } - Nearest: { "spatial": { "near": { "lat": -38.0, "lon": 145.0, "metres": 50000 }, "nearest": 25 } } - Bbox: { "spatial": { "withinBbox": { "minLat": ..., "minLon": ..., "maxLat": ..., "maxLon": ... } } } Virtual select fields: - SpatialWkt, Lat, Lon, Srid, DistanceMetres - Direct selection of the underlying spatial column is not supported. Request format: - Provide arguments inside a top-level object named "req". - Example: { "req": { "field": "value" } } - Requests without the "req" wrapper are invalid.

Queries a single allowed table in Asset Vision Data Services using structured filters, sorting, and paging. Table selection is limited to the allowlisted tables exposed by the connector. Field naming: - Plain column names such as AssetName or InspectionID are supported. - Table-prefixed field names are not supported by this endpoint. Supported operators: - eq, ne, gt, gte, lt, lte, contains, startswith, endswith, in, isnull, isnotnull. Special values: - '@me' (or 'me') may be used on supported user-related fields to refer to the current authenticated user. Examples: - where: { field: "InspectionID", op: "isnotnull" } - orderBy: { field: "CreatedDate", dir: "desc" } Request format: - Provide arguments inside a top-level object named "req". - Example: { "req": { "field": "value" } } - Requests without the "req" wrapper are invalid.

Creates a chart specification from aggregated data for bar, line, or pie charts. This endpoint returns a ChartSpec and does not produce a visible image. It is intended for use with aggregated results such as grouped counts or summaries. Field requirements: - xField must match a group-by field present in the aggregate result. - yField must match an aggregate alias present in the aggregate result (for example, JobCount). Chart types: - bar: compares values across categories - line: shows values across an ordered sequence such as time - pie: shows proportional breakdowns Request format: - Provide arguments inside a top-level object named "req". - Example: { "req": { "field": "value" } } - Requests without the "req" wrapper are invalid.

Renders a PNG chart image from a ChartSpec and returns its URL. This endpoint produces a user-visible chart image from a valid chart specification. ChartSpec values are typically produced by the chart specification endpoint or another compatible source. Input requirements: - chart must contain a valid ChartSpec with title, type, xField, yField, and points. - width and height control the output image size. Request format: - Provide arguments inside a top-level object named "req". - Example: { "req": { "field": "value" } } - Requests without the "req" wrapper are invalid.

Renders aggregated data as formatted text charts such as ASCII bars, sparklines, or emoji bars. This endpoint returns a text-based chart representation suitable for chat display. It is intended for aggregated results such as grouped counts or summaries. Styles: - ascii: wide bars using █ characters - sparkline: compact bars using block characters - emoji: friendly bars using coloured blocks such as 🟦🟥🟩🟨 Request format: - Provide arguments inside a top-level object named "req". - Example: { "req": { "field": "value" } } - Requests without the "req" wrapper are invalid.

Resolves a user phrase such as 'work order', 'CRM', or 'customer request' to the best matching Module.ModuleName values. This endpoint searches existing ModuleName values and returns ranked suggestions. The returned candidates can be used in Module queries or filters on ModuleName. Optional filters: - Contract (exact match) - Region (exact match) Returns: - candidates [{moduleName, score, count, reason}] Request format: - Provide arguments inside a top-level object named "req". - Example: { "req": { "field": "value" } } - Requests without the "req" wrapper are invalid.

Resolves a user phrase to the best matching sub-register types. In Asset Vision, sub registers are Module records attached under another record via Module.ParentSourceTableName and Module.ParentSourceTableID. Example phrases may include 'sub register', 'activity notes', 'CRM tab', or 'safety checklist'. Returns ranked Module.ModuleName candidates for matching sub-register modules. Matched results can be used with compatible child-module joins and ModuleName filters. Request format: - Provide arguments inside a top-level object named "req". - Example: { "req": { "field": "value" } } - Requests without the "req" wrapper are invalid.

Opens the connected Asset Vision dashboard UI.

Updates the Priority of a Job via the AV Cloud Services API. Use this to adopt 'fix_priority_anomaly' suggestions surfaced by analyse_schedule (e.g. when items of identical work are split across priority tiers). newPriority is the rank (integer 1-4) stored in Job.Priority — HIGHER = more urgent. Each rank maps to a Terminology key: 1=PriorityLow, 2=PriorityMedium, 3=PriorityHigh, 4=PriorityUrgent. The label the tenant uses for each rank is configurable (one tenant may call rank 4 'Urgent', another 'Critical', another 'P1'). If the user mentions priority by label, call get_about first and resolve their term to the rank via tenantTerminologyOverrides — match the row whose Key is one of PriorityLow/Medium/High/Urgent and whose tenantLabel matches the user's word. Do not guess. Confirm the proposed change with the user before calling. When adopting a multi-item fix_priority_anomaly suggestion, call this tool once per item — there is no bulk endpoint. Request format: - Provide arguments inside a top-level object named "req". - Example: { "req": { "field": "value" } } - Requests without the "req" wrapper are invalid.