AI Field Agents — Record-Level AI Built Into Your Fields
Give any supported Field its own AI agent—generating, analyzing, and enriching Record data from a prompt you control, with every change tracked in the audit trail.
What Is AI Field Agents?
AI Field Agents are record-level AI assistants embedded directly into Fields on a SmartSuite Table. Each agent combines a custom prompt with the Record's own field data to generate, analyze, enrich, or update that Field's value—on demand or automatically whenever referenced data changes—and logs every action as AI-driven activity. AI Field Agents are listed on all current SmartSuite plans and work across supported Field types including Title, Text, Number, SmartDoc, Status, Single Select, and Date.
Prompt Builder
Configure Agents with Prompts That Reference Record Data
Each agent is defined in Field Settings with a Prompt Builder that mixes static instructions and field-reference variables like {{Issue Description}}—so the agent reasons over the Record it lives on, not generic context. At least one field variable anchors every prompt to real data.
Agents write their output back into the Field itself, across types from Title, Text, and Text Area to Number, Single Select, Multiple Select, Status, Link, Address, Date, and SmartDoc.
Execution Modes
Run Agents on Demand or the Moment Data Changes
Choose manual execution—triggered from the grid, the record page, or a widget—or automatic execution, where the agent re-runs whenever a referenced field changes. A Stop control cancels an in-progress run at any time.
Clear execution states keep users oriented: a processing indicator while the agent works, an "Updated by AI" marker on success, and explicit error states for a disabled agent, exhausted free requests, an invalid API key, or a broken prompt.
Extended Context
Extend Agents with Internet Search and Reference Files
Admins can attach files at the agent level—persistent reference context the agent uses across every Record it touches, from style guides to assessment criteria. An optional Internet Search capability, enabled workspace-wide by an admin and then toggled per field, lets agents supplement Record data with current external information.
Context stays a deliberate choice: what an agent can see is decided by admins and builders, never by the model.
Traceability
Track Every AI-Driven Change in the Audit Trail
Every agent action is logged as AI-driven activity—what changed, and which agent changed it—so reviewers can distinguish AI updates from human edits at a glance. If a referenced field is deleted, a broken-prompt alert surfaces the issue instead of letting the agent fail silently.
That traceability is what makes record-level AI workable in regulated environments: enrichment at scale, with a review trail behind every value.
Central Governance
Govern Agents Centrally Through the AI Center
AI Field Agents are enabled workspace-wide from the AI Center, where admins manage provider connections and decide which ones agents may use. A Connection Selector on each field lets builders pick a specific approved provider and model—credentials stay with admins.
Every workspace starts with the SmartSuite Trial connection and 100 free AI requests; when they run out, agents stay configured and runs pause with an inline error until a provider connection is added.
AI Field Agents
AI Field Agents keep governed Record data enriched and current—risk registers, control libraries, and vendor files stay complete with no exports, no copy-paste, and a full audit trail behind every change.
Generate a plain-language risk description from structured likelihood, impact, and category fields—refreshed automatically whenever the underlying assessment data changes, with each update logged as AI-driven activity.
Draft a test conclusion from structured control and result fields so testers start from a consistent, reviewable baseline. Reviewers see exactly which values AI proposed before sign-off.
Distill a dense policy document into a concise summary field the attestation audience can read at a glance. Automatic mode keeps the summary current as the policy text evolves through review.
Propose a vendor risk tier in a Single Select field from structured inputs like data access, criticality, and contract scope—reviewers adjust where judgment says otherwise, and the audit trail records both the proposal and the change.
Turn on Internet Search for an agent to add current external regulatory context to a compliance Record—and govern which AI connections the agent may use through the AI Center.
AI Field Agents keep governed Record data enriched and current—risk registers, control libraries, and vendor files stay complete with no exports, no copy-paste, and a full audit trail behind every change.
AI Field Agents are record-level AI assistants embedded into individual Fields on a Table. Each agent runs a custom prompt against the Record's own field data to generate, analyze, enrich, or update that Field's value—with every action logged in the audit trail.
Supported types are Title, Text, Text Area, Number, SmartDoc, Single Select, Multiple Select, Status, Link, Address, and Date—covering both narrative fields and structured values.
They can. In automatic mode an agent re-runs whenever a field it references changes; in manual mode a user triggers it from the grid, record page, or a widget. A Stop control cancels an in-progress run either way.
Every agent action is recorded as AI-driven activity—what changed and which agent made the change—so compliance and IT reviewers can separate AI updates from human edits. This traceability helps regulated teams stay audit-ready while using AI at scale.
Yes, optionally. A Workspace Administrator enables Internet Search at the workspace level, and builders then toggle it per field—letting selected agents supplement Record data with current external information while others stay limited to workspace data.
Every workspace includes 100 free requests through the SmartSuite Trial connection. When they are exhausted, agents remain configured but runs fail with an inline error until an admin adds a paid provider connection in the AI Center.
The SmartSuite pricing page lists AI Field Agents as available on all current plans. Enabling the feature workspace-wide and managing provider connections requires Workspace Administrator permissions.
Enrich Every Record Automatically—and Keep the Audit Trail
Turn on your first agent with the built-in trial connection and watch fields generate, analyze, and update themselves—with every change logged.