Run Actions on Contacts and Companies
The Run Action button on contact and company detail pages lets you execute any AI skill against a single record. The AI receives the full record data as context, so outputs are specific to that person or company — no copy-paste required.
How It Works
- Open any contact or company detail page
- Click the lightning bolt icon in the page header
- Search or scroll to select an action
- The action runs with this record’s data automatically injected as context
- Results appear in a panel — you can copy or close when done
The panel stays open while the action runs and shows a completion time when finished.
What Actions Are Available
The dropdown shows all enabled AI actions configured for your organization. These are the same actions available in the AI Agents section.
Actions that appear here:
| Type | Description | Example |
|---|---|---|
| Research | Summarize company news, find trigger events, identify pain points | ”Summarize recent funding news for this company” |
| Personalization | Generate an opening line, write a tailored email, draft a connection note | ”Write a LinkedIn opener for this contact” |
| Analysis | Qualification scoring, fit assessment, competitive context | ”Score this lead against our ICP” |
| Enablement | Meeting prep notes, talk track suggestions, objection handling | ”Generate meeting prep notes for this account” |
Actions must be enabled in AI Agents to appear in this list. Only AI-powered (LLM) actions are shown here — automation and webhook actions are not.
Context Injection
When you run an action from a contact page, the action receives:
- First name, last name, job title, company name
- Email address and phone number
- Location, LinkedIn URL
- Any custom fields on the record
When run from a company page, the action receives:
- Company name, domain, industry
- Employee count, revenue range, location
- Technologies, keywords, description
- Any custom fields on the record
The AI uses this data for variable substitution in action templates and as background context in its prompt.
Global Context Included
Beyond the record data, every action also receives:
- Brand Brief — your company positioning
- Value Propositions — key differentiators
- ICP definition — who you sell to
- Offer Brief — pricing and packaging
- Personas — buyer role profiles
This context comes from Studio’s Global Context. Update it at Studio → Global Context to improve every action at once.
Result Panel
Results render in-line with formatting (headings, lists, links). You can:
| Button | Action |
|---|---|
| Copy | Copy the full result to clipboard |
| Copy as Markdown | Copy with markdown formatting preserved |
| Regenerate | Re-run the action (useful if the first output wasn’t quite right) |
| Save to Record | Store the result as a note on the contact or company |
| Send to Copilot | Continue iterating in Copilot chat |
Saving Results
When you click Save to Record, the result is added to:
- The contact or company’s Notes tab
- The CRM (if sync is enabled) as an activity note
- The Inbox timeline as an “AI Generated” entry
Saved results can be referenced in future actions — they’re part of the record context going forward.
Running Actions at Scale
Single-record actions are for ad-hoc work. To run the same action against many records, use one of these:
| Method | Best For | Location |
|---|---|---|
| List bulk action | Run against every record in a list | Data → Lists → Select list → Run Action |
| Campaign step | Run as part of a sequence | Engage → Sequencer → Add AI Action Step |
| Workflow trigger | Auto-run when a contact hits a condition | Agents → Workflows |
| Enrichment column | Run as a computed column on every record | Data → Custom Fields → AI Column |
Permissions and Credits
Who Can Run Actions
| Role | Permissions |
|---|---|
| Admin | Run any enabled action, edit actions |
| Member | Run any enabled action |
| Viewer | Cannot run actions |
Disable individual actions for specific roles at AI Agents → Action settings → Role visibility.
Credit Costs
Each action consumes credits based on model tier:
| Model Tier | Cost per Run | Typical Actions |
|---|---|---|
| Fast | 1 credit | Short responses, summaries |
| Standard | 3–5 credits | Most personalization and research |
| Deep | 10–20 credits | Deep research with web access |
| Thinking | 15–40 credits | Qualification scoring, complex analysis |
Running actions from detail pages counts against your monthly credit quota. Check remaining credits in Settings → Billing.
Error Handling
If an action fails, the panel shows the error reason:
| Error | Cause | Fix |
|---|---|---|
| Missing required field | The action needs a field that’s empty on this record (e.g., LinkedIn URL) | Enrich the record first |
| Insufficient credits | Your credit balance is below the action’s cost | Top up in Settings → Billing |
| Model timeout | The LLM didn’t respond in time (rare) | Click Regenerate |
| Action disabled | An admin disabled this action after you opened the page | Ask your admin to re-enable |
| Rate limited | Too many concurrent actions from this user | Wait 30 seconds and retry |
Activity Log
Every action run is logged:
- Who ran it (user email)
- What action was run
- Which record it ran against
- When it ran (timestamp)
- Cost in credits
View the log at Settings → Activity → AI Actions. Use this for compliance review, usage auditing, or understanding team behavior.
Related
- AI Agents → — Create and configure actions
- Campaigns → — Run actions at scale across audiences
- Copilot Skills → — Context-aware skills in the Studio editor
- Workflows → — Auto-run actions on triggers
- Global Context → — Shared context every action uses