Parallel AI integration researchA little less busywork with Parallel AI.
Research only. Grunt is prelaunch, live service is paused, and live provider connections and outcomes remain unverified. The operations and examples below are not confirmed live Grunt features or approved supplier commitments.
Explore possible Parallel AI workflows with Grunt. Documented operations include Agent: Returns when the conversation last changed without loading any message; Agent: Creates a new AI agent with the specified configuration the agent can; Agent: Permanently deletes the specified agent and all associated data. Actions on this page are research candidates from public integration definitions, not confirmation that Grunt support is live.
Exploratory scope only: these candidates come from public integration definitions and are not confirmed as live Grunt features.
Parallel AI
“Explore whether a Grunt integration could returns when the conversation last changed without loading any message, then creates a new ai agent with the specified configuration the agent can in Parallel AI. Show me the result and wait for my approval before changing anything.”
A workflow idea to explore.
Illustrative steps and outcome; no real work is performed here.
- 01Agent: Returns when the conversation last changed without loading any message
- 02Agent: Creates a new AI agent with the specified configuration the agent can
- 03Review the result before anything changes
A clearer next step, with the final say still yours.
Explore possibilities with Parallel AI.
Researched operations, not a list of enabled Grunt actions. Actual support, account permissions, supplier terms, and live acceptance require separate verification.
533 researched capabilities across 40 groups
- Returns when the conversation last changed without loading any message
- Creates a new AI agent with the specified configuration the agent can
- Permanently deletes the specified agent and all associated data
- Returns the full agent configuration including channel settings
- Returns a paginated list of conversations for all agents in the company
- Returns the full conversation including all messages exchanged between
- Returns a paginated list of agents belonging to the company associated
- Updates the agent configuration only include fields that need to be
- Converts the provided text to speech using elevenlabs
- Call the delete audio
- Returns a list of available text to speech models and their capabilities
- Returns both public voices and company specific cloned voices
- Returns a paginated list of audio files for the company
- An agent screenshots and reads the source itself to assemble the whole brand
- Call the kit create brand kit
- Call the kit delete brand kit
- Call the kit get brand kit
- Default a paginated list with listall true returns the full list
- Call the kit save brand kit
- CSS Head scripts guidelines with automatic S3 upload for images
- Runs the content engine image pipeline AI image or canvas agent per the
- Call the kit upload brand kit image
- Cancels the specified browser task if it s still pending or running
- Poll this endpoint to check the status of a submitted browser task
- Returns a list of browser session integrations linkedin twitter etc
- Submits a browser task for asynchronous execution returns a task ID
- Renders a self contained HTML canvas at exact pixel dimensions and returns a permanent public URL for social posts story graphics quote cards and link preview art costs 0 1 credits per call
- Default to free stock music for music soundtracks reels and stories use search stock music then import stock music use paid create music ONLY when the user explicitly requests custom AI generated music never automatically fall back to paid generation when stock search import fails try different keywords or tracks and report if none are suitable
- Import a selected search stock music result by stockaudioid rechecks its indexed license and matching license link on the original source page saves an MP3 with source license metadata and returns a permanent public audiourl plus measured duration in seconds use audiourl in lt audio src URL loop data volume 0 2 gt lt audio gt in an animated canvas no audio generation credits are charged sources that cannot be verified are rejected choose another track imports are bounded by the configured file size and duration limits default 40 mib and the first ten minutes
- Search pexels stock videos for animated canvases returns direct videourl MP4 urls preview images dimensions duration and creator attribution use the exact videourl in lt video autoplay loop muted playsinline gt with object fit cover sized to the canvas layer readable text above it set animated true when calling create canvas or set animated true in the canvas builder video frames are seeked with the animation timeline in exports use separate audio tracks for sound still pngs have no motion paginate with page pagesize this tool is available whenever canvas tools are enabled
- Character based estimate close enough for claude code s context
- Point claude code or any anthropic SDK at
- Lets claude code discover platform models when
- Fully compatible with openai s API specification for seamless
- Call the delete company
- Call the get companies
- Call the get company
- Call the get company tags
- Call the leave company
- Call the save company
- Call the transfer ownership
- Call the update companies order
- Call the update company features
- Call the users add company user
- Call the users get company users
- Call the users get member subscription type
- Call the users get my role
- Call the users remove member
- Call the users resend invite
- Call the users update company user
- Call the users update company user features
- Call the assign content
- Call the bulk schedule content
- Shares the app builder s agent loop and typed SSE protocol text
- Call the canvas design chat upload
- Based on email deliverability best practices and spam filter guidelines
- Call the clone public prompt
- Pure text analysis the same checks wordpress SEO plugins run no model
- Call the create canvas design
- Call the create canvas design thread
- Use this to insert a content piece directly skipping AI ideation generation
- Call the delete canvas design
- Call the delete canvas design thread
- Call the delete content
- Call the delete content capaign
- Call the delete content strategy
- Call the delete profile
- Call the delete prompt
- Call the delete voice
- Call the download content as pdf
- Call the export content ideas
- This endpoint is used at send time to generate small personalized text snippets
- Supports email sms linkedin message and voice message content types
- Call the generate content from idea
- Call the generate content strategy with ai
- Call the generate profile
- Spintax allows for variation in messages using RANDOM option1 option2 syntax
- Call the generate voice
- Either linkedin url or link id must be provided the profile type will be
- Call the get assigned content
- Call the get canvas design
- Call the get canvas design thread
- Call the get content campaigns
- Call the get content images
- Call the get content statistics
- Call the get content strategy
- Call the get profile
- Call the get prompt
- Call the get specific content strategy
- Call the get voice
- Strips invisible homoglyph characters then runs the text through the
- Call the import content ideas
- Call the list canvas design threads
- Call the list canvas designs
- Call the list content
- Call the list content strategies
- Call the list profiles
- Call the list prompts
- Call the list public prompts
- Call the list voices
- Call the publish content
- Call the render canvas design
- Seeks the design s CSS animation timeline frame by frame so the canvas
- Call the save content
- Call the save content campaign
- Call the save content strategy
- Call the save profile
- Call the save prompt
- Call the save voice
- Call the select hook and generate
- Message sends the partial reply is already saved
- Call the trigger content campaign
- Call the update assignment status
- Call the update canvas design
- Call the update canvas design thread
- Call the update content status
- Call the update scheduled publish date
- Create one project per ideal customer profile or product line each run
- Soft deletes the project its opportunities stop appearing in listings
- Hides it from the default listing re running the project will not
- Creates a record in the content engine and writes the piece in the
- Returns the full record including the score breakdown generation brief
- Infers a project name competitor domains and seed keywords from the
- Pass projectid to scope results to one project s ICP omit it to see
- Each project targets one domain and one ideal customer profile A company
- Newest first poll this after triggering a run the first entry carries
- Returns it to new so it appears in the default listing again
- Runs asynchronously through the SERP AI answer keyword metrics
- Creates a daily keyword monitor across google AI and chatgpt for up to
- Changing the domain competitors seed keywords or ICP affects the next
- Call the assign employee departments
- Call the assign user departments
- Call the create department
- Call the delete department
- Call the get departments
- Call the get employee departments
- Call the get user departments
- Call the update department
- Call the get documentation page
- Use this first to see what areas of documentation exist then call
- Omit parentid to list top level categories or pass a parent page ID
- Matches are ranked by how many query tokens appear in the title weighted
- Creates a support ticket of type feature request tagged as a
- Extracts text content from uploaded files without saving them to the knowledge base
- Creates a copy of the document in the specified target folder
- Supports two modes
- Creates a folder at the specified path
- Permanently removes the document and all associated embeddings
- If delete true also deletes all documents inside the folder
- Returns document details including name path type and tags
- Returns the full text content of the document
- Returns the full text content extracted from the document
- Returns all documents and folders at the specified path
- Returns the folder structure with optional tag information
- Returns documents deleted within the specified number of days
- Accepts either a single documentid or a documentids list and moves
- Updates the folder location and propagates the change to all child documents
- Updates the folder name and propagates the change to all child documents
- Restores a previously deleted document from the recycle bin
- Performs semantic search across the knowledge base
- Creates a new document or updates an existing document with the provided content
- Updates the document content optionally update name and tags
- Sets the auto apply tags for documents created in this folder
- Supports PDF DOCX TXT and other text based file formats
- Send a message to a specific AI employee and receive a response based on
- Returns immediately with a chat ID that can be used to poll for the response
- Call the create employee
- Takes a persona ID and creates an employee with the persona s attributes
- Call the delete employees
- Poll this endpoint with the chat ID returned from the async chat endpoint
- Call the get employee
- Call the get employee persona
- Call the get employee persona categories
- Call the get employee personas
- Call the get employees
- Returns employees for the company associated with the API key
- Call the save employees
- Call the search employee personas
- Takes a source image and applies a template overlay from a brand image
- Creates an image using the specified model GPT image or google and
- Permanently removes the image and any associated data
- Call the download stock image
- Returns the image metadata including urls and generation details
- Call the get stock image providers
- Returns a list of available image generation models with their
- Returns a paginated list of images with their metadata
- Call the search stock images
- Accepts image files PNG JPG JPEG GIF WEBP up to 10MB
- Creates integrations inbox records and sender profiles for each inbox
- Call the delete draft
- Call the delete inbox
- Call the disable warmup
- Call the edit reply with AI
- Call the enable warmup
- Call the generate compose content
- Call the generate reply
- Call the get drafts
- Call the get inbox statistics
- Call the get message details
- Call the get warmup analytics
- Supports filtering by message types comma separated and inbox ids
- Only returns shared inboxes or private inboxes owned by the user
- Returns all inboxes regardless of sharing status useful for sender profile dropdowns
- Call the list messages
- Call the list warmup inboxes
- Call the save inbox
- Call the send composed email
- Call the send draft
- Call the send reply
- Call the set warmup filter tag
- Call the transfer inbox ownership
- Call the update inbox
- Use the returned ID when setting a sequence s crmintegrationid with
- Use this to find ids for sequence and sender profile channel fields common
- Pass companyid as a query param and the integration in the path use a
- Use this to pick a number for a sender profile s phone or a sequence s
- Pass companyid and the heyreachintegrationid as query params use a returned
- POST Companyid the salesforgeintegrationid and a workspaceid use a
- POST Companyid and the salesforgeintegrationid use a returned workspace ID
- Call the delete api key
- Call the get api key
- Call the list api keys
- Call the save api key
- Call the upsert api key
- Call the pages create landing page
- Call the pages create landing page thread
- Call the pages delete landing page
- Call the pages delete landing page thread
- Call the pages get landing page
- Call the pages get landing page preview
- Call the pages get landing page thread
- Call the pages get landing pages
- Shares the app builder s agent loop and typed SSE protocol text
- Call the pages landing page chat upload
- Call the pages list landing page threads
- Call the pages publish landing page
- Call the pages save chat history
- Message sends the partial reply is already saved
- Call the pages unpublish landing page
- Call the pages update landing page
- Call the pages update landing page content
- Call the pages update landing page thread
- Provide any combination of name email phone company domain linkedin
- At least one of filters segment audience is required combining them
- Takes the same criteria as the generate endpoint but instead of returning
- Describe true returns the schema map every field the generate endpoint
- Thousands of B2B B2C segments searched never listed returns matching
- Call the add header
- Lookup columns are computed read only columns whose values are derived live
- Call the add row
- Creates multiple rows with the provided data optionally creates new
- Table actions are automations attached to a list as an action column e g
- Rows are grouped by their values in columnnames within each group the
- Call the delete header
- Soft delete the stored results are kept so dismissing a search from the
- Call the delete list
- Call the delete rows
- Call the delete table action
- Copies every active row column definition cell value cell research and
- Call the edit header
- For lists under 10k rows returns CSV directly
- Returns aggregate counts of how many rows are pending processing
- Call the get action summary
- The results are returned exactly as they were when the search ran so they
- Result rows are omitted here the listing carries the label result count
- Call the get list
- Returns every action and lookup column that can be attached to a list
- Returns only the paginated rows payload pair this with
- Returns the list s name type and ordered column definitions headers
- Call the get lists
- Returns all rows that were created after the specified ISO timestamp
- Call the get public list metadata
- Call the get public list rows
- Call the get row history
- Use this to poll an asynchronous operation e g deduplication that
- Works for any saved search whichever surface produced it a sales navigator
- Returns immediately with a background task ID the scrape and import run
- Copies the specified rows to a target list optionally removes
- Adds the specified rows to a sequence as new members optionally
- Pass selectedrows to run on specific rows or omit it to run on every row
- Call the save list
- The search runs in the background this does NOT return the leads it
- Call the stop table action
- The request body is the list object include ID plus any fields to change
- Call the update row
- Apply the same column value assignments set to every row matching the
- Only campaigns in scheduled status can be cancelled campaigns that are
- A Newly created campaign is always in draft status and does not send anything
- Author the email body as bodydocument the visual builder JSON the
- The underlying send history and recipient rows are retained
- Listings search aggregate metrics and the send pipeline but is retained
- Call the campaigns delete template
- Call the campaigns get campaign
- Send status open count click count and error message where applicable
- Plus open and click rates across all of the company s marketing campaigns
- Default 7 max 90 optionally restrict to a single campaign with
- Call the campaigns get template
- Captured each time the campaign s email body changes
- Returns all non deleted campaigns ordered by creation date newest first
- A Version is captured whenever the visual document and therefore the
- Templates are reusable email designs subject visual builder body
- Itself captured as a new version so it is always reversible only allowed
- The restore is itself saved as a new version so it can always be undone
- The campaign must have a senderprofileid listid subject and bodyhtml
- Uses the campaign s configured sender profile subject and body but does
- Scheduled cancelled or failed status editing a scheduled campaign drops
- Changing bodydocument re renders bodyhtml on the server and saves a new
- Calls the specified tool with the provided arguments and returns the result in MCP format
- Returns all available tool definitions in MCP format including input schemas
- Returns server capabilities protocol version and metadata
- Call the create memory route
- Call the delete memory route
- Call the get memories list
- Call the get memory
- Call the toggle memory active route
- Call the update memory route
- Returns a list of AI models available for chat completions
- Creates a monitor to track specified keywords across social media platforms
- Permanently deletes the specified monitor associated mentions are
- Marks the mention as dismissed hiding it from the default view
- Uses AI to generate reply suggestions in four different styles
- Returns full details for a mention including the post content author
- Returns counts of mentions broken down by status new seen dismissed
- Returns a dense zero filled series for the total and for each platform
- Returns the full configuration for a keyword monitor including its
- Returns social media mentions found by your keyword monitors mentions
- Returns all keyword monitors for the current company monitors can be
- Temporarily stops the monitor from checking for new mentions the monitor
- Reactivates a paused monitor so it will begin checking for new mentions
- Immediately triggers the monitor to check all configured platforms for
- Updates the monitor configuration only include fields that need to be
- Caller must be an owner or admin only owners may grant the owner role
- The organization starts on the free tier the caller is its first owner
- Each entry includes a role indicating the caller s relationship
- Call the list organization members
- Call the remove organization member
- Call the update my organization
- Call the update organization member
- Call the get company graph
- Each person unifies every touchpoint across smart lists sequences
- Call the get people stats
- Touchpoints are grouped by source type and ordered newest first this is the
- Where the profile s touchpoints say where a person appears this says how
- Resolves matches through the identity graph so records that belong to the
- Links a workspace project to a chat for file access
- Creates a copy of a file at the destination path
- Creates a project with optional description AI instructions and sharing settings
- Creates a new directory at the specified path
- Creates a file or directory at the specified path
- Deletes the project and all associated files
- Permanently removes a file from the project
- Removes the project link from a chat
- Returns the file in the requested format as a download
- Returns the attached project or null if none
- Returns project if found and accessible to the user
- Returns file details including content
- Returns list of files in the specified directory path
- Returns nested tree structure with all files and directories
- Returns paginated list of projects with optional search and sharing filters
- Moves a file from source path to destination path
- Returns the file with publicslug and publicurl for sharing
- Removes the public URL and makes the file private again
- All fields are optional only provided fields will be updated
- Replaces the content of an existing file
- Scheduletype is one time needs scheduledat recurring needs
- Call the tasks delete scheduled task
- The copy starts paused so it can be renamed and reviewed before it begins
- Called when the user picks a template so the create task form opens with
- By default this triggers a fresh run pass executionid to evaluate a run
- Call the tasks get scheduled task
- Call the tasks get scheduled task evaluation
- Call the tasks get template setup
- Call the tasks list all executions
- Each template s persona the AI employee that runs the task is resolved
- Call the tasks list scheduled tasks
- Call the tasks list task daily logs
- Call the tasks list task executions
- Call the tasks pause scheduled task
- Soft deletes its executions evaluations and daily logs notes and zeroes
- Call the tasks resume scheduled task
- Runs the same executor the scheduler uses so it records an execution with
- Headline stats across the company then a page of tasks with their run
- Call the tasks update scheduled task
- Overwrites the prompt and tool settings with the template s latest keeping
- A Queued audit stops immediately A running one finishes its current wave
- Intended for putting your own domain next to its competitors each domain
- Call the create project route
- Endpoint reference https docs dataforseo com v3 find the endpoint
- Call the delete audit route
- Call the delete project route
- Returns domain rating estimated monthly organic traffic total backlinks
- Searches the technology index for domains whose live HTML contains the
- Module pages call this when a site is typed freeform the domain gets a
- Priced from the vendor s list rates and the settings supplied tracked
- Call the get audit route
- Call the get project route
- Pass issueid to see only the pages carrying one particular check which is
- Call the list audits route
- Returns every keyword the company has metrics for including the ones a
- Call the list project backlinks route
- Call the list projects route
- Call the pause project route
- Each row is one AI platform s answer to one tracked prompt already scored
- Call the project ai visibility trend route
- Call the project backlink history route
- Call the project competitors route
- Audits plus per scan scores over the last days for the trend chart
- Call the project lighthouse snapshot route
- Call the project link profile route
- Call the project rank history route
- Expands the seed keyword into related terms and returns volume CPC and
- Call the resume project route
- Call the run ai readiness route
- Body may carry modules SEO MODULE keys to run only those
- Sets the saved flag on keywords the company already has metrics for saving
- Pass domain to flag which connection s site covers a project s domain
- Seeding a monitored website s keywords and AEO prompts which are questions
- Crawls the domain from its homepage and checks each page for on page SEO
- Clears the saved flag only the keyword and its metrics stay so a later
- Call the update keyword route
- Call the update project route
- Used when a recipient takes a tracked action off channel and the user wants to
- Call the add sequence members
- Call the add sequence step
- Call the add unsubscribe
- Call the approve all company executions
- Call the approve sequence executions
- Call the bulk add unsubscribes
- Call the bulk update profiles
- Supported variables
- Call the check email unsubscribe
- Call the create profile
- Call the delete all company executions
- Call the delete profile
- Call the delete sequence
- Call the delete sequence executions
- Accepts an optional memberdisposition with targetstepid for
- Call the duplicate sequence route
- Call the edit sequence execution
- Call the execute all company executions
- Call the execute sequence executions
- Efficiently collects all activity data without pagination limits
- Call the generate sequence
- Call the get all company executions
- Call the get all sequences hourly metrics
- This includes standard member fields and custom userdata fields from sequence members
- Finds the member across different parts of the platform
- Call the get sequence
- Uses database level pagination for efficiency with large datasets
- Call the get sequence execution
- Call the get sequence executions
- Call the get sequence hourly metrics
- Call the get sequence member timeline
- Call the get sequence members
- Call the get sequence metrics
- Call the get sequence step details
- Powers the delete confirmation prompt which has to tell the user how many
- Call the get sequence step metrics
- Call the get sequences
- Call the get unsubscribes
- Call the import from smart list
- Call the list profiles
- This is the source of truth for what steps a sequence can contain each entry
- Call the preview sequence execution
- Regenerates the subject and content for the execution based on the step s settings
- Call the remove sequence member
- Call the remove unsubscribe
- This allows failed or executed steps to be retried
- By default partial omitted or true only the fields present in the body
- Call the send test email
- Marks them as executed so the sequence can continue
- Call the test sequence step
- Call the trigger sequence member
- Call the update profile
- Changed omitted fields keep their saved value and the other steps in the
- Returns the skill s markdown instruction text read the relevant skill
- Returns each skill s slug and title fetch a skill s full instructions with
- Submit a bug report feature request question or other support issue
- Call the create support ticket message
- Call the get support ticket
- Call the list support ticket messages
- Regular users only see their own tickets supports filtering by status
- Include only the fields you want to change status and response fields
- Call the create tag
- Call the delete tag
- Call the get tags
- Call the update tag
- Calculates time saved based on automated actions performed in the last 30 days
- Takes the same startdate enddate useruid and feature filters as
- Call the get org usage histogram
- Caller must be an organization owner or admin optional useruid and
- Call the get org usage history summary
- Call the get usage
- Call the get usage by features last30 days
- Call the get usage histogram last30 days
- Call the get usage leaderboard
- Filtered to the user s own usage when authenticated via a company API key
- Returns total credits tokens requests plus breakdowns by model and feature
- Call the create video
- Call the delete video
- Returns the video data including URL and status
- Call the get video models
- Returns the current status of video generation when complete
- Call the get video styles
- Returns a paginated list of videos for the company
- Request body multipart form data
- Call the delete voice
- This endpoint fetches voices from elevenlabs API and caches them
- Call the get voice
- Query params
- Request body
- Call the update voice
- Disambiguates a phrase into the trending topics it could refer to e g
- Returns name address phone website hours rating and coordinates
- Returns photo thumbnails and full size image urls for a place identified
- Returns reviews rating text date reviewer for a place identified by
- Fetches the page and returns it as clean markdown default or raw HTML
- Renders the page captures it as a PNG full scrollable page by default
- Returns job listings title company location description for a query
- Returns titles urls publishers and dates for matching news coverage
- Finds businesses matching a query like coffee shops in austin with
- Returns titles urls snippets and dates for the best matching pages
- Pass a page URL screenshotted in a real browser OR a direct imageurl
- Removes a file on a review branch a branch tombstone the branch is
- Returns performance accessibility best practices and SEO scores plus core
- Shows the review branches you or the owner have created and how many files
- Each entry maps a public route e g about to the file that renders it
- Start here returns each website s ID and name so you can audit or optimize
- Returns the full content of a file so you can inspect it before proposing a
- Returns a score plus per page state and issues worst first titles meta
- Reports broken includes missing layouts template cycles and invalid
- Returns the changed files and their before after i e exactly what the
- Applies a fix the branch is REQUIRED and must not be main the change
Proposed setup considerations
The research notes below describe a possible setup. They are not instructions for an available Grunt connection; actual steps depend on supported operations and provider approval.
- 01
Start with a small, useful task.
Describe a routine step involving Parallel AI. Keep a first version easy to review and change.
- 02
Plan the account access.
Check which Parallel AI account, data, and permissions a safe connection would need. Start with the least access that can do the job.
- 03
Test the details together.
Validate a sample against a disposable account. Confirm which records are read, what could change, and how errors or duplicates are handled.
- 04
Keep the say-so.
Decide which steps could run on their own and which would need your review. Expand access only when the workflow needs it.
RESEARCH NOTES
Questions about the concept.
These notes describe possible uses. Customer availability remains unverified.
What could a Grunt integration do with Parallel AI?
The documented operations include Agent: Returns when the conversation last changed without loading any message; Agent: Creates a new AI agent with the specified configuration the agent can; Agent: Permanently deletes the specified agent and all associated data; Agent: Returns the full agent configuration including channel settings. These are research candidates from public integration definitions and need validation and Grunt implementation before they can be treated as supported capabilities.
Could I keep a workflow read-only?
The research candidates do not confirm an implemented Parallel AI connection or a read-only access mode. If built, verify least-privilege scopes and reject writes at the app API boundary.
Will every action run automatically?
If a connection is built, choose which steps could run on their own and which should pause for review. Changes should follow the permissions and approval rules you set.
Do I need to review actions in Parallel AI?
A future integration should let you decide which work could run on its own and which steps need your review.
There’s a Grunt for that
A little less busywork.
A little less busywork.
A little more you.
Read the current prelaunch status and confirmed prices.
View pricing status Explore the simulation