- Brand
- Carbide
- Category
- Operations
- Primary Subcategory
- ERP & Operations Resource Tools
Integration details
Description
Carbide helps contracting teams run their business operations. Connect your organization to ChatGPT to manage customers and requests, book appointments, plan production jobs, coordinate crews, configure integrations, and diagnose operational issues. Every action follows your Carbide organization role, product access, and OAuth permissions. Use only ordinary business contact and project information. Do not submit health information, biometrics, government identifiers, payment-card details, passwords or authentication codes. Connect credential-based services securely in the Carbide web app.
- Integration type
- Plugin
- Verification status
- Not applicable
- Platform
- ChatGPT
- Primary Subcategory
- ERP & Operations Resource Tools
- Secondary Subcategories
- None listed
- Brand
- Carbide
- Access
- Account required
- First tracked
- 2026-11-20
- Tool count
- 338
- Market
- US
The Primary Subcategory used for this profile’s headline score.
Other Subcategories where the Integration is listed.
Not discoverable in ChatGPT yet
ChatGPT now suggests Plugins on its own when they match a user's request, but not every Plugin is being surfaced yet.We haven’t seen Carbide suggested yet. Sign up and we’ll track it, and your Discoverability Score will appear here once ChatGPT starts surfacing it.
What discovery looks like

Get alerts for Carbide
Get updates when Carbide’s Discoverability Score or category rank changes.
Competing in ChatGPT ERP & Operations Resource Tools
View Category338 tools agents can invoke
Mark an organization member active so they can sign in and be scheduled. Pass the membership id from list_organization_members at the top level; there are no input fields. You cannot change your own membership. Only owners and admins may call this.
Add the property at a verified address (address_id from find_or_create_verified_address) to a company in the current organization, creating the property if it is new. Returns property_id and the property's Zillow status.
Add the property at a verified address (address_id from find_or_create_verified_address) to a contact in the current organization, creating the property if it is new. Returns property_id and the property's Zillow status.
Give an existing member a CSR (dispatcher) profile. Under input pass membership_id (from list_organization_members; the member must be active and must not already be a CSR). Returns the new profile_id and a message; a previously archived CSR profile is restored rather than duplicated. Only owners and admins may call this.
Give an existing member an estimator profile so they can be assigned estimate appointments. Under input pass membership_id (from list_organization_members; the member must be active and must not already be an estimator) and project_area_ids, the project areas they cover (required; ids come from list_project_areas). Returns the new profile_id and a message; a previously archived estimator profile is restored rather than duplicated. Only owners and admins may call this.
Deprecated 2026-11-17; removed on or after 2026-12-17. Use apply_job_revision instead. Pass the import's id as revision_id. Apply an explicitly reviewed ready import once. Pass import_id and future_state or review using IDs from the proposal and current tenant catalogs. A changed job refuses stale application; retry and review again.
Apply an explicitly reviewed ready revision of an existing job once. Pass revision_id and future_state (kind revision) or review (kinds details and additional_work) using IDs from the proposal and current tenant catalogs. A changed job refuses stale application; retry and review again.
Apply a previously simulated production plan after reviewing its conflicts. Pass under input the same changes, the simulation's base_digest and job_ids, and any acknowledgements it requires. Returns the schedule calculation while it is queued; it runs in the background, so poll get_schedule_calculation with its id until its status is succeeded (the plan is applied and result holds the applied scenario) or failed.
Archive a crew by id, hiding it from the crew and scheduling screens while keeping its history and member assignments. Use unarchive_crew to bring it back. Only owners and admins may call this, and only when the organization has the job scheduling product.
Archive a crew member by id, hiding them from the crew and scheduling screens while keeping their history. Archiving is refused while the person is the lead of an active crew: give that crew another lead with update_crew_details first. Use unarchive_crew_member to bring them back. Only owners and admins may call this, and only when the organization has the job scheduling product.
Archive a custom request field. Pass the field id from list_custom_fields at the top level. The field is archived rather than erased, so answers already collected keep their history, and its label becomes free for a new field; its published API ID stays reserved. The call is refused while the field is still referenced by a HubSpot, Pipeline CRM or PaintScout field mapping, by a placeholder token in an email or calendar invite template, or by a published request form version; the error names each place, so remove those references first. Only owners and admins may call this.
Archive (delete) a production job in the current organization, whatever its status. The job, its blocks and its schedule leave the jobs board, the schedule and search, and stop holding crew capacity. Its request can be promoted to production again.
Name the crew members working one day of a production block, replacing that day's names. Days are numbered from 0 over the block's productive working days (the block's `days` list). An empty list means nobody is named that day; reset_block_day with fields ["members"] makes the day use the block's names again.
Replace the assigned crew members for a production block.
Assign a customer service representative to a company.
Assign an estimator to a company.
Assign a customer service representative to a contact, or clear it with null. The user must be an active CSR of the organization.
Assign an estimator to a contact.
Assign a customer service representative to a request.
Run Schedule Job on an unscheduled production job: Carbide finds each qualified crew's earliest opening, ranks the crews with the crew routing rules, and places the job on the best crew with a tentative start date. The job needs a work block with labor hours, a production value or a duration, and the organization an active crew. The job is `analyzing` while it runs (usually a few seconds); read the result with `get_job`: `tentatively_scheduled` with a crew and start date when placed, or back to `unscheduled` with the reason in its activity when it could not be.
Book an appointment for a request using an exact option returned by find_appointment_options: pass its estimator_id as estimator_user_id and its start and end as start_at and end_at. mode first books the request's first appointment and never moves a booked one; repeating the exact same call returns the booked appointment, so a retry is safe. A request may hold several booked appointments: mode add books another one and leaves the others as they are, while schedule (the default) without an appointment_id books the request's appointment or moves its booked in-person appointment. appointment_type virtual (always added) needs location_address_id, the verified address the estimator holds it from; invite_customer (off by default) adds the customer to the event and its messages, and video_meeting (following invite_customer when left out) creates a Google Meet or Teams link on the organization calendar. Carbide saves the booking first and adds its Google or Outlook events moments later; a calendar that refuses the event doesn't undo the booking, and the appointment's calendar sync state shows it.
Cancel an appointment and record the cancellation details. A booked appointment needs a cancellation_reason. An appointment the estimator removed from their own calendar is already cancelled and awaits a reschedule; cancelling it again with outcome cancelled makes it definitive (the request is cancelled and the chosen messages are sent) and keeps its recorded reason unless you pass one.
Stop a running HubSpot sync. Pass, under input, sync_type — people, companies or deals. The queued work is dropped and the record type's sync state is marked cancelled; records already synced stay. Calling it when nothing is running is harmless. When HubSpot is not connected the call fails with "Connect HubSpot first." Only owners and admins may call this.
Cancel the Pipeline CRM sync that is running for one record type. Pass, under input, sync_type: people, companies or deals. There is no id — it acts on the organization's Pipeline CRM connection. The queued job is cancelled and the sync summary under provider_metadata.pipeline_crm is marked cancelled; records already imported stay, and the cursor is left where it stopped so the next run picks up from there. Use it when a sync is stuck or when trigger_pipeline_crm_sync reports one already running. Only owners and admins may call this.
Move a production job to another status, as the job page does. Pass the job id at the top level and, under input, status (unscheduled, tentatively_scheduled, scheduled, in_progress, completed, paused or canceled) and an optional comment for the job's activity. tentatively_scheduled, scheduled, in_progress and completed need the job's planned start date; set it with update_job_plan first.
Re-check an email sending domain with the email provider and store what comes back. Pass, under input, domain_id — an id from list_sending_domains — and optionally trigger_verification: true, which asks the provider to start verifying the domain when it has not been verified yet. Use it after the DNS records from list_sending_domains have been created at the customer's DNS host. The result is the refreshed domain with its status, dns_records and email_sending_identities; when the domain reaches verified, the pending sender (role default_candidate) becomes the organization's default sender in the same call. Only owners and admins may call this.
Check whether the Carbide id and URL properties exist on the HubSpot contact, company and deal objects. There are no arguments. Carbide writes its own record ids and deep links into carbide_contact_id, carbide_company_id, carbide_request_id, carbide_job_id and the matching _url properties so records can be matched back and opened from HubSpot; syncing is unreliable until they are all there. The result reports each property and whether it is complete. Run sync_hubspot_connection_id_properties to create the missing ones. When HubSpot is not connected the call fails with "Connect HubSpot first." Only owners and admins may call this.
Mark a production job checklist item complete.
Drop an estimator's personal appointment limit so the organization limit from get_organization applies to them again. Pass, under input, estimator_profile_id from list_estimator_profiles. Returns entry_id, the estimator profile id, and a message with an empty value and unit. Only owners and admins may call this.
Clear an estimator's territory, removing every ZIP code and the service-area map drawn from them. Pass estimator_profile_id under input, an id from list_estimator_profiles. Automatic assignment then stops matching them on location. Only owners and admins may call this.
Forget where the last HubSpot sync of one record type stopped. Pass, under input, sync_type — people, companies or deals. The next updates run then starts from the beginning instead of from the last successful point, which is what to do when records were missed or a mapping changed and history has to be re-read. It cannot run while that record type is syncing — cancel the run first. When HubSpot is not connected the call fails with "Connect HubSpot first." Only owners and admins may call this.
Forget where the last Pipeline CRM sync of one record type stopped, so the next run starts from the beginning instead of from the last change. Pass, under input, sync_type: people, companies or deals. There is no id — it acts on the organization's Pipeline CRM connection. Nothing is deleted and no sync starts: follow it with trigger_pipeline_crm_sync to re-import everything. Use it when records are missing or stale because an earlier run skipped them. Only owners and admins may call this.
Complete a PaintScout import once with reviewed details and customer/property IDs. Optionally attach to its eligible request or auto-schedule. Creation is transactional; auto-schedule refusals leave no job.
Change one CRM field mapping. Pass id — the id of the row from list_crm_field_mappings — at the top level, not under input, and, under input, only the parts that change: carbide_resource, carbide_field and carbide_field_label for the Carbide side, hubspot_object, hubspot_property, hubspot_property_label, hubspot_property_type and hubspot_field_type for the remote side, or position to reorder. Anything left out keeps the value it has. Take new values from the same options tools create_crm_field_mapping uses, and expect the same refusals for managed remote properties and incompatible types; a new HubSpot hubspot_property needs its hubspot_property_type, unless the row already has one. The provider, direction, and connection, integration or audience a mapping belongs to cannot be changed: remove the mapping and create the one you want instead. Only owners and admins may call this.
Set the From address customer email is sent from. Pass, under input, email and optionally display_name (the name customers see) and reply_to. The email must be on a domain the organization controls — public mailbox providers such as gmail.com or outlook.com are rejected, because they cannot be verified. Carbide registers the domain with the email provider if it is new and returns the identity together with its sending_domain, including the DNS records to create at the customer's DNS host. The identity stays default_candidate, and email keeps going out from Carbide's own address, until the domain verifies; call check_email_sending_domain_verification once the DNS records exist, and the identity then becomes the default. Only owners and admins may call this.
Choose which HubSpot pipeline and deal stage a Carbide event moves a deal to. Pass id — the id of the row in get_hubspot_connection's deal_stage_settings list whose lifecycle is the event you mean — at the top level, not under input, and, under input, enabled: whether the event moves the deal at all, plus pipeline_id, pipeline_label, stage_id and stage_label copied from one entry of list_hubspot_connection_deal_stages (the entry's entry_id and name are the stage, its pipeline_id and pipeline_name the pipeline). The five lifecycles are inbound_draft_creation (a deal arriving from HubSpot creates a draft request in Carbide), request_creation (a request is created in Carbide), appointment_booked, appointment_canceled and request_closed. There is one row per lifecycle and no tool to create one: HubSpot seeds all five when it is connected. Only the fields passed change, so read get_hubspot_connection first. The result is the deal-stage row. Only owners and admins may call this.
Turn syncing on or off for one HubSpot record type. Pass id — the id of the row in get_hubspot_connection's sync_states list whose sync_type is the record type you mean (people, companies or deals) — at the top level, not under input, and, under input, inbound_enabled: whether HubSpot records flow into Carbide, and outbound_enabled: whether Carbide records are pushed to HubSpot. Both are optional and only the ones passed change, so read get_hubspot_connection first to see what is already on. There is exactly one row per record type, and there is no tool to create one: HubSpot seeds all three when it is connected. Turning a record type off does not undo what has already synced, and turning it on does not backfill — run trigger_hubspot_connection_sync for that. The result is the sync-state row. Only owners and admins may call this.
Set the Meta (Facebook) pixel the organization's public booking pages report conversions to. Pass, under input, pixel_id — the numeric pixel id from Meta Events Manager, digits only, with no spaces or letters. Calling it again replaces the pixel id already stored. The result is the integration row, with the pixel id under provider_metadata.meta. Use disconnect_integration to stop reporting to Meta. Only owners and admins may call this.
Choose when Carbide creates the first PaintScout quote for an unlinked request. Pass id — the PaintScout integration id from list_integrations — at the top level and trigger under input: request_created or appointment_booked, with optional quote_status (draft by default). Pending creation retains the settings captured when triggered. Disabling outbound quotes cancels pending creation. The setting applies to future request activity and does not queue older requests when changed. Existing linked quotes continue to update. A new quote's customer and company are created in PaintScout with it when they aren't linked yet, even where people or companies outbound sync is off. The current choice is provider_metadata.paintscout.quote_creation_trigger in list_integrations. Only owners and admins may call this.
Turn one direction of one PaintScout record type's sync on or off. Pass id — the id of the paintscout row from list_integrations — at the top level, not under input, and, under input, sync_type: people, companies or quotes (quotes support outbound only); direction: inbound to pull that record type from PaintScout into Carbide, or outbound to push Carbide's changes to PaintScout (only the fields that changed, Carbide winning a conflict); and enabled: true or false. One call sets one direction of one record type, so changing both directions takes two calls; the current flags are under provider_metadata.paintscout in list_integrations. Enabling inbound needs the PaintScout webhook in place and fails with "Create the PaintScout webhook under Real-time sync before enabling inbound sync." — call create_paintscout_webhook first, then repeat this call. Enabling inbound also queues the first sync of that record type. Only owners and admins may call this.
Connect a custom domain to the organization's public booking page. Pass, under input, hostname — a subdomain such as book.example.com. Apex domains (example.com) are rejected, as are hostnames already connected to another organization. Carbide registers the hostname with its certificate provider and returns the same settings map as get_custom_booking_domain_settings, with the hostname in candidate_domain and status pending. Tell the customer to create a CNAME record at their DNS host from that hostname to cname_target; connecting a second hostname replaces any candidate that has not activated yet. Then call refresh_custom_booking_domain until the domain reports active. Only owners and admins may call this.
Finish connecting Google Analytics by choosing which data stream Carbide sends events to. Pass, under input, ga_account_id, ga_property_id and ga_data_stream_id — the three ids of one stream from list_google_analytics_stream_options. Somebody has to authorize Google Analytics in a browser first; until they have, the stream options report authorized false and this call fails. Carbide creates the measurement-protocol secret in Google Analytics itself and keeps it encrypted, so no secret is asked for or returned. The result is the connection with its ids, measurement_id and status; calling it again moves the connection to a different stream. Only owners and admins may call this.
Start, cancel or reset a PaintScout sync. There is no id — it acts on the organization's PaintScout connection. Pass, under input, sync_type: people, companies or quotes, and operation: start to queue an inbound contact or company sync now (the inbound sync has to be enabled first, or the call fails with "Enable PaintScout <type> inbound sync first."), cancel to stop the run in progress, or clear_cursor to forget where the last successful run stopped so the next start re-reads everything from PaintScout. clear_cursor is refused while a run is active — cancel it first. The result is a status: queued, cancelled or cleared; watch the run itself under provider_metadata.paintscout in list_integrations. Without a PaintScout connection the call fails with "Connect PaintScout first." Only owners and admins may call this. Quote sync is outbound-only: starting an inbound quote sync is refused; cancel and clear_cursor remain available to clean up legacy quote runs.
Create an appointment reminder. Pass under input: hours_before (required, 1 to 720, unique within the organization) and enabled (defaults to true). Only owners and admins may call this.
Create a scheduled production block for a job. Without producer_slots the block gets a default team from the organization's role settings: crew_size seats (3 when omitted), one for each producing role required per job that fits the block's trades, the rest taking turns between the fitting producing roles with no maximum per job, else going to the first producing role with no maximum; when every producing role has a maximum, the rest fill the room those maximums leave and the team stops there. The job's producers take seats of their own role, adding one when needed, and crew_size becomes the number of seats. A slot sent without role_id takes its named person's role, or the block's extra-seat role when it names nobody, and keeps no role only when no role has room for it. A role's seats may not exceed its maximum per job.
Add a booking reminder emailed to customers who have not booked hours_after hours (1 to 720) after submitting a request. Pass hours_after and optionally enabled (defaults to true) under input. Each hours_after value may only be used once per organization. Only owners and admins may call this.
Create a company customer record. A csr_user_id must be an active CSR, and address_id and estimator_profile_id must belong to the organization.
Add an email address to a company.
Add a phone number to a company.
Create a contact customer record. A company_id must be a company of the organization; selected_place takes the place resolve_address returned.
Carbide ChatGPT Plugin FAQ
How the directory, categories and Discoverability Score work.
Read the methodologyHow do I improve Carbide's ChatGPT Plugin discoverability?
The levers are the listing surface agents actually read: names, descriptions, keywords, tool metadata, and registry health. Which lever matters depends on where discovery breaks, which is what continuous measurement shows.
What are Carbide alternatives on ChatGPT?
As of 2026-11-20, Carbide competes with (주)한산테크 MES, 101, 1barn, 1XA ERP, ADI Admin, AeroPlan IQ Connect, Aginera, Agriax and 155 more in ChatGPT ERP & Operations Resource Tools, ranked by public Discoverability Score.
Where is this profile measured?
This profile uses the geography attached to the latest public registry snapshot: US. Locale tags are intentionally omitted.