Answered from these docs only, by a model that cannot see your account. Check the pages it cites.

What Roiva keeps from each platform

For every platform Roiva connects to, with those it can't connect yet marked: what a sync reads, which records and fields Roiva keeps, and for how long.

View as Markdown View as JSON

Answered from these docs only, by a model that cannot see your account. Check the pages it cites.

Generated from what Roiva ships, on every deploy.

What Roiva reads from each platform it connects to, and what it keeps. It's written for a security or vendor-risk review that has to answer whether Roiva holds your customers' or your staff's personal data, and where.

If your review hasn't cleared yet, an account can start without any of this: with Aggregate Figures Only on, it holds only the figures your people type, import, or confirm from a Roiva suggestion, and can't be connected to anything. See Aggregate Figures Only: a note for your risk team.

What a sync keeps

A sync reads the records it needs from the platform, such as tickets, deals or bills, and keeps two things:

  • Records, trimmed as they're saved to the fields listed in the second table. The rest of the platform's record, such as a ticket's subject and messages, a contact's email or a customer's address, is dropped, never stored. Beside those fields Roiva keeps values it works out from them, such as a date, an amount and its currency or whether a deal was won, and, where the last column says so, a value from beyond them.
  • Totals, such as a month's ticket count or spend. A platform with nothing in the second table keeps only totals.

A cost or value entry booked from a record keeps its reference, date and amount, and can keep the vendor's or customer's name. A sync's history lists the deals, customers and vendors it couldn't match to an initiative, by name. The connection itself keeps the name of what it's connected to, such as a company or workspace. For Salesforce that's the name of the person who connected it, and for Google Workspace and Microsoft 365 their email address and the names and owners of the calendars they can see. A few pickers on the form for linking a connection to an initiative, such as a help desk's list of agents, read names live from the platform; they're cached for up to an hour and not kept.

In your account, Owners and Admins can open synced records, as can a finance reviewer an Admin names — inside a month's finance evidence packet, not through the connection pages. Every member sees the totals.

How long Roiva keeps it

None of it is deleted on a schedule: it's your account's data, kept while the account is. Some syncs also remove a record the platform has deleted, the next time they run.

  • Disconnect stops syncing and keeps everything, including the encrypted credentials, so the connection can be brought back. For Google Workspace, Jira, QuickBooks, Slack, Xero and Zendesk it also asks the platform to revoke Roiva's access (for Zendesk, a connection that signed in; delete an API token in Zendesk's Admin Center); for the rest, remove Roiva in the platform's own settings too. Atlassian doesn't say whether a grant was there to end, so for Jira remove Roiva in your Atlassian app settings too if you need to be sure.
  • Delete Connection asks the same platforms to revoke Roiva's access, then deletes the connection, its credentials, its records and its sync history, except a bill an allocation rule split. The totals and the cost and value entries it booked stay on your initiatives.
  • Deleting the account deletes all of it. Database backups are kept for 30 days.

Managing a connection and Your data: exports and deletion cover each step, and What each platform books covers what a record books once it's kept.

By platform

Marked coming soon: Brex and Ramp. They can't be connected yet, as the Integrations page says: Roiva's app for each is waiting on the platform's approval. Each row says what Roiva will keep once the platform can be connected.

Platform What Roiva reads What it keeps
American Express Nothing: you upload statement files. Each row's cardholder name and address are dropped as the file is read, and the file itself isn't kept. Records of card transactions, plus the totals worked out from them.
Anthropic Usage by model and cost by workspace for the last 12 months, cost by API key for the last year, and daily Claude Code usage for the last 3 months, from the Admin API. Never a prompt or a response. The Claude Code report names each user by email or API key name; Roiva adds each day up and drops the names before anything is saved. Records of cost lines, usage by model, cost by API key and daily Claude Code totals, plus the totals worked out from them.
AWS Daily cost by service from Cost Explorer for the last 12 months, and the same filtered by a cost-allocation tag where an initiative names one. With the optional Amazon Quick permissions, Quick's lists of users, groups and group members, to count who holds each subscription. Records of daily cost by service, Amazon Quick seat counts and Amazon Quick seat counts per group, plus the totals worked out from them.
Azure Monthly cost by service from Cost Management for the last 12 months, and the same filtered by a tag where an initiative names one. Records of monthly cost by service, plus the totals worked out from them.
BigQuery Job counts, bytes processed and slot time from INFORMATION_SCHEMA.JOBS, added up by month inside BigQuery, for about the last six months. Never a query's text or who ran it. Your own tables only through the metric queries you write, and from those only the period, value, entity and sample_size columns, one row per month. Totals only, such as a month's count or spend. No record is kept. For each metric query you write, every version of its text and the rows each version returned: a month, a number, a record count, and a breakdown's value, which is whatever the query selects as its entity. These stay after the connection is deleted, because the figures computed from them point to them.
Brex (coming soon) Card transactions posted in the last 12 months. Nothing about cardholders, and no expense reports. Records of card transactions, plus the totals worked out from them.
ChatGPT Enterprise Daily Codex totals for the workspace for the last 3 months, asked for as workspace totals so no developer is named in them. Each member's daily ChatGPT and Codex usage, which names the member by ID and email: Roiva replaces the ID with its own digest and keeps no email or name. The workspace's member list, from which it keeps two counts. Records of daily Codex totals, daily ChatGPT counts, ChatGPT users by month and seat counts, plus the totals worked out from them.
Claude Enterprise Daily analytics for each member for the last 3 months, of which Roiva uses the Claude Code figures and the number of chat messages, the organization's daily summary of assigned seats, and, as far back as Anthropic's reports go, what usage cost by member, user group and product, and each member's message counts in each chat Project. Each member row names the member by email; Roiva adds Claude Code up by day, keeps a digest of each member who chatted, and drops the names before anything is saved. Records of daily Claude Code totals, daily Claude chat totals, Claude chat users by month, monthly Claude Enterprise seat counts, monthly cost by Project and product and daily cost by user group, plus the totals worked out from them.
Databricks Monthly cost by product, at list price, from Databricks' system.billing tables, for the last 12 months. Never who ran a job or what it was called. Your own tables only through the metric queries you write, and from those only the period, value, entity and sample_size columns, one row per month. Records of monthly cost by product, plus the totals worked out from them. For each metric query you write, every version of its text and the rows each version returned: a month, a number, a record count, and a breakdown's value, which is whatever the query selects as its entity. These stay after the connection is deleted, because the figures computed from them point to them.
Freshdesk Tickets updated in the last 12 months, with their response and resolution times, and the IDs of deleted and spam tickets so those can be removed. Freshdesk sends each ticket's subject and requester; Roiva keeps only the fields the field-by-field table lists. Every solution category, folder and article, without the article's text, to count the published articles customers can see. Records of solution articles and tickets, plus the totals worked out from them.
Gemini Code Assist Daily totals of code suggestions offered and accepted, and lines of code accepted, from Cloud Monitoring for the last six weeks, summed across every IDE and programming language. Records of daily usage totals, plus the totals worked out from them.
Gemini in Google Workspace Google's Gemini audit log: an entry for each thing someone did with Gemini, naming the person by email and Google profile ID, the app and the kind of action. Roiva replaces the person with its own digest of the profile ID before anything is saved. Records of Gemini users by month and Gemini audit log read by month, plus the totals worked out from them.
GitHub From the repositories the connection can see, pull requests closed and deployments made in the last 6 months, with each merged pull request's line counts; each organization's daily Copilot totals and billing for the last 3 months; and its Copilot seat list, each night. Never a pull request's description, reviews or comments. Records of pull requests, deployments, daily Copilot totals, monthly Copilot seat counts and billing lines, plus the totals worked out from them.
Glean Glean's Insights overview for each calendar month, asked for with per-user Insights turned off, so that it lists no one. Roiva reads only the number of people who used Glean's Assistant and keeps no name or email. Records of Glean Assistant users by month, plus the totals worked out from them.
Google Cloud Your budgets, and monthly cost by project and service from your Cloud Billing export in BigQuery. The first sync covers 12 months and later ones the last two. Labels only where a budget filters on them. Records of budgets, plus the totals worked out from them.
Google Workspace Events on the calendars you choose, from 12 months back to the end of this month, including all-day, out-of-office and focus time. Google sends each event's title, description and attendees; Roiva keeps none of them. Records of calendar events, plus the totals worked out from them.
HubSpot Deals, contacts and tickets, all of them, with only the properties the field-by-field table lists, so never a contact's name, email or phone or a ticket's subject or content. Where proposal stages are chosen on the connection page, each deal's date of entering them, kept only as the dates on the deal. With the optional marketing permissions, campaigns, marketing emails and blog posts from the last 12 months, and monthly website sessions by traffic source. Records of deals, contacts, tickets, campaigns, marketing emails and blog posts, plus the totals worked out from them.
Intercom Conversations created in the last 12 months. Intercom sends each conversation with its messages and the people in it; Roiva keeps only the fields the field-by-field table lists. Records of conversations, plus the totals worked out from them.
Jira Issues moved to Done in the last 3 months, with only the fields the field-by-field table lists and your site's story-point field. Never an issue's description or comments. Records of issues, plus the totals worked out from them.
Looker The saved Looks an initiative names, each run as the API user. Roiva keeps only each result's numbers, by month and column, and drops its text columns. Totals only, such as a month's count or spend. No record is kept.
Make Your team's scenarios and their run logs from the last 60 days. Roiva keeps run counts and credits per scenario, by name. Totals only, such as a month's count or spend. No record is kept.
Microsoft 365 Events on the calendars you choose, from 12 months back to the end of this month. Microsoft sends each event's subject, organizer and attendees; Roiva keeps none of them. Teams notifications, if you add a webhook, post the same updates as Slack. Records of calendar events, plus the totals worked out from them.
Microsoft 365 Copilot Microsoft's Microsoft 365 Copilot usage report: a row per licensed user with their name (or Microsoft's hash of it, where names are concealed) and the last day they used Copilot in each app. Roiva replaces the name with its own digest before anything is saved. Records of Copilot license counts and Copilot users by month, plus the totals worked out from them.
n8n Workflows, and executions from the last 12 months without their data: Roiva asks n8n to leave out what each run processed. It keeps run counts per workflow, by name. Totals only, such as a month's count or spend. No record is kept.
NetSuite Invoices that aren't voided, dated in the last 12 months, and vendor bills and credit card charges that aren't voided, dated in the last 15, where one first synced when already more than 12 months old only counts active suppliers and books no cost, nine columns each; the base currency the books are kept in; monthly revenue and expense totals for the last 12 months; and the chart of accounts, departments, classes and locations; purchase orders dated in the last 12 months, with their status; and each vendor bill line's amount and the department, class and location it's tagged with, by name, for AI Spend's departments. Never a line's memo or description, an address or a contact. Records of invoices, vendor bills, credit card charges, purchase orders and income statement totals, plus the totals worked out from them.
OpenAI Usage by model and cost by line item and project for the last 3 months, from the Admin API. Never a prompt or a response, and never which user or API key made a request. Records of cost lines and usage by model, plus the totals worked out from them.
Pipedrive Deals, open, closed and archived, all of them, and their pipelines and stages. Pipedrive sends its whole deal record; Roiva keeps only the fields the field-by-field table lists, plus any custom field named where it's linked as a value source. Where proposal stages are chosen on the connection page, each deal's change history (recents:read): Roiva keeps only when the deal entered each stage, never the other changes or who made them. Records of deals, plus the totals worked out from them.
QuickBooks Invoices and bill payments dated in the last 12 months; bills and purchases dated in the last 15, where one first synced when already more than 12 months old only counts active suppliers and books no cost; the profit and loss report for each of the last 12 months; and the chart of accounts, classes, locations and budgets. QuickBooks sends each document's contact details; Roiva keeps only the fields the field-by-field table lists. Purchase orders dated in the last 12 months, counted and never booked. Records of invoices, bills, purchases, bill payments, purchase orders and profit and loss reports, plus the totals worked out from them.
Ramp (coming soon) Card transactions from the last 12 months. Ramp sends each cardholder's name and department with them; Roiva keeps neither. Records of card transactions, plus the totals worked out from them.
Salesforce Opportunities and cases, all of them, and Messaging sessions from the last 12 months with the Omni-Channel work records that say who took each one, with only the fields the field-by-field table lists, plus any custom opportunity field named where it's linked as a value source. Where proposal stages are chosen on the connection page, the stage history rows for those stages (opportunity, stage and date), kept only as the dates on the opportunity. Never a case's subject or description, a conversation's messages, or a contact. Records of opportunities, cases and Messaging sessions, plus the totals worked out from them.
Slack The list of channels Roiva can post to. Never a message: Roiva doesn't have the permission to read them. It posts the updates you choose, such as initiative changes, approvals and notes, which name the people involved, to the channel you pick. Nothing read from Slack. The workspace's name, the channel list, and the bot token, encrypted.
Snowflake Credits used, query counts and average run times from the ACCOUNT_USAGE views, added up by month inside Snowflake, and daily spend by usage type (Cortex AI by service type) from ORGANIZATION_USAGE, for the last 12 months. Never a query's text or who ran it. Your own tables only through the metric queries you write, and from those only the period, value, entity and sample_size columns, one row per month. Records of daily spend by usage type, with Cortex AI by service type, plus the totals worked out from them. For each metric query you write, every version of its text and the rows each version returned: a month, a number, a record count, and a breakdown's value, which is whatever the query selects as its entity. These stay after the connection is deleted, because the figures computed from them point to them.
SQL Server Your own tables only through the metric queries you write, and from those only the period, value, entity and sample_size columns, one row per month. Never a view's text: where the login can read one, only a digest computed on your server. Totals only, such as a month's count or spend. No record is kept. For each metric query you write, every version of its text and the rows each version returned: a month, a number, a record count, and a breakdown's value, which is whatever the query selects as its entity. These stay after the connection is deleted, because the figures computed from them point to them.
Stripe Successful charges from the last 12 months, each with its customer; every subscription; and, to work out tax, the invoice payments and checkout sessions from the same months, which aren't kept. Stripe sends each customer's email and address; Roiva keeps only their name. Records of charges and subscriptions, plus the totals worked out from them.
Xero Authorized and paid sales invoices dated in the last 12 months; bills and spend-money transactions dated in the last 15, where one first synced when already more than 12 months old only counts active suppliers and books no cost; the profit and loss report for each of the last 12 months; and the chart of accounts, tracking categories and budgets. Xero sends each document's contact details; Roiva keeps only the fields the field-by-field table lists. Purchase orders dated in the last 12 months, counted and never booked. Records of invoices, bills, purchase orders, spend-money transactions and profit and loss reports, plus the totals worked out from them.
Zapier Nothing: a step you add to a Zap posts each run to Roiva. Records of Zap runs, plus the totals worked out from them.
Zendesk Tickets created in the last 12 months, with each ticket's metrics, from Zendesk's incremental export. Zendesk sends each ticket whole, subject, description and requester included; Roiva keeps only the fields the field-by-field table lists. Every Help Center article, without its body, and the user segments that say who can see one, to count the published articles customers can see. Records of Help Center articles and tickets, plus the totals worked out from them.

Each kind of record, field by field

The fields are the ones Roiva keeps from the platform's own record; everything else in it is dropped before the record is saved. The last column names anything in those fields that can identify a person or hold free text, and anything kept beside them.

Record Fields kept Names and free text
American Express card transactions Date, Description, Appears On Your Statement As, Amount, Reference, Category, Account # The merchant's name and descriptor, which can name a person trading on their own. The card by the last digits American Express prints, never the cardholder.
Anthropic cost lines model, token_type, cost_type, amount, currency, description, workspace_id, starting_at, ending_at, context_window, service_tier, speed No person. Anthropic's description of each cost line, and the workspace it's in, by ID and by the name your team gave it.
Anthropic usage by model model, starting_at, ending_at, uncached_input_tokens, cache_read_input_tokens, output_tokens, service_tier, context_window, workspace_id, cache_creation (ephemeral_1h_input_tokens, ephemeral_5m_input_tokens), server_tool_use (web_search_requests) None.
Anthropic cost by API key starting_at, ending_at, group (api_key_id, workspace_id, model, service_tier, context_window, cost_type, token_type), cost (amount, currency) No person. Each API key's cost by model and day, the workspace it's in, and the key's ID and the name your team gave it. Never who made the key or who it acts as. Under Aggregate figures only, key names are neither read nor shown.
Anthropic daily Claude Code totals date, accepted, rejected, proposals, acceptance_rate, lines_added, lines_removed, sessions, commits, pull_requests, active_users, records_seen None. The day's totals and a count of active users.
AWS daily cost by service date, service, amount, currency, estimated None in the cost records. The cost-allocation tag key and value an initiative names are kept with its settings and on the costs it books, so a tag whose value is a person's name or email keeps it.
AWS Amazon Quick seat counts as_of, professional, enterprise, licensed, inactive, bi_only, other, groups, ungrouped, in_several_groups, identity_region, group_sets (professional, enterprise, groups) None. How many people hold each Amazon Quick subscription in the month, and how many are in no group or several. Roiva reads Quick's list of users to count them and keeps no name, email or ARN.
AWS Amazon Quick seat counts per group as_of, group_key, group_name, professional, enterprise, licensed, bi_only Each Quick group's name, as your Quick admins named it, and how many of its members hold each subscription. A group named after a person keeps that name.
Azure monthly cost by service date, service, amount, currency None in the cost records. The tag key and value an initiative names are kept with its settings and on the costs it books, so a tag whose value is a person's name or email keeps it.
Brex card transactions (coming soon) id, card_id, type, initiated_at_date, posted_at_date, amount (amount, currency), merchant (raw_descriptor, mcc) The merchant's descriptor, which can name a person trading on their own. The card by Brex ID, never the cardholder. The transaction's own description is read and dropped, not kept.
ChatGPT Enterprise daily Codex totals object, start_time, end_time, totals (chats, messages, threads, turns, credits, cost_usd, estimated_cost_usd, uncached_text_input_tokens, cached_text_input_tokens, text_output_tokens, text_total_tokens), code_attribution (lines_of_code (added, removed, committed, total_committed_lines, committed_percentage)) None.
ChatGPT Enterprise daily ChatGPT counts date, chat_users, messages None. How many members sent a ChatGPT message that day, and how many messages.
ChatGPT Enterprise ChatGPT users by month user_key, first_active_date A digest of each member that Roiva made and can't reverse, with no name or email, and the first day they used ChatGPT in the month.
ChatGPT Enterprise seat counts as_of, chatgpt_seats, codex_seats None. How many members hold a ChatGPT seat and how many a Codex-only one.
Claude Enterprise daily Claude Code totals date, accepted, rejected, proposals, acceptance_rate, lines_added, lines_removed, sessions, commits, pull_requests, active_users, records_seen None. The day's totals and a count of active users.
Claude Enterprise daily Claude chat totals date, chat_users, messages, records_seen None. How many members sent a chat message that day, and how many messages they sent.
Claude Enterprise Claude chat users by month user_key, first_active_date A keyed one-way digest of each member who chatted in the month, made from Anthropic's id for the member, and the first day they chatted. No email, name or message is kept.
Claude Enterprise monthly Claude Enterprise seat counts as_of, assigned_seats, pending_invites, monthly_active_users None. Seats assigned, invitations pending and how many members were active in the 30 days before, as of the month's last available day, from Anthropic's organization summary.
Claude Enterprise monthly cost by Project and product date, amount_usd, line_kind, line_id, line_name, estimated No one by name, ID or email. A month's cost for a chat Project, by its ID and the name your team gave it, or for a product. Anthropic's reports list each member's cost and messages; Roiva works out each Project's cost from them and drops the members before anything is saved, so a Project one person uses shows that person's estimated cost under its name.
Claude Enterprise daily cost by user group date, amount_usd, rbac_group_ids No one by name, ID or email. What a day's usage cost for a set of user groups, by each group's ID and the name your team gave it. Anthropic's report lists each member's cost; Roiva adds it up by group and drops the members before anything is saved, so a group of one shows that person's cost under the group's name.
Databricks monthly cost by product period, product, cost None.
Freshdesk solution articles id, title, status, folder_id, folder_name, category_id, agent_id, created_at, updated_at, folder_visibilities, tags The title, tags and folder name, which your team writes, and the author by Freshdesk ID. Not the article's text.
Freshdesk tickets id, status, priority, type, source, created_at, updated_at, responder_id, group_id, product_id, is_escalated, fr_escalated, tags, stats (first_responded_at, resolved_at, closed_at, reopened_at, agent_responded_at, status_updated_at) No subject, description, requester or email address. The responding agent and group by Freshdesk ID. Tags, which are free text your team sets. Beside them, a one-way digest of the requester's ID, keyed to this connection, which only tells whether two tickets came from the same customer.
Gemini Code Assist daily usage totals date, project, code_suggestions_count, code_suggestions_accepted_count, code_lines_accepted_count None.
Gemini in Google Workspace Gemini users by month user_key, first_active_date, last_active_date, apps A digest of each person that Roiva made and can't reverse, with no name or email, the first and latest day they used Gemini in the month and the apps they used it in.
Gemini in Google Workspace Gemini audit log read by month month, read_from, read_through, gap, active_users None. How much of the month's log Roiva has read, and how many people used Gemini in it.
GitHub pull requests id, number, repo, title, state, draft, created_at, updated_at, closed_at, merged_at, additions, deletions, changed_files, user (id, login, type) The title, which is free text, and the author by GitHub ID and username.
GitHub deployments id, repo, environment, production_environment, transient_environment, ref, sha, task, description, created_at, updated_at, creator (id, login) The description, which is free text, the branch or tag, and who started it by GitHub ID and username.
GitHub daily Copilot totals org, day, organization_id, code_generation_activity_count, code_acceptance_activity_count, loc_suggested_to_add_sum, loc_added_sum, daily_active_users, weekly_active_users, monthly_active_users, user_initiated_interaction_count, daily_active_copilot_code_review_users, pull_requests (total_reviewed_by_copilot, total_copilot_suggestions, total_copilot_applied_suggestions) None. Totals for the organization, not for any one person.
GitHub monthly Copilot seat counts org, as_of, seats, active_30d None. How many Copilot seats each organization assigned and how many of their holders used Copilot in the 30 days to the day Roiva read them. GitHub's seat list names each holder; Roiva counts them and keeps no names.
GitHub billing lines organizationName, org, date, product, sku, repositoryName, unitType, quantity, pricePerUnit, grossAmount, discountAmount, netAmount None.
Glean Glean Assistant users by month month, read_through, read_at, chat_users, updated_at, instance None. How many people used Glean's Assistant in the month, and when Roiva read it.
Google Cloud budgets name, displayName, etag, amount (specifiedAmount (currencyCode, units, nanos)), thresholdRules (thresholdPercent, spendBasis), budgetFilter (calendarPeriod, creditTypesTreatment, projects, services) No person. The budget's name and its whole filter, including any labels it filters on, which your team chose.
Google Workspace calendar events id, status, created, updated, recurringEventId, eventType, recurrence, start (dateTime, date, timeZone), end (dateTime, date, timeZone), attendees (responseStatus, organizer, self, optional, resource), conferenceData (conferenceSolution (name)) No title, description, location or attendee. When an event was, how long it ran, how many were invited and accepted, and whether it recurred or was online. Which of your chosen calendars it was on, and a person's calendar ID is their email address.
HubSpot deals id, createdAt, updatedAt, archived, properties (dealname, dealstage, hs_is_closed, hs_is_closed_won, amount, deal_currency_code, closedate, createdate, pipeline, hubspot_owner_id, hs_object_id, hs_lastmodifieddate) The deal's name, which often names the customer, and its owner by HubSpot user ID. The value of any custom property set to match on where it's linked as a value source. A value entry booked from a won deal keeps its name.
HubSpot contacts id, createdAt, updatedAt, archived, properties (lifecyclestage, createdate, hs_lead_status, leadsource, hs_object_id, lastmodifieddate) No name, email or phone. One record per contact, identified by its HubSpot ID, with its lifecycle stage and lead source.
HubSpot tickets id, createdAt, updatedAt, archived, properties (hs_pipeline_stage, createdate, closed_date, hs_ticket_priority, source_type, hs_object_id) None. No subject, content or requester.
HubSpot campaigns id, createdAt, updatedAt, properties (hs_start_date, hs_end_date, hs_name) The campaign's name.
HubSpot marketing emails id, name, publishDate, stats (counters (delivered, open)) The email's internal name. No subject, content or recipients.
HubSpot blog posts id, publishDate, state, created, updated None. No title, content or author.
Intercom conversations id, state, open, priority, created_at, updated_at, admin_assignee_id, team_assignee_id, ai_agent_participated, source (type, delivered_as), statistics (time_to_admin_reply, time_to_first_close, count_reopens, count_assignments, count_conversation_parts, first_close_at, last_close_at), conversation_rating (rating, created_at), ai_agent (resolution_state), tags (tags (id, name)) No message, author, contact or rating comment. The assigned teammate and team by Intercom ID. Tag names. Beside them, a one-way digest of the first contact's ID, keyed to this connection, which only tells whether two conversations came from the same customer.
Jira issues id, key, fields (summary, created, resolutiondate, statuscategorychangedate, assignee (accountId), issuetype (id, name), status (id, name, statusCategory (key)), priority (id, name)) The issue's summary, which is free text. Its assignee by Atlassian account ID, never a name or email.
Microsoft 365 calendar events id, isAllDay, isCancelled, isOnlineMeeting, start (dateTime, timeZone), end (dateTime, timeZone), attendees (type, status (response)), recurrence (pattern (type, interval)), responseStatus (response) None. No subject, organizer or attendee. When an event was, how long it ran, how many attended and responded, and whether it recurred or was online.
Microsoft 365 Copilot Copilot license counts report_refresh_date, enabled_users, active_users_28d, identity None. How many users hold a license and how many used Copilot in the last 28 days.
Microsoft 365 Copilot Copilot users by month user_key, last_activity_date, apps, identity A digest of each user that Roiva made and can't reverse, with no name or email, the last day they used Copilot in the month and the apps they used it in.
NetSuite invoices id, tranid, trandate, foreigntotal, foreignamountunpaid, exchangerate, currency, entity_id, entity_name The customer's company name, on every invoice. A value entry booked from an invoice keeps it.
NetSuite vendor bills id, tranid, trandate, foreigntotal, foreignamountunpaid, exchangerate, currency, entity_id, entity_name The vendor's company name.
NetSuite credit card charges id, tranid, trandate, foreigntotal, exchangerate, currency, entity_id, entity_name, card_account The vendor's company name, and the name of the credit card account in your books.
NetSuite purchase orders id, tranid, trandate, foreigntotal, currency, status, status_name, entity_id, entity_name The vendor's company name.
NetSuite income statement totals The month's revenue, expense and cost of goods sold totals, which Roiva's query adds up inside NetSuite. None.
OpenAI cost lines object, line_item, project_id, api_key_id, start_time, end_time, amount (value, currency) No person. The project it's billed to and the API key it was made with, each by ID and by the name your team gave it. Never who owns the key. Under Aggregate figures only, key names are neither read nor shown.
OpenAI usage by model object, model, start_time, end_time, input_tokens, output_tokens, input_cached_tokens, input_audio_tokens, output_audio_tokens, num_model_requests, project_id, batch None.
Pipedrive deals id, title, value, currency, status, pipeline_id, stage_id, owner_id, probability, is_archived, is_deleted, add_time, update_time, close_time, won_time, lost_time, expected_close_date The deal's title, which often names the customer, and its owner by Pipedrive user ID. The value of any custom field set to match on where it's linked as a value source. A value entry booked from a won deal keeps its title. Not the person or organization the deal is with.
QuickBooks invoices Id, DocNumber, TxnDate, TotalAmt, HomeTotalAmt, ExchangeRate, CurrencyRef (value, name), DepartmentRef (value, name), MetaData (CreateTime, LastUpdatedTime), TxnTaxDetail (TotalTax), LinkedTxn (TxnId, TxnType), Line (Id, LineNum, Description, Amount, DetailType, AccountBasedExpenseLineDetail (AccountRef (value, name), ClassRef (value, name), TaxCodeRef (value, name)), ItemBasedExpenseLineDetail (Qty, UnitPrice, ItemRef (value, name), ClassRef (value, name), TaxCodeRef (value, name)), SalesItemLineDetail (Qty, UnitPrice, ItemRef (value, name), ClassRef (value, name), TaxCodeRef (value, name)), LinkedTxn (TxnId, TxnType)), DueDate, Balance, HomeBalance, CustomerRef (value, name) The customer's name on every invoice, whether or not an initiative uses it. If you invoice individuals, that's a person's name. Each line's description, which is free text. A value entry booked from an invoice keeps the customer's name. No email, address or phone.
QuickBooks bills Id, DocNumber, TxnDate, TotalAmt, HomeTotalAmt, ExchangeRate, CurrencyRef (value, name), DepartmentRef (value, name), MetaData (CreateTime, LastUpdatedTime), TxnTaxDetail (TotalTax), LinkedTxn (TxnId, TxnType), Line (Id, LineNum, Description, Amount, DetailType, AccountBasedExpenseLineDetail (AccountRef (value, name), ClassRef (value, name), TaxCodeRef (value, name)), ItemBasedExpenseLineDetail (Qty, UnitPrice, ItemRef (value, name), ClassRef (value, name), TaxCodeRef (value, name)), SalesItemLineDetail (Qty, UnitPrice, ItemRef (value, name), ClassRef (value, name), TaxCodeRef (value, name)), LinkedTxn (TxnId, TxnType)), DueDate, Balance, HomeBalance, VendorRef (value, name), APAccountRef (value, name) The vendor's name, which can be a person trading on their own, and each line's description. No address or contact details.
QuickBooks purchases Id, DocNumber, TxnDate, TotalAmt, HomeTotalAmt, ExchangeRate, CurrencyRef (value, name), DepartmentRef (value, name), MetaData (CreateTime, LastUpdatedTime), TxnTaxDetail (TotalTax), LinkedTxn (TxnId, TxnType), Line (Id, LineNum, Description, Amount, DetailType, AccountBasedExpenseLineDetail (AccountRef (value, name), ClassRef (value, name), TaxCodeRef (value, name)), ItemBasedExpenseLineDetail (Qty, UnitPrice, ItemRef (value, name), ClassRef (value, name), TaxCodeRef (value, name)), SalesItemLineDetail (Qty, UnitPrice, ItemRef (value, name), ClassRef (value, name), TaxCodeRef (value, name)), LinkedTxn (TxnId, TxnType)), PaymentType, Credit, AccountRef (value, name), EntityRef (value, name, type) The payee's name and type, which can be an employee or a customer, and each line's description.
QuickBooks bill payments Id, DocNumber, TxnDate, TotalAmt, HomeTotalAmt, ExchangeRate, CurrencyRef (value, name), DepartmentRef (value, name), MetaData (CreateTime, LastUpdatedTime), TxnTaxDetail (TotalTax), LinkedTxn (TxnId, TxnType), Line (Id, LineNum, Description, Amount, DetailType, AccountBasedExpenseLineDetail (AccountRef (value, name), ClassRef (value, name), TaxCodeRef (value, name)), ItemBasedExpenseLineDetail (Qty, UnitPrice, ItemRef (value, name), ClassRef (value, name), TaxCodeRef (value, name)), SalesItemLineDetail (Qty, UnitPrice, ItemRef (value, name), ClassRef (value, name), TaxCodeRef (value, name)), LinkedTxn (TxnId, TxnType)), PayType, VendorRef (value, name), CreditCardPayment (CCAccountRef (value, name)), CheckPayment (BankAccountRef (value, name)) The vendor's name, and the name of the bank or card account that paid.
QuickBooks purchase orders Id, DocNumber, TxnDate, POStatus, TotalAmt, VendorRef (value, name), CurrencyRef (value), MetaData (CreateTime, LastUpdatedTime) The vendor's name, which can be a person trading on their own. No lines or address.
QuickBooks profit and loss reports The month's profit and loss report as QuickBooks returns it: account rows and totals, with no customer or vendor breakdown. None.
Ramp card transactions (coming soon) id, state, card_id, merchant_name, merchant_descriptor, merchant_category_code, user_transaction_time, synced_at, entity_amount (value, currency) The merchant's name and descriptor. The card by Ramp ID, never the cardholder. The memo — free text the cardholder writes, so it can hold anything — is read and dropped, not kept.
Salesforce opportunities Id, Name, StageName, IsWon, IsClosed, Amount, CurrencyIsoCode, CloseDate, CreatedDate, OwnerId, Type, RecordTypeId, Probability, attributes (type) The opportunity's name, which often names the customer, and its owner by Salesforce user ID. The value of any custom field set to match on where it's linked as a value source. A value entry booked from a won opportunity keeps its name.
Salesforce cases Id, CaseNumber, Status, IsClosed, Priority, Origin, Type, CreatedDate, ClosedDate, AccountId, OwnerId, attributes (type) No subject, description or contact. The customer's account and the case owner by Salesforce ID only.
Salesforce Messaging sessions Id, Status, AgentType, EndedByType, ChannelType, CreatedDate, EndTime, EndUserMessageCount, CaseId, OwnerId, attributes (type), agent_work (Id, BotId, BotType, UserId, Status, CreatedDate) No messages and no customer. The person or AI agent each session was routed to and the user or queue that owns it, by Salesforce ID only, and the case it was linked to by ID. Beside them, a one-way digest of the customer's Messaging user ID, keyed to this connection, which only tells whether two sessions came from the same customer.
Snowflake daily spend by usage type, with Cortex AI by service type usage_date, usage_type, currency, usage_in_currency None.
Stripe charges id, object, status, amount, amount_captured, amount_refunded, currency, customer, description, invoice, payment_intent, paid, captured, refunded, created The customer's name, or the cardholder's when the customer has none. If you sell to individuals, that's a person's name. The charge's description. A value entry booked from a charge keeps the customer's name. No email, address or card details.
Stripe subscriptions id, object, status, currency, customer, created, canceled_at, cancel_at_period_end, current_period_start, current_period_end, items (data (id, quantity, price (id, unit_amount, currency, product, recurring (interval, interval_count)))) None. The customer by Stripe ID only.
Xero invoices Type, InvoiceID, InvoiceNumber, Reference, Status, Date, DueDate, LineAmountTypes, CurrencyCode, CurrencyRate, SubTotal, TotalTax, Total, AmountDue, AmountPaid, AmountCredited, FullyPaidOnDate, UpdatedDateUTC, Contact (ContactID, Name), LineItems (LineItemID, Description, Quantity, UnitAmount, LineAmount, ItemCode, AccountCode, AccountID, TaxType, TaxAmount, Tracking (TrackingCategoryID, TrackingOptionID, Name, Option)), Payments (PaymentID, Date, Amount, Reference, CurrencyRate) The customer's name on every invoice, whether or not an initiative uses it. If you invoice individuals, that's a person's name. The reference, each line's description and each payment's reference, which are free text. A value entry booked from an invoice keeps the customer's name. No email, address or phone.
Xero bills Type, InvoiceID, InvoiceNumber, Reference, Status, Date, DueDate, LineAmountTypes, CurrencyCode, CurrencyRate, SubTotal, TotalTax, Total, AmountDue, AmountPaid, AmountCredited, FullyPaidOnDate, UpdatedDateUTC, Contact (ContactID, Name), LineItems (LineItemID, Description, Quantity, UnitAmount, LineAmount, ItemCode, AccountCode, AccountID, TaxType, TaxAmount, Tracking (TrackingCategoryID, TrackingOptionID, Name, Option)), Payments (PaymentID, Date, Amount, Reference, CurrencyRate) The supplier's name, which can be a person trading on their own, the reference, and each line's description. No email, address or phone.
Xero purchase orders PurchaseOrderID, PurchaseOrderNumber, Date, DateString, Status, Total, CurrencyCode, UpdatedDateUTC, Contact (ContactID, Name) The supplier's name, which can be a person trading on their own. No lines, address or contact details.
Xero spend-money transactions Type, BankTransactionID, Reference, Status, Date, LineAmountTypes, IsReconciled, CurrencyCode, CurrencyRate, SubTotal, TotalTax, Total, UpdatedDateUTC, Contact (ContactID, Name), BankAccount (AccountID, Code, Name), LineItems (LineItemID, Description, Quantity, UnitAmount, LineAmount, ItemCode, AccountCode, AccountID, TaxType, TaxAmount, Tracking (TrackingCategoryID, TrackingOptionID, Name, Option)) The payee's name, which can be a person, such as an employee paid back, the reference, and each line's description.
Xero profit and loss reports The month's profit and loss report as Xero returns it: account rows and totals, with no customer or supplier breakdown. None.
Zapier Zap runs run_id, zap_id, zap_name, status, occurred_at The Zap's name, which your team chose. Anything else the step posts is dropped.
Zendesk Help Center articles id, title, locale, source_locale, draft, user_segment_id, section_id, author_id, created_at, updated_at, edited_at, outdated, user_segment_ids, label_names The title and labels, which your team writes, and the author by Zendesk ID. Not the article's body.
Zendesk tickets id, status, type, priority, created_at, updated_at, assignee_id, group_id, brand_id, ticket_form_id, support_type, tags, via (channel), satisfaction_rating (score), metric_set (solved_at, reopens, reply_time_in_minutes (calendar, business)) No subject, description, requester or satisfaction comment. The assignee, group and brand by Zendesk ID. Tags, which are free text your team or your triggers set. Beside them, a one-way digest of the requester's ID, keyed to this connection, which only tells whether two tickets came from the same customer.