[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"productUpdate-en::codatum-agent-execution-budget-and-bigquery-cost:en":3,"productUpdatesNearDate-en:timestamp-2026-10-14T08:00:00+09:00:en":20},{"id":4,"title":5,"slug":6,"description":7,"body":8,"tags":9,"coverUrl":13,"coverMediaUrl":14,"ogImageUrl":15,"createdAt":16,"updatedAt":16,"datePublished":17,"locale":18,"updateType":19},"1q8yFsZmqFZFlyG4hsxmVE","Set an Execution Budget for Codatum Agent's BigQuery Queries","codatum-agent-execution-budget-and-bigquery-cost","Codatum Agent stops BigQuery queries that exceed your execution budget before billing, and Session info now shows bytes billed and slot time.","\u003Cp>Codatum Agent now applies an execution budget—a scan size limit you set—to the BigQuery SQL it runs. You can also see the bytes billed and slot time for BigQuery queries run in a session.\u003C\u002Fp>\u003Cp>The execution budget is set per user and per workspace. If left empty, a budget of 1024 GB is used.\u003C\u002Fp>\u003Ch2>Key features\u003C\u002Fh2>\u003Cul>\u003Cli>\u003Cp>\u003Cb>Stops before billing\u003C\u002Fb>: A query that needs more scan than your budget is not run and is shown as \u003Cb>Not executed\u003C\u002Fb>. Nothing is billed\u003C\u002Fp>\u003C\u002Fli>\u003Cli>\u003Cp>\u003Cb>Approve and re-run\u003C\u002Fb>: Codatum Agent either narrows the query or explains why it needs to re-run it and asks for approval. Once approved, the limit is raised only as far as the scan size shown\u003C\u002Fp>\u003C\u002Fli>\u003Cli>\u003Cp>\u003Cb>Applies to report exploration too\u003C\u002Fb>: The same budget applies to queries run through Explorer, not just SQL\u003C\u002Fp>\u003C\u002Fli>\u003Cli>\u003Cp>\u003Cb>Query cost at a glance\u003C\u002Fb>: \u003Cb>Session info\u003C\u002Fb> in the chat screen shows the total \u003Cb>Bytes billed\u003C\u002Fb> and \u003Cb>Slot time\u003C\u002Fb> for BigQuery queries\u003C\u002Fp>\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>Example use cases\u003C\u002Fh2>\u003Cul>\u003Cli>\u003Cp>Let Codatum Agent work with large tables without worrying about unexpected scan volume\u003C\u002Fp>\u003C\u002Fli>\u003Cli>\u003Cp>Check Session info at the end of an investigation to see how many BigQuery resources the analysis used\u003C\u002Fp>\u003C\u002Fli>\u003Cli>\u003Cp>Agree on a budget across your team and estimate the query cost of using Codatum Agent\u003C\u002Fp>\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>Related documentation\u003C\u002Fh2>\u003Cul>\u003Cli>\u003Cp>\u003Ca href=\"https:\u002F\u002Fdocs.codatum.com\u002Ffeatures\u002Fai-agent\u002Foverview#approval\" target=\"_blank\" rel=\"noopener noreferrer\">Approving operations\u003C\u002Fa>\u003C\u002Fp>\u003C\u002Fli>\u003Cli>\u003Cp>\u003Ca href=\"https:\u002F\u002Fdocs.codatum.com\u002Ffeatures\u002Fai-agent\u002Foverview#user-settings\" target=\"_blank\" rel=\"noopener noreferrer\">Settings\u003C\u002Fa>\u003C\u002Fp>\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>Availability\u003C\u002Fh2>\u003Cp>Change the budget under \u003Cb>Execution budget (BigQuery)\u003C\u002Fb> in \u003Cb>Settings\u003C\u002Fb> in the chat input. You can set any value of 0.1 GB or more in 0.1 GB steps, or turn on \u003Cb>No limit\u003C\u002Fb>.\u003C\u002Fp>\u003Cp>The scan size limit and the bytes billed \u002F slot time display apply to BigQuery only. For other data sources, Codatum Agent explains its reason and asks before running a query it expects to scan a large amount of data. In \u003Cb>Allow all\u003C\u002Fb> approval mode, an over-budget query is re-run with a raised limit without asking.\u003C\u002Fp>",[10],{"id":11,"name":12},"new_feature","New Feature","","https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002F3i3I4Ac2XqphZXMLcucr98\u002Fe0cec360d104af8fa7357b19eaf12587\u002Fproduct-update-agent-execution-budget.png","https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002F2jySu0AJQg0jVwr5l9w8UX\u002F435be9d3b5e5715db806d75037fdb870\u002Fproduct-update-agent-execution-budget-og.png","2026-10-13T23:00:23.604Z","2026-10-14T08:00:00+09:00","en","standard_release",[21,33],{"id":22,"title":23,"slug":24,"description":25,"body":26,"tags":27,"coverUrl":13,"coverMediaUrl":29,"ogImageUrl":30,"createdAt":31,"updatedAt":31,"datePublished":32,"locale":18,"updateType":19},"HBuf9FCqusakQzEtgw6Ij","Upflow and Cursor Added to Agent Connectors","agent-connectors-upflow-and-cursor","Upflow and Cursor are now built-in Agent Connectors, letting Codatum Agent reference Upflow deals and ask Cursor Cloud Agents to open pull requests.","\u003Cp>Upflow and Cursor are now available as built-in Agent Connectors. From Codatum Agent, you can look up opportunities and sales conversations in Upflow, or ask a Cursor Cloud Agent to implement changes.\u003C\u002Fp>\u003Cp>Each user connects with their own account. Upflow previously required contacting us; it can now be added to a workspace and used right away.\u003C\u002Fp>\u003Ch2>Key features\u003C\u002Fh2>\u003Cul>\u003Cli>\u003Cp>\u003Cb>Upflow\u003C\u002Fb>: Connects through Upflow's official remote MCP. Search opportunities and check their current state, review sales conversation analytics, topics, and transcriptions, and search customer voices linked to opportunities\u003C\u002Fp>\u003C\u002Fli>\u003Cli>\u003Cp>\u003Cb>Cursor\u003C\u002Fb>: Use a personal Cursor API key to start Cloud Agents that implement changes in a GitHub repository and open pull requests. Check progress and send follow-up instructions from your conversation with Codatum Agent\u003C\u002Fp>\u003C\u002Fli>\u003Cli>\u003Cp>\u003Cb>Personal authentication\u003C\u002Fb>: Upflow uses each user's own Upflow account, and Cursor uses each user's own API key. What can be read or changed follows the permissions of that account\u003C\u002Fp>\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>Example use cases\u003C\u002Fh2>\u003Cul>\u003Cli>\u003Cp>Combine Upflow sales conversation records with product usage and revenue data in your warehouse to analyze deal status\u003C\u002Fp>\u003C\u002Fli>\u003Cli>\u003Cp>Search customer voices and compare them with product usage to prioritize improvements\u003C\u002Fp>\u003C\u002Fli>\u003Cli>\u003Cp>Turn an issue found during analysis into a fix by asking a Cursor Cloud Agent to open a pull request\u003C\u002Fp>\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>Related documentation\u003C\u002Fh2>\u003Cul>\u003Cli>\u003Cp>\u003Ca href=\"https:\u002F\u002Fdocs.codatum.com\u002Fworkspace\u002Fexternal-integrations\u002Fagent-connectors\" target=\"_blank\" rel=\"noopener noreferrer\">Agent Connectors\u003C\u002Fa>\u003C\u002Fp>\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>Availability\u003C\u002Fh2>\u003Cp>Adding a connector requires permission to change workspace settings. After it is added, each user authenticates with their own account. Actions that Codatum Agent takes through a connector are confirmed according to your approval mode.\u003C\u002Fp>\u003Cp>Cursor's Cloud Agents API is offered by Cursor as a public beta. Codatum is not notified when a Cloud Agent finishes, so ask Codatum Agent to check its status.\u003C\u002Fp>",[28],{"id":11,"name":12},"https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002F4ljyT26hme1gQ73fBlUc7N\u002F676e9a4fb4df51a9a6acb27e80c59d86\u002Fproduct-update-agent-connectors-upflow-cursor.png","https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002F4bWht2ZkvQpt3ebNNKhLoV\u002Fc948d8a0f930e7f1911e7000289b16de\u002Fproduct-update-agent-connectors-upflow-cursor-og.png","2026-10-11T23:00:18.006Z","2026-10-12T08:00:00+09:00",{"id":34,"title":35,"slug":36,"description":37,"body":38,"tags":39,"coverUrl":13,"coverMediaUrl":43,"ogImageUrl":44,"createdAt":45,"updatedAt":45,"datePublished":46,"locale":18,"updateType":19},"1cOiO76cIIABbAphv7ozlu","Choose How Codatum Agent Asks for Approval with Approval Modes","codatum-agent-approval-modes","Approval modes (Manual, Auto, and Allow all) replace Ask and Agent modes and control when Codatum Agent asks before acting.","\u003Cp>You can now choose how often Codatum Agent asks for your confirmation before acting, using approval modes. Approval modes replace the previous Ask and Agent modes.\u003C\u002Fp>\u003Cp>Select an approval mode from the bottom left of the chat input. The default is \u003Cb>Auto\u003C\u002Fb>, and existing users also start in \u003Cb>Auto\u003C\u002Fb>.\u003C\u002Fp>\u003Ch2>Key features\u003C\u002Fh2>\u003Cul>\u003Cli>\u003Cp>\u003Cb>Three approval modes\u003C\u002Fb>: \u003Cb>Manual\u003C\u002Fb> asks before side effects, such as running SQL or editing files. \u003Cb>Auto\u003C\u002Fb> asks only before high-impact actions, such as external writes or deletes. \u003Cb>Allow all\u003C\u002Fb> runs every operation without asking\u003C\u002Fp>\u003C\u002Fli>\u003Cli>\u003Cp>\u003Cb>Decide on the approval card\u003C\u002Fb>: When an operation needs confirmation, choose \u003Cb>Deny\u003C\u002Fb>, \u003Cb>Allow once\u003C\u002Fb>, or \u003Cb>Always allow\u003C\u002Fb> on the approval card\u003C\u002Fp>\u003C\u002Fli>\u003Cli>\u003Cp>\u003Cb>Manage operations allowed without asking\u003C\u002Fb>: Operations you mark \u003Cb>Always allow\u003C\u002Fb> run without asking in any mode from then on. Review and delete them under \u003Cb>Allowed without asking\u003C\u002Fb> in \u003Cb>Settings\u003C\u002Fb> in the chat input\u003C\u002Fp>\u003C\u002Fli>\u003Cli>\u003Cp>\u003Cb>Switch mid-conversation\u003C\u002Fb>: A new mode applies from the next operation, even while a request is running\u003C\u002Fp>\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>Example use cases\u003C\u002Fh2>\u003Cul>\u003Cli>\u003Cp>Work in \u003Cb>Auto\u003C\u002Fb> day to day, confirming only important actions such as writes to external services\u003C\u002Fp>\u003C\u002Fli>\u003Cli>\u003Cp>Switch to \u003Cb>Manual\u003C\u002Fb> when working with a new data source or an important notebook, and review each step\u003C\u002Fp>\u003C\u002Fli>\u003Cli>\u003Cp>Mark operations you use repeatedly as \u003Cb>Always allow\u003C\u002Fb> to skip routine confirmations\u003C\u002Fp>\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>Related documentation\u003C\u002Fh2>\u003Cul>\u003Cli>\u003Cp>\u003Ca href=\"https:\u002F\u002Fdocs.codatum.com\u002Ffeatures\u002Fai-agent\u002Foverview#operation-modes\" target=\"_blank\" rel=\"noopener noreferrer\">Approval modes\u003C\u002Fa>\u003C\u002Fp>\u003C\u002Fli>\u003Cli>\u003Cp>\u003Ca href=\"https:\u002F\u002Fdocs.codatum.com\u002Ffeatures\u002Fai-agent\u002Foverview#approval\" target=\"_blank\" rel=\"noopener noreferrer\">Approving operations\u003C\u002Fa>\u003C\u002Fp>\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>Availability\u003C\u002Fh2>\u003Cp>The approval mode and the list of operations allowed without asking are saved per user and per workspace. The mode you select also carries over to new conversations in that workspace. Which tools Codatum Agent can use is determined by permissions and AI profile settings, not by the approval mode.\u003C\u002Fp>\u003Cp>Routine runs approve every operation automatically; the approval mode and the allowed list do not apply to them.\u003C\u002Fp>",[40],{"id":41,"name":42},"improvement","Improvements","https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002F5wFnTzVgQj3DZpay4wNZTX\u002F79fef0995354e85e2d80fa3d19210dad\u002Fproduct-update-agent-approval-modes.png","https:\u002F\u002Fimages.ctfassets.net\u002Fggtw2zqmifs5\u002FRPgVp7Dwh13QutdYF7s2Z\u002Ff59bd18f565a08667ced1f08eaafc02e\u002Fproduct-update-agent-approval-modes-og.png","2026-10-08T23:00:10.499Z","2026-10-09T08:00:00+09:00"]