Data export and conversation review in Tigy: key differences

Review one execution

To investigate a call, use runs and usage. Locate agent, period and execution identifier. Review available records and compare conversation with the external destination outcome whenever an action was involved.

Request an export

In settings, find the data and access area. Request an export, refresh the request list and track its status. When ready, download the file before it expires. It uses JSON, a text format that organizes data by names and values. Requesting an export does not mean the file is already available.

Handle downloaded files

Exports may contain personal information. Store them according to purpose and share only with people who need access. Exporting does not remove your account and should not be a shortcut to distribute a whole team’s records.

Distinguish account and workspace

Personal account, member participation and workspace are different resources. Remove team access through the people area. Before requesting deletion, check ownership and integrations depending on the resource shown in confirmation.

Verify request outcomes

Track state and conditions shown by the interface. Access removal should not be described as immediate purging of shared, financial or call records. Consult applicable documents and accountable owners for purpose and retention.

Which resource should you use to investigate a call?

Locate the run by agent, period and identifier. Compare available records with the task: what was said, which fields were confirmed and the responsible system’s result. Account export serves another purpose and should not be presented as a complete report of every team call.

To copy account data, track the request in settings and download the file when ready, before the stated expiry. It uses JSON, a text format that organizes information by names and values. It is not necessarily a ready-to-read report; check which information it includes. Requesting an export does not mean the file has already been produced.

Choose evidence according to the question you need to answer

Before requesting an export, define the question under investigation. If the issue concerns an agent response in a call, the corresponding run provides the review path. If the purpose is obtaining account data, privacy settings serve a different need. Recordings, transcripts, and export files have different scopes. Choosing the largest available file does not guarantee it contains appropriate evidence.

At a fictional company, an owner wants to understand why the agent confirmed an incorrect appointment. Locate the conversation by agent and time, inspect its transcript, and review the action where available. The calendar system establishes what was recorded. An account export does not replace those checks. In another case, someone wants data provided by the personal export process. The documented route is requesting it through the relevant settings area.

Record purpose before downloading files. A short note can identify resource, period, and question: “check the time confirmed in this conversation.” This helps limit access and sharing. One reviewer may need only a reference and excerpt, while another needs a recording to assess pronunciation. Do not distribute all material to everyone merely for convenience. Evidence selection should reflect each person's actual review task.

If a search does not find expected information, check resource scope and availability at that run state. Absence from one file does not establish absence from every system. Nor does it prove a commercial operation never happened. Find the appropriate source before concluding. Evidence should relate to the question, and conclusions should stay within what it demonstrates. A failed search can indicate the wrong resource rather than missing underlying data.

This choice reduces unnecessary copies and clarifies review. Begin by identifying the correct resource, rather than accumulating files in the hope that one contains the entire story. Keeping scope explicit also lets another reviewer understand why a particular file was used and what remains outside the investigation.

Track the request until the file is available

The documented export process begins in Settings, in the data-and-access area. Request an export and refresh requests to track its state. Creating the request does not mean the file is ready. When available, download the JSON before its indicated expiry. Check interface state instead of assuming completion because the request button was used.

In a fictional example, someone requests data and returns later to check. If state does not yet allow download, there is no need to create many identical requests merely to try accelerating it. Track the existing request through offered options. If an error appears, preserve its description and state for the appropriate investigation without sharing credentials alongside evidence. A reproducible account of the step is more useful than repeated untracked requests.

After downloading, store the file according to purpose. JSON is a structured data format; having the file does not automatically make it an editorial report or complete conversation history. Identify scope and available fields before using it for another question. Do not assert that a data type does not exist in the product simply because it is absent from that export. The export's documented purpose remains relevant when interpreting its contents.

Share only with people needing access for the defined purpose. The file may contain personal information. Technical investigation can often use a reduced description or run reference instead of the complete JSON. Concrete retention and access choices should follow documents and procedures applicable to the organization. Keep downloaded copies included in that consideration rather than treating them as outside the original data process.

Exporting does not delete an account. These are different actions with distinct results and effects. Do not use an export as evidence of removal or assume downloading ended access. If the objective becomes deletion, review that resource and process separately. This distinction prevents navigation mistakes and incorrect expectations about what the request accomplished.

Review a conversation using details available for its channel

To investigate service, select the workspace and locate the run using agent, period, and available filters. Open details and compare the interaction with its expected outcome. Transcript, recording, context, and actions may differ in availability according to state, channel, and configuration. Text chat does not create voice recordings. Interpret absence in the correct context instead of concluding data loss before checking the execution type.

For pronunciation or interruption issues, an available voice recording provides evidence the transcript may not show. A correctly transcribed sentence may still have been difficult to hear. For lookup issues, inspect parameters and results where available. Recording alone does not establish that the tool used the right identifier. Select the evidence combination appropriate to the failure. Different questions about the same run can require different details.

In a fictional call, the customer corrects an order number and the agent answers using the previous reference. Review should examine the correction in conversation and the lookup actually performed. For external writes, check the responsible system. A “completed” conversation state does not prove the task was correct. The caller may have ended the interaction because useful help was unavailable. Treat conversation state and task outcome as separate findings.

Keep the run reference and a scenario description. For reproduction in tests, use fictional information preserving the relevant condition whenever possible. Do not copy a customer's complete record into a test document merely because it seems quicker. Reproduce behavior rather than expanding circulation of individual records. A precise reduced scenario often gives reviewers a clearer basis for understanding the issue.

After correction, repeat the case and a nearby variation. Old evidence explains the failure; the new test establishes whether the change resolved it. Connect them through a clear description without treating an account export as a substitute for call investigation or verification of external effects. This keeps the review tied to observable behavior and appropriate system evidence.

Distinguish account, workspace, and member access

Personal accounts, workspaces, and member participation are different resources. Removing someone from a team does not require deleting the shared workspace. Exporting does not remove an account. Account deletion should not be presumed to immediately eliminate every record connected to collective operations. Documentation advises reviewing the correct resource and dependencies before a deletion request.

At a fictional company, a colleague stops maintaining an agent. The team needs to review access and account-linked integrations. That differs from ending the workspace containing agents, tools, documents, and people. Deleting shared resources may affect everyone's work. Choose the appropriate route for ending participation in member and permission settings according to available options. The action should match the actual objective rather than serve as a broad reset.

If the person owns workspaces, documentation advises resolving that situation before account removal. Review who remains responsible and which external-service credentials or integrations depend on that person's responsibilities. Do not use deletion to investigate an access issue. First check account, invitation, workspace, and permissions; configuration failures need not become destructive changes. Preserve continuity for legitimate users while correcting the specific problem.

Read interface confirmation and verify the selected resource when deletion is genuinely intended. Track the request result rather than assuming immediate completion. Shared, financial, and call records remain subject to applicable retention conditions as described in documentation. Check concrete decisions against relevant documents and procedures instead of turning a navigation guide into a universal data-handling promise. Export, member removal, and deletion need separate interpretations.

For agent-quality review, stay focused on service and necessary evidence. Resource distinctions help request the correct file, share less information, and avoid changes unrelated to investigation. A well-defined analysis produces a verifiable conclusion without conflating personal access, account data, and team operations. Keep unresolved data-scope questions explicit rather than inferring broad outcomes from a single file or interface action.