Privacy notice for candidates and hiring teams.

What Puddle Collects

Puddle may collect account information, customer workspace information, candidate invite details, interview scheduling information, audio and video recordings, transcripts, rubric notes, rubric scores, draft rubric observations, rankings, recommendations, summaries, reviewer activity, device diagnostics, support messages, and security logs.

Customers may also provide role rubrics, job context, candidate materials, source signals, and reviewer notes so Puddle can run a structured interview and produce review materials.

How Puddle Uses Data

• To create and operate candidate interview rooms.
• To record, transcribe, summarize, and organize interview records for the hiring team.
• To apply the customer-defined rubric and generate AI-assisted notes, transcript references, rubric scores, rankings, recommendations, draft rubric observations, and review materials for human review.
• To maintain security, debug the product, prevent misuse, and support customers and candidates.
• To comply with legal, contractual, accounting, or security obligations.

Google Calendar Connections

Google user data is not sold, used for advertising, or used to train general-purpose AI models. Puddle's use and transfer of information received from Google APIs adheres to the Google API Services User Data Policy, including its Limited Use requirements.

When a hiring-team member connects Google Calendar, Puddle receives the member's Google account identity and authorization to list accessible calendars and read calendar events. The member selects a calendar in Puddle. Google grants access to the requested Calendar scopes; Puddle uses the selected calendar for synchronization.

Synchronized event data can include event identifiers, titles, descriptions, organizer and attendee email addresses, dates, times, time zones, recurrence information, and cancellation or update status. Puddle uses this information to identify recruiting meetings, match participants to customer recruiting records, and support interview briefings and follow-up workflows within the connected customer organization.

This Calendar connection requests read-only Calendar access. It does not request Gmail access or permission to create, edit, or delete Google Calendar events. Relevant recruiting information may appear in briefings delivered to authorized members of the customer's connected Slack workspace.

Calendar AI Discovery and Rule Suggestions

When you choose Find interview examples in Calendar setup, Puddle automatically samples up to 100 events from the selected calendar's past 28 days before you choose individual examples. Puddle sends sampled event titles (limited to 240 characters each), durations, recurrence indicators, and temporary sample identifiers to an Anthropic model through Amazon Web Services Bedrock to identify likely recruiting meetings. The sample excludes events marked private, confidential, cancelled, or private-copy events. It does not send event descriptions, attendee/contact records, organizer addresses, Google authorization tokens, or raw Google event identifiers. Titles may still contain names or other personal information; this is not anonymous data.

When you select Calendar examples and choose Use these examples, Suggest rules or Revise suggestions, Puddle sends those examples' titles, your labels, meeting durations, substituted organizer and recurring-series identifiers, and any correction you enter to an Anthropic model through Amazon Web Services Bedrock. Titles and corrections can contain names or other personal information. The rule-suggestion request excludes full event descriptions, attendee/contact records, Google authorization tokens, and event records you did not select. The earlier discovery sample is a separate AI request. Included and excluded examples and your labels help evaluate the proposed rules. Substituted identifiers do not make the entire request anonymous.

Puddle uses the response to propose rules for recognizing your recruiting meetings. You review and confirm the rules into a draft, inspect the matching meetings and recipients, and separately enable automation. Browsing the calendar and applying confirmed matching rules do not themselves invoke this suggestion model. Calendar access remains read-only.

For Calendar discovery and rule suggestions, Puddle uses AWS Bedrock US inference profiles; processing can occur across supported AWS regions within the United States. It is not restricted to the region hosting Puddle's database. Google user data, including discovery samples, selected examples and derived results, is not used to train general-purpose AI models.

Slack Connections

When a customer connects Slack, Puddle stores the workspace installation and its authorization credential. Puddle processes messages that mention the app and direct messages sent to it, interaction payloads, and Slack member identifiers and email addresses needed to associate Slack users with authorized Puddle organization members.

Puddle uses this information to respond to recruiting requests, deliver interview briefings, and support configured recruiting workflows. Content sent to Slack is also processed by Slack under the customer's Slack settings and agreements. Installing the app does not itself grant access to another customer's Puddle data.

Integration Credentials, Disconnection, and Deletion

Puddle stores Google refresh tokens and Slack installation tokens encrypted and associates connections with a customer organization. These credentials allow configured background workflows to operate after the initial connection.

Disconnecting an integration in Puddle removes its stored authorization credential and stops future access through that connection. Disconnection does not immediately erase previously synchronized records, briefings already delivered to Slack, or audit records. Customers may request deletion by emailing hello@usepuddle.com; the retention and deletion provisions below also apply to integration data.

Users can also revoke Google's authorization in their Google Account settings. Workspace administrators can remove the Slack app using Slack's app management controls. Revoking access does not recall information already delivered to another service.

Protection of Google User Data

Puddle encrypts stored Google authorization credentials and associates integration connections with a customer organization. Access to Calendar data and configured workflows is checked against authenticated organization membership and connection permissions. AI discovery and rule-suggestion requests use limited data projections rather than sending complete Calendar records or authorization credentials to the model.

Google API data and data derived from it are not used or transferred for creating, training, or improving foundational or general-purpose AI/ML models, including through third-party providers. Substituted identifiers and limited samples reduce disclosure but do not make event titles or user corrections anonymous. Disconnection, retention, and deletion are described in the sections above and below.

AI Boundaries

Puddle is designed to support structured hiring review. Puddle may generate rubric scores, ranked lists, recommendations, draft observations, and other decision-support materials for the hiring team. Puddle does not itself hire, reject, or communicate employment outcomes to candidates. Hiring companies are responsible for reviewing interview materials, providing any required human review or reconsideration process, and deciding how candidates move through their process.

Facial recognition, emotion recognition, voiceprint identification, attractiveness, facial expression, eye movement, accent, race, color, religion, sex, pregnancy, childbirth or related medical conditions, national origin, citizenship or immigration status where protected, age, disability, genetic information, sexual orientation, gender identity, veteran status, arrest or conviction record where protected, zip code, and other protected characteristics or proxies are not inputs for rubric scores, rankings, recommendations, or job-fit outputs.

Session Diagnostics

Puddle may process limited session data to operate and secure the interview room, such as connection quality, device and browser diagnostics, microphone and camera permission status, session start and end timestamps, reconnects, interruptions, and recording status.

Puddle should not treat session diagnostics as a substitute for job-related answers or customer-defined rubric criteria.

Sharing and Subprocessors

Puddle shares data with the customer that invited the candidate and with service providers that help run the product, such as cloud hosting, authentication, email, live video, storage, analytics, support, and AI model providers.

Puddle does not sell candidate personal information or interview recordings, and does not share candidate personal information for cross-context behavioral advertising. Customer agreements may further restrict how customer and candidate data is processed, retained, exported, or deleted.

Retention and Deletion

Puddle retains candidate data for the period needed to provide the service, support customer review, maintain security records, and satisfy legal or contractual obligations. Customer agreements may set shorter or longer retention periods for particular data categories.

Candidates or customers may request deletion by contacting Puddle. Some records may need to be retained when required by law, customer employment-record obligations, dispute preservation, fraud prevention, backup restoration cycles, or security recordkeeping.

Candidate Rights and Accommodation

The hiring company usually determines the purposes of the interview and is the primary contact for employment-related requests. Puddle helps process candidate requests as required by customer agreements and applicable law.

Candidates may contact Puddle or the hiring company to request access, correction, deletion, accommodation, or an alternative interview process. Candidates should not be penalized for requesting a reasonable accommodation or legally required alternative process.

Jurisdiction-Specific Notices

Some locations may require additional candidate notices, consent language, bias-audit information, automated-decision disclosures, deletion rights, appeal or reconsideration rights, or alternative-process information, especially when AI-assisted scores, rankings, or recommendations are used in hiring review. The hiring company is responsible for providing location-specific notices required for its hiring process, and Puddle is designed to support those workflows where configured.