Ambient Assist AI Privacy Notice
Effective date: November 9, 2026
Ambient Assist AI, L.L.C.
Jurisdiction: Louisiana, United States
201 Rue Beauregard STE 202, Lafayette, LA 70508, US
1 Who we are and when this notice applies
Ambient Assist AI, L.L.C. ("Ambient," "we," "us," or "our") provides Ambient Assist AI. This notice explains how we handle personal information through our website, business communications, accounts, and organization deployments. You can reach us at contact@ambientassist.ai for general questions, support@ambientassist.ai for ordinary support and request intake, privacy@ambientassist.ai for privacy questions, and security@ambientassist.ai for security reports.
An organization subscribes to Ambient Assist AI and chooses how to use it with its enrolled users and devices. The service observes enabled work activity, maintains organizational knowledge, generates business intelligence, and assists users with their work. It can process communications continuously while capture is running, including outside named meetings. Your organization determines the purposes and configuration of its deployment and should give you its own monitoring and privacy notices before collection begins.
For information we process on an organization's instructions, we act as its service provider or processor under our agreement. For our own sales communications, business administration, and applicable security and compliance records, we determine the purposes of our processing. A customer may itself be processing information for another organization; our agreement addresses that relationship. These roles depend on the actual activity and applicable law.
Our initial commercial offering is the cloud organization service and enrolled desktops for US business customers. Personal-desktop deployment and phones are excluded from this offering.
2 Information we receive
Collection depends on the organization's settings, the device's permissions, and the features in use. The categories below describe available processing; they do not mean every category is enabled for every person.
| Category | Information and source |
|---|---|
| Account and organization information | Email addresses, account and membership identifiers, roles, invitations, organization and team information, enrolled-device identifiers and names, device public keys, and sign-in information. Sources include users, customer administrators, and sign-in services. |
| Speech and transcripts | Microphone input and computer audio, including other people on a call or near a microphone, are transcribed on the desktop. Transcripts are sent to the organization's cloud service. Short audio buffers are discarded after transcription unless meeting audio recording is enabled. |
| Desktop activity | Enabled signals can describe application focus, window titles, idle periods, input rhythm, window actions, app switching, dialogs, menus, and scrolling. Input rhythm indicates activity such as typing or mouse use; it does not capture a stream of individual keys or identify mouse-click targets. |
| Screen and clipboard text | Where enabled, text is read from application windows using accessibility features or local optical character recognition, and plain text is read from the clipboard. Administrators can select an application list or enable Read every app to read every otherwise eligible window except private applications. Read every app is off by default. Application text can include content displayed in background windows or unsent drafts. The screen-text feature does not retain the images used for optical character recognition. |
| Meeting recordings | Depending on settings, a meeting may retain audio or audio with video of the computer's virtual screen. Screen video can include multiple monitors. This is separate from ordinary transcription and screen-text extraction. |
| User contributions and generated records | Chat messages, notes, wiki edits, comments, action items, work outputs, timelines, recaps, coaching suggestions, facts, heads-ups, and intelligence records. Some records are inferences or summaries rather than verified statements. |
| Business communications | Information provided through contact forms, demos, support, and commercial correspondence, including email, message text, and the page or button associated with an inquiry; website lead records also include timestamps and confirmation/contact status. Billing uses invoices plus ACH transfers to the business bank account. This method does not authorize automatic account debit. |
| Service operations | Administrative audit events, deletion records, device and job identifiers, errors, and operational measurements. Central traces and metrics use a restricted set of operational attributes rather than prompts, replies, or tool arguments. Service logs can include exception text and some identifying metadata. Trace filtering does not guarantee exclusion of content or personal information from every application or service log, including standard output and error streams. |
Information captured from work applications or conversations may include information about people who are not Ambient users. Depending on the actual content and enabled features, it may include sensitive information such as location, health or financial information, credentials, or private communications. The fact that a category is technically collectable does not authorize a customer to collect it in every circumstance.
On an enrolled desktop, speaker labeling applies to computer-audio transcript lines while capture is running, including outside meetings. It creates temporary speaker representations held in memory to distinguish utterances. The labeling state resets when the next meeting starts; the representations are not stored or matched across meetings. Transcript speaker labels can remain with the transcript.
3 Capture controls and notices
Organization administrators choose the enabled desktop awareness signals, private applications, screen-text settings (an application list or Read every app), recording settings, and microphone and computer-audio rules. Audio rules can differ inside and outside meetings. The organization can permit a user to change a mute setting or hold that setting under its policy. A user's permitted Private and Muted choices survive a restart; an organization-held setting takes precedence while that rule applies.
The desktop window shows capture controls, meeting and recording status, and a card explaining organization settings. A minimized window, static tray icon, or assistant animation should not be treated as an indication that transcription has stopped. A recording indicator and a transcription setting describe different activities. The application can run capture without its main window open.
Private-application exclusions and other protective filters reduce certain capture. They do not guarantee that every sensitive item or incidental conversation will be excluded. Microphones may hear nearby people, and computer audio may include other call participants. The service does not automatically announce its listening or recording to people through Teams, Zoom, Google Meet, or other meeting platforms. The organization and its users must carry out the notice and permission process applicable to their deployment.
4 How we use information
We process customer information to provide the configured service, maintain work and organizational records, answer authorized questions, generate summaries and suggestions, run permitted work functions, provide support, protect the service, and carry out the customer's agreed retention, export, and deletion instructions.
We use our business-contact and administrative information to respond to inquiries, arrange demonstrations, administer subscriptions, manage accounts and billing, address security and privacy requests, maintain appropriate evidence, and comply with applicable legal obligations.
The service is intended to support work and management. Managers may consider information it supplies in their decisions. It does not replace the organization's responsibility for employment decisions, lawful practices, or checking material facts. AI outputs may be incomplete or inaccurate and can reflect an inference about a person's activity rather than an established fact.
We do not use customer information for our own general-purpose model training, advertising profiling, or unrelated cross-customer purposes. Any future such use requires the customer's separate written authorization and lawful arrangements. This statement concerns Ambient's own use. Customer-selected AI providers are addressed separately below.
5 Access within your organization
Your records are part of your organization's deployment. They should not be understood as a private personal assistant space inaccessible to your employer.
Owners and administrators can access substantial user records, including meeting recordings and transcripts, journals, chat, coaching, wikis, timelines, and intelligence through the available service views. They can continue to access retained records of removed members. Ordinary web views do not expose every stored raw observation or the live transcript outside meetings.
Team leads and organization leaders see authorized heads-ups and intelligence. Their assistants can retrieve more detailed records within their permitted organizational reach, including journals, chat, coaching, wikis, work records, and meeting speech transcripts. Those assistant tools do not retrieve meeting recordings or raw desktop observation lines. Answers can remain in the asking person's chat, which owners and administrators can read.
Team and organization knowledge features can incorporate permitted work notes and wiki material. Individual heads-ups may be delivered to team leads. Ordinary peers do not receive general access to another person's records.
An organization's owners can receive an exit export of retained organization records as stored, together with the associated vault and a manifest identifying exclusions, subject to the availability and withdrawal rules in Section 9. This may include stored records not presented through ordinary web views. An organization that downloads an export controls its subsequent copies.
Administrative changes are recorded in an audit log. Individual reads of content are not comprehensively recorded, and the service does not send a user a notice each time someone reads a record.
6 Our personnel and service providers
We use AWS for hosting, storage, identity, email delivery, key protection, and operational monitoring. The service's primary cloud hosting configuration specifies AWS US East (Northern Virginia), us-east-1. The hosting region does not establish where authorized people access data or where customer-selected AI services process information.
Authorized Ambient personnel can administer the infrastructure that holds customer data, including records, recordings, and the vault. Although the operator dashboard does not provide ordinary member-content browsing, its Exit page provides download links to the whole exit export described in Section 9, subject to that export's stated exclusions. Providing those links does not create an application audit-log entry. The operator can also issue invitations assigning the Owner role; those invitations are recorded in the audit log. Infrastructure access can reach customer information. We do not represent that every such access requires a customer's prior approval or is visible in a customer read log.
We disclose information to service providers as needed for the functions described in our agreement and applicable provider register. We may also disclose information when legally required, to address security or legal claims, or as part of a business transaction subject to appropriate safeguards and applicable notice requirements.
Business email is routed through Zoho's mail-hosting service. AWS SES sends website confirmation emails, and inquiry confirmations include the outbound Cal.com scheduling link described in Section 12. We use business correspondence and scheduling information to respond to inquiries, arrange demonstrations, and carry out the business purposes described in Section 4. External scheduling services have their own applicable terms and privacy information.
7 Customer selected AI services and other destinations
The organization selects its AI service and supplies its own credentials. Supported configurations include Anthropic, OpenAI, OpenRouter, a customer-designated public HTTPS OpenAI-compatible service, and an appropriate Ollama endpoint. Credentials are protected centrally using AWS key-management encryption and are not placed on enrolled desktops. AI requests originate from our cloud service.
The application uses one configured provider for an organization and does not select another provider as an application-level fallback when that provider fails. An intermediary such as OpenRouter can have its own routing and provider defaults. The customer must consider those settings and the downstream services involved.
Selected structured values in some stored lines are replaced with placeholders and their original values held in a vault. This is reversible protection, not anonymization. It does not remove names, topics, or the general substance of work. It does not cover all content: typed chat can be sent to a model as entered; recordings, recaps, names, work records, and intelligence are not covered by the same redaction process. The screen and clipboard content gate controls whether that material is sent through the configured model-processing paths; it does not mean that the material is absent from the organization's stored record.
Model-provider handling, including provider-side retention, training, abuse monitoring, and human review, depends on the customer's provider agreement, chosen service, and settings. Our own copies and responsibilities remain governed by our agreement and this notice. We do not promise zero retention or universal redaction at every provider.
Registry discovery is disabled in the cloud organization release described here. Application controls prevent cloud research and work functions from searching the Smithery registry, and organization administrators cannot enable registry search through settings. This is an application restriction, not a network-isolation guarantee; configured AI-provider and other permitted external communications continue. The release does not recall or delete earlier registry queries or results, or establish what was previously sent; retained records remain subject to the retention and deletion rules below. Organization-configured external MCP servers are not part of the service described in this notice unless separately added to the contracted offering with updated disclosures.
Desktop setup may contact Hugging Face to download a transcription model and Microsoft to install WebView2 when needed. Those downloads are distinct from the ordinary routing of customer observation content. External sites reached through a scheduling or other link have their own notices.
8 Retention
The organization receives its agreed retention schedule through its contract or onboarding record. Ambient's operator applies the schedule; organization administrators do not currently view or change the retention periods in the product. The following are the standard defaults, subject to the customer's agreed schedule and applicable legal obligations.
| Record category | Default period or condition |
|---|---|
| Completed meeting recordings | 90 days from upload completion |
| Raw speech and activity lines and live transcript | 180 days from the relevant line time |
| Feed records including YOU/REPLY chat copies | 365 days from the record time |
| Persisted agent snapshot | Record-built public context and private fields in the same snapshot; no independent age-based sweep. Rebuilding and private message-array limits are explained below. |
| Raw unanswered chat | Removed on answer commit; no age expiry while pending |
| Timeline, journal and meeting records including eligible journal chat copies | 730 days from the relevant record, journal day, or meeting end |
| Stored model turn records including turn-context copies | 90 days from the beginning of the turn |
| Work run outputs and completed or declined jobs | 365 days from run completion or the job's relevant last change |
| Wikis, comments, heads-ups and intelligence | No general age-based expiry; the wiki-claim quarantine rule below is an exception. Otherwise retained until deleted individually, with the relevant team, or with the organization |
| Vault values | Removed when no retained row of the value's owner references them |
| Revoked-device registry rows and accounts without membership | Normally 30 days from revocation or the end of the last membership, respectively |
| Expired invitation records | 90 days after expiry |
| Administrative audit and deletion ledger records | Six years; audit identity information can remain while relevant audit rows remain |
| Exit administration records | No scheduled expiry, including after organization deletion; includes the organization's name, operator notes, requester, and device and member identifiers |
| Device departure and revocation guard records | Retained without a scheduled expiry to identify devices and prevent renewed access after removal; these include identifiers and public keys rather than captured content |
| Operational logs and traces | Configured 30-day retention for the shared API/worker application log group and aws/spans, with asynchronous deletion after expiration rather than an exact erasure deadline. This is not a period for every AWS log store; other service-generated records can have separate retention settings. |
| Content-free operational metrics | Up to 15 months under the monitoring service's retention |
| Sales leads | 730 days from the latest contact under the configured lead-retention rule, with earlier operator deletion; subject to applicable rights and lawful retention exceptions |
Wiki claims without a valid citation move to a Quarantine section. A quarantined claim is dropped on the next page write after 30 days in quarantine; without a subsequent page write, it can remain beyond that period.
The 365-day default applies to YOU/REPLY chat rows in the feed, not every representation of a conversation. Raw unanswered messages are removed when an answer is committed and have no age-based expiration while pending. Where the wiki is enabled and an interaction is nonprivate, chat can also be copied to the journal and follows the 730-day records schedule. Applicable rights and lawful preservation obligations can require different treatment of particular copies.
The persisted agent snapshot holds record-built public context, including day and meeting blocks, recent lines and wiki text with associated identifiers or cursors. Private messages and discussions occupy logically separate fields in that same snapshot, not a separate table. Public context is not a 40-message replay. Each private discussion's message array retains up to 40 message entries, dropping oldest user/assistant pairs; this limit does not cap the discussion list or its request, source-person or turn-reference arrays.
Public-context reset does not clear private fields. The snapshot contents have no independent age-based sweep; the 90-day turn-record and 365-day feed-chat periods do not automatically apply to snapshot copies. Scoped meeting, day/time-range and document deletion marks public context for rebuilding on a later qualifying request; it does not independently rewrite or erase the persisted agent snapshot. Without that request, affected context can remain. Rebuilding from remaining records does not erase independently retained derivatives or turn-context copies. A permission check can block private-discussion reuse without erasing the stored discussion.
Whole removed-member record deletion removes that person's agent snapshot, including private fields; ordinary membership removal does not. Organization deletion removes tenant-owned state. Deleting one person's records does not establish deletion of copies in another member's private discussion; those copies need assessment as part of the applicable deletion request.
For active organizations, an age-based process is scheduled daily. Eligible records are normally removed at the next successful pass, approximately within a day; this is not an exact deletion deadline. Failed runs or an organization's failed pass can delay cleanup. Ordinary age-based sweeping pauses while an organization is paused or leaving. Routine cleanup then follows its exit process or the next applicable active sweep; separately authorized scoped deletion remains available while the organization is held, as described below.
The production configuration specifies seven days of automated database backup retention. Manual and final snapshots are reviewed in a monthly operator process and are selected for deletion once at least 30 days old, unless a documented reason requires continued retention. This is not a guarantee that every snapshot is removed on day 30. Relevant retained snapshots and their planned deletion dates are identified in the exit completion process.
9 Exit export and deletion
After the last day of the contractual period, if no renewal is in place, we initiate managed organization offboarding through our operator under the contract's period boundaries. Expiration does not itself start an automatic exit timer. Other managed exits are initiated by the operator following an authorized owner's written request or under a documented contract-end arrangement accepted by the customer. We confirm the request date and time, the operator's actual freeze start and scheduled end, export selection and authorized recipient. The reversible freeze lasts 30 days from the operator action that starts it, rather than from receipt of the customer's request. This timing does not restart or extend applicable legal request deadlines. The service refuses new incoming customer data and ordinary access; desktops stop capture when they receive the state, and owners retain supported export access only while the export remains available. Normal jobs are held; work already in flight can finish and is drained at commit. Cancellation restores the earlier active or paused status.
The reversible freeze gives the customer an opportunity to re-establish service before deletion. Re-establishment requires an authorized written request, an agreed continuation or renewal arrangement, and the existing cancellation, status and safety checks; it is not automatic. Early termination, including termination of an affected service portion, follows the applicable contract-end arrangement identified in the Order or otherwise agreed in writing. Termination of a portion does not itself authorize deletion of the whole organization or termination of unrelated service portions.
The exit export is optional and requested by default. A requested original export must be written before managed deletion can proceed. Until first written, it has no export expiry and remains required even after the 30-day freeze. An exit without a requested export has no export prerequisite. The package contains the included organization tables as stored, including the organization's sign-in domains, the associated vault, available completed recordings, administrator-visible audit records, and a manifest. It excludes model turn histories, internal bookkeeping and queues, account sign-in identity bindings and sign-in-provider configuration, provider secrets, and device departure guard tables. Files are available through temporary download links. The first export's availability period ends 30 days after it is written and can extend beyond the freeze. Any eligible replacement keeps that first export's original expiry; it does not receive a new 30-day window. Object cleanup follows the processes and limits below. An applicable deletion duty may require earlier removal or a changed handoff package; ordinary export availability is not a legal exception.
After the freeze deadline, managed deletion can proceed only while the exit remains in progress and the applicable export prerequisite is met. If an already-written export is withdrawn, a requested replacement that is not yet written holds managed deletion only until the first export's original expiry. After that expiry, no replacement is generated and export access is not reopened; deletion may proceed subject to the other exit requirements and safeguards. An original requested export that has never been written still holds deletion. An expedited final commit follows the written-request and approval process below and the same export rules. Deletion drains work, rechecks state, deletes tenant-owned live organization records and queues object and identity cleanup; completion is then checked. Legal-request handling and outstanding cleanup obligations continue. On the customer's written request, including a request made in the Order, our operator issues a completion statement after verification describing performed work and remaining copies. This is not a universal promise of deletion within 30 days or certification of every copy. The exit administration record remains after organization deletion with no scheduled expiry, as described in Section 8; it is separate from the six-year audit log and deletion ledger. Audit and deletion evidence, eligible accounts, exports, backups, metadata guards and endpoint copies have the separate treatment described here.
Expedited final commit requires a clear written request from the customer's authorized exit requester acknowledging and approving the expedition, and our acknowledgment and approval of that request. All existing authority, requested-export, status, locking, drain/recheck and legal safeguards remain required. A casual request, generic termination or decision to decline an export does not authorize bypassing them. Separately required legal action remains due under applicable law and is not delayed by this expedition procedure.
Separate direct controls bypass the managed freeze and export prerequisite. An authorized owner may directly delete an organization after typing its exact name, including while it is paused. Account deletion removes the person's own space and organizations with no other active members; a sole owner cannot leave an organization through account deletion while other active members remain. Other organizations may retain contributions and audit information. These routes warn that deletion cannot be undone, but do not provide the managed cancellation or export opportunity.
An organization can request deletion of a meeting, recording, a person's day or range, a document, or a removed member's record while active or while paused or leaving. For a held organization, the operator must confirm the exact held status before proceeding. Authorization, a dry run and review, deletion records, and restore safeguards remain applicable. The organization stays held: scoped deletion does not resume ingestion or ordinary processing, release the device hold, extend the freeze, or alter the conditions for managed commit. Some derived text needs additional review. Technical scopes and account or exit processes do not reduce applicable privacy rights or extend legal deadlines.
If scoped deletion affects an existing exit export or handoff package, the service withdraws that package by clearing the information used to offer downloads. It then replaces the stored file contents with placeholders containing none of the withdrawn content, one object at a time. A clean replacement may be provided only if an export was requested and the first export's original availability window is still open. An exit without a requested export remains without one. No new export or handoff package is created, and export access is not reopened, after the original export deadline. Withdrawal or replacement does not extend the exit freeze.
Withdrawal and cleanup are not simultaneous or instantaneous. Already-issued download links and ongoing downloads are not immediately revoked; an issued link may continue to serve content until that object is overwritten or the link expires. Such links last no more than 15 minutes and are capped by the original export deadline. Copies already downloaded outside the service remain outside Ambient's control. Storage cleanup is asynchronous; interrupted or failed cleanup, or writing still in progress, can leave content until later storage cleanup. Withdrawal and cleanup instructions remain tracked for retry and after a database restore. The service does not guarantee removal of every object at the exact expiry time.
10 Copies on desktops
Desktop delivery queues normally remove acknowledged lines and otherwise are bounded by age and capacity. Successfully uploaded meeting media is removed locally after acknowledgment. Unsent recordings may remain for the user's decision.
When a desktop receives a valid departure response, it stops capture under either policy. Under clear, it attempts to remove the applicable organization outboxes, acknowledged or discarded meeting folders, pushed settings, and enrollment information, and retries failed steps. A person may choose Leave anyway before cleanup is complete, so some files can remain. Unsent recordings remain until the person elects to keep or delete them; personal mute and view preferences and application logs can remain. Under keep, no local data is cleared. A device that never receives the response cannot be confirmed as stopped or cleared. File removal is not represented as forensic secure erasure.
11 Privacy requests and choices
For information controlled by your organization, contact the privacy contact identified in your organization's notice. You can also submit a request to support@ambientassist.ai or contact privacy@ambientassist.ai. We will help identify the responsible organization and assist in accordance with our role, the agreement, and applicable law. Where we are responsible for answering, we acknowledge, respond and take required action within the applicable legal periods, including any lawfully available extension. Where we assist your organization, forwarding or verification does not restart its clock unless the governing law expressly uses a different trigger. We verify identity and authority as appropriate and do not disclose another person's information merely because someone requests it.
The support address is a request-intake channel; it does not replace authentication, routing to the responsible organization, or any additional request method required by law. Ordinary support objectives and support hours do not extend privacy-rights or incident-notification deadlines.
Depending on the law that applies, rights can include access, correction, deletion, obtaining a copy, and certain objections, restrictions, opt-outs, or appeals. Exceptions may apply, including appropriate preservation and legal obligations. If we decline a request we are responsible for answering, we will provide the explanation and appeal information required by applicable law. We will not penalize a person for exercising a legally protected privacy right.
Organization settings and operating-system permissions affect capture, but do not replace any notices or consents required for other participants. Ask your administrator if a setting is held by the organization or if you need an alternative work or meeting arrangement. Ending a membership stops authorized service access; it does not itself erase all records the organization retains.
12 Website technologies and business practices
The marketing website's homepage and pages for leaders, managers and individuals are designed to use first-party styles, scripts and icons, with local static illustrations. In that design, these pages use no cookies, analytics, advertising pixels, third-party frames or embeds, or external font requests. This description covers those marketing pages, rather than every future page or the separate workplace application. The illustrations do not capture a website visitor's screen or audio or activate the enrolled workplace application's capture features.
The website stores aa-theme in the browser's localStorage to remember the visitor's light or dark display preference.
The contact form submits an email address, an optional message, the page pathname, the call-to-action text and an anti-bot field to the site's own leads endpoint, with browser credentials omitted. The stored lead record contains email, message, page/call-to-action information, timestamps and confirmation/contact status. These website inquiry records are separate from customer-directed workplace processing.
AWS SES sends the confirmation email. The scheduling address is https://cal.com/ambientassistai/intro-call. It is an outbound link, not an embedded scheduling widget, and Cal.com is not loaded merely by viewing these marketing pages. Following the link takes the visitor to that external service and its applicable notices and terms.
The lead database does not store IP addresses or user-agent information. Hosting nevertheless handles network traffic, including IP addresses. The website and application configuration does not enable optional access-log delivery for the static website's CloudFront distribution and S3 bucket and for the application load balancer. The regional AWS WAF filters application load-balancer traffic; it is not attached to the static site's CloudFront distribution and does not provide that WAF coverage for ordinary static-page browsing. On the relevant application request paths it uses IP-based rate limiting, metrics and sampled requests, which can include addresses, request paths, headers and timing. No full WAF logging destination is specified in that configuration. These settings do not eliminate network and security processing or separate records that AWS may maintain in providing its services. The 30-day retention setting in Section 8 does not apply to every operational or service record.
Lead records use the 730-day period measured from the latest contact described in Section 8, with earlier operator deletion. This configured period is not an exact calendar-month period and does not authorize retention for unrelated purposes or override applicable rights and lawful retention exceptions.
In our current business practices, we do not sell or exchange prospect or contact information with another business; upload lists of people's email addresses or other identifiers to an advertising platform to match them to accounts, build lookalike audiences or retarget them; or give identifiable lead or customer-contact information to a marketing partner for that partner's own advertising or targeting.
The service is offered to organizations for business use and is not directed to children. Organizations should identify any deployment that could involve children's information or other specially regulated content so the parties can assess that use before enabling it. Authorized user accounts are limited to individuals who are at least 18 years old. Individuals under 18 may not hold or use an authorized user account; no exceptions apply. Parental, guardian, employer or customer permission, lawful employment and emancipation do not create an under-18 exception. This chronological-age account rule does not mean that every meeting participant, nearby person, bystander or other individual described in captured information is an adult, or that minors' information cannot enter the service. Signing authority and contractual capacity, applicable upstream-provider requirements, and required recording or privacy permissions remain separate.
13 Changes and contact
We will update this notice when our practices change and identify the effective date. We will provide additional notice or obtain permission where required. A change to this notice does not itself amend a customer's negotiated contract or authorize a new use of its information.
Privacy: privacy@ambientassist.ai. Security: security@ambientassist.ai. Ordinary support and request intake: support@ambientassist.ai. General inquiries: contact@ambientassist.ai. Postal correspondence: Ambient Assist AI, L.L.C., 201 Rue Beauregard STE 202, Lafayette, LA 70508, US.