Version 2026-10-05.1. Effective when published and, for contract terms, accepted. Current beta operation; real-money billing is not available.
Reading-copy editorial notice (16 October 2026): this website reading copy incorporates a factual company-status correction only. THREADLINE LDA was incorporated in Portugal on 9 October 2026 (NIPC / matrícula 519681169). The title and version shown above identify the historical accepted document; this is not a new accepted version, agreement or novation, and the original accepted documents and acceptance evidence remain archived unchanged.
1 How AI is used
Threadline uses OpenAI to turn selected professional information, public-web/company research, vendor settings and user instructions into summaries, signals, Account Intelligence and draft messages or emails. Inputs can include non-public company information and custom prompts you supply. Output is AI-assisted even when fluent or confident.
2 Review every output
Output may contain invented facts, mistaken attribution, bias, stale information or unsuitable wording. Check material names, dates, financial claims, source links and relevance. A cited URL does not prove the model’s interpretation is correct. Do not treat output as legal, financial, employment or other professional advice.
You choose the final content, recipient, channel and timing. Threadline does not automatically send outreach. Review and edit before copying, sharing or sending. Similar output can be generated for others; no guarantee of exclusivity, copyright protection or non-infringement is made.
3 Inputs and prohibited uses
Submit only information you may lawfully process and disclose to the service and its providers. Do not include passwords, secrets, irrelevant private details, sensitive personal information, criminal allegations or children’s data. Do not submit another party’s confidential information without authority for the specific processing.
Do not use Threadline for harassment, spam, deception, impersonation, discrimination or unlawful targeting of sensitive traits. Do not use it for solely automated or high-impact decisions about employment, credit, insurance, housing, education, health, law enforcement or migration.
Public professional information remains personal data. The responsible controller must establish a lawful basis, provide required Article 14 notices, honour objections and choose a lawful retention and outreach process. A generated email is not permission to send unsolicited marketing.
4 Provider processing and retention
OpenAI receives relevant inputs and produces outputs. OpenAI states that API data is not used for model training by default. Threadline does not use customer content to train a general-purpose model. Standard abuse-monitoring records may be retained for up to 30 days, subject to provider exceptions and law; store:false does not remove all retention. Threadline does not promise Zero Data Retention or EU-only AI processing.
The Privacy Notice describes recipients, transfers and Threadline’s separate storage of saved outputs, usage records and account information. Deleting a browser cache does not delete those server or provider copies.
5 Transparency and correction
Threadline identifies its research and drafting features as AI-assisted. Do not misleadingly present an inference as verified fact or conceal an AI disclosure where the law or context requires it. Specific EU AI Act transparency duties depend on the feature, content and provider/deployer role; not every ordinary sales email has the same labelling rule.
Report incorrect, harmful or sensitive output to [email protected] with the relevant record or request reference and minimal necessary detail. A correction may need to cover saved outputs and shared briefs as well as the source field. Recipients of copied/exported material may need a separate correction.
6 Responsibilities and rights
You remain responsible for your own lawful inputs, review and communications. Threadline remains responsible for its own design, processing, security, disclosures and mandatory obligations. The main Terms govern permitted use and liability. Nothing in this notice excludes statutory consumer remedies or data-subject rights.
Threadline is currently operated jointly by Diogo Baptista and Pedro Castro. Correspondence address: Rua Aristides Sousa Mendes n8 6o andar, 1600-413, Lisboa, Portugal. Contact: [email protected]. THREADLINE LDA was incorporated on 9 October 2026 (NIPC / matrícula 519681169); the effective transfer of the provider role to it has not yet been confirmed, so it is not the present contracting party.
Company case studies and use-case links — 26 October 2026
Supplement 2026-10-26.1
This notice describes the case-study library introduced with Threadline 1.7.146. It takes effect for this functionality when the corresponding app and backend are released. It does not change earlier accepted terms or record a new agreement.
When you generate your company profile, Threadline uses the company name, product or service category and website to find public customer stories and use-case pages on that website and through web search. You can also submit public page links. This work continues in the background and does not require uploading a document. The search and extraction use OpenAI; the saved library is held in Threadline’s Supabase backend. No additional provider or Chrome permission is introduced by this feature.
Threadline reads public pages and stores their source links, titles, customer and industry names, stated teams and responsibilities, a short description of the problem and use case, reported outcomes, a short supporting excerpt and source/check dates where available. Pages are processed to extract these facts; full source documents are not retained as library attachments. Public source material can contain professional names or quotations. Source facts and AI summaries may be incomplete or inaccurate. The source link remains available for review.
Your company library is associated with your account. In a managed Teams workspace, members use the workspace owner’s company library; the owner manages its links. The library is not a public directory. When you request a new message or regenerate one, Threadline may supply a limited set of previously read facts to the existing writing model. The model decides whether they are relevant. It does not search case-study pages during Reveal, and a missing or unavailable library does not prevent writing. Saving case studies or generating a draft does not send outreach.
You can remove links in the Library tab during setup or on the web app’s Library page. Removal excludes the related evidence from future library inputs and prevents a discovery job already in progress from adding that removed link back. It does not rewrite earlier saved messages or retract text you copied or sent. Changing the company identity replaces its active library. Account deletion removes the account’s library; providers’ logs and backups follow their separate retention schedules. Facts checked more than 90 days ago are excluded from writing, which is a freshness limit rather than automatic deletion. Regenerating a profile can refresh a library older than 30 days, and the Find links again control requests a refresh subject to service limits.
Existing rights, security controls, provider disclosures and contact routes continue to apply. Contact [email protected] for data-rights requests or concerns about a source.
Spoken call playbooks and personal call settings — supplement 2026-10-28.1
From extension 1.7.151, the cold-call playbook has Open, Ask then listen, Brush-off and Close. Rep dialogue is separate from hypothetical prospect answers and coaching. The existing AI writing request generates the message and its playbook together. Regenerating a message also updates its playbook. An explicit playbook refresh uses the full current LinkedIn draft, supplied profile and signal, seller evidence, and available account intelligence. It preserves that draft. Existing bounded provider/error recovery still applies. These are generated suggestions, not verified prospect statements or guaranteed outcomes.
My opener is optional personal text saved to the signed-in Threadline account in Supabase and displayed verbatim. It is not generated, included in the playbook AI request, or included in a shared brief. The rep can replace it or save an empty value. It is available when the same account opens a brief on another device.
The optional sent declaration records that the rep says they sent a specific LinkedIn message, with a chosen date. Copying, generating or editing a draft does not mark it sent. The saved record contains account and prospect identifiers, a SHA-256 message fingerprint, sent date, revision and update time; this new table does not retain another copy of the message body. It is account-scoped and draft-specific. Clearing the declaration removes the date while retaining the fingerprint and revision to prevent stale writes. The declaration does not verify delivery and does not send, read or monitor LinkedIn messages.
Saved call settings and declaration records persist until replaced or account deletion; they are included in the existing account export and removed by the account purge/deletion flow. Previously generated drafts, shared copies, provider records and backups retain their existing lifecycle. Sent status can affect the opening and email brush-off only after the rep explicitly refreshes the playbook. Refresh is available on Starter and above under the existing regeneration entitlement.
This update adds no Chrome permission, provider, automatic outreach, live call recording or microphone access. Existing privacy and AI disclosures continue to apply. Accepted contract text and acceptance history remain unchanged. Contact [email protected] for privacy assistance.
Save this call playbook notice
Simplified call playbook — supplement 2026-10-28.2
From extension 1.7.153, the cold-call playbook displays only Open, Ask then listen, Brush-off and Close. My opener, the sent-message checkbox/date, save controls and separate playbook refresh controls are removed. The updated extension does not read or write those personal call settings. Opening the playbook or editing a draft does not trigger an extra AI request. Normal message generation and regeneration continue to generate the message and playbook together.
New playbook generation does not use saved sent declarations or claim that a message was already sent. Existing saved wording is retained until normal regeneration. Earlier opener and sent-declaration records are not deleted by this interface update; they retain the existing account-scoped protection, export and account deletion/purge lifecycle described in supplement 2026-10-28.1. Older installed versions may still expose their previous controls until updated.
This supplement supersedes the earlier description of available call settings in the updated extension. It introduces no new provider, data category, purpose, retention period, Chrome permission, outreach automation, pricing or contract acceptance. AI output remains suggested wording for the rep to review.
Save this simplified playbook notice
Manual Library use cases — Supplement 2026-10-29.1
From Threadline 1.7.156, the Library also accepts a title and plain text for a case study or use case that is not available online. A public link, uploaded document and team tags are not required. Threadline stores the supplied text in its Supabase backend. Saving or editing this text does not ask an AI model to extract or rewrite it. Ready means saved and available for generation; it does not mean independently verified. Manual entries are labelled Added by you.
The existing writing model may receive saved manual entries alongside public Library evidence when you request a new message or regenerate one. Their contents are seller-supplied information, not instructions. A general use case does not establish an actual customer relationship or achieved result. Only add information you are entitled to use, and anonymize anything you do not want included in generated outreach. Library text can contain business-confidential information or professional personal data if you choose to enter it. The Library is not a public directory, but details used in a draft may appear in text you copy, send or deliberately share. Saving an entry does not send outreach.
An account controls its own Library. In a managed Teams workspace, members can use the owner's Library and only the owner can add, edit or remove entries. Manual entries are stored separately from website discovery and are not overwritten by Find links again. Unlike retrieved public facts, they do not expire after the public evidence freshness interval; the owner maintains them until removal, replacement of the active company identity, product-data purge or account deletion. An account's own Library is included in its data export. Provider logs and backups retain their existing schedules.
Edits and removals affect future Library inputs. They do not rewrite saved drafts, cached signal packages or previously shared snapshots, and do not retract copied or sent text. To use changed Library information in an existing draft, request normal regeneration where available. The feature introduces no new AI provider, Chrome permission, outreach automation, price or acceptance event. It supplements the URL-only Library description in notice 2026-10-26.1; earlier accepted contract text and acceptance records remain unchanged. Contact [email protected] for data-rights requests.