TPDOC AI, arm's length automated

Privacy Policy

Last updated 11 November 2026

Overview

This policy explains what personal data TP Doc Gen AI handles, why we handle it, who else can see it, how long we keep it, and what you can ask us to do about it.

Where your organisation has a signed engagement letter or data-processing agreement with us, that document governs and takes precedence over this page.

1.Who we are and what this covers

TP Doc Gen AI is a transfer-pricing documentation, benchmarking and filing platform operated by Steadfast Business Consulting LLP (“SBC”, “we”, “us”) and built by EchorTech. This policy explains what personal data the platform handles, why, and what you can ask us to do about it.

It applies to the TP Doc Gen AI web application, the client portal on tenant subdomains, and the public marketing pages on this site. It does not apply to information you give us offline, or to a third-party site we happen to link to.

2.Our role: controller and processor

The platform handles two different kinds of data, and our responsibility differs between them. Reading this section first makes the rest of the policy easier to place.

  • Engagement data — the entities, financials, related-party transactions, documents and studies a firm loads to prepare a transfer-pricing file. The firm that holds the engagement is the controller of that data. We act as its processor and handle the data only on its documented instructions.
  • Account and website data — the details of the person signing in, support correspondence, enquiries submitted through the contact form, and traffic measurement on the public pages. For these we are the controller.

If you are a taxpayer, a director or an employee whose details were loaded by an advisory firm, that firm decides what is collected and how long it is kept. Send your request to them first; we will support them in answering it.

3.Information we collect

You give us

  • Account details — name, work email address, phone number, job role, the firm you belong to, and your password in hashed form.
  • Engagement content — entity and group details, financial statements, related-party transaction schedules, agreements, prior-year studies and any other document uploaded to support a filing.
  • Correspondence — enquiries, support requests, review notes and messages exchanged inside the platform.

We generate or observe

  • Derived records — extracted fields, functional analyses, comparable sets, profit-level indicators, ranges and generated report sections.
  • Audit events — who changed what, when, and from which perspective. Every state-changing action writes one.
  • Technical data — IP address, browser and device type, pages requested, timestamps and referring page, used to operate the service and keep it secure.

We do not ask for, and the platform is not designed to hold, special-category personal data such as health, biometric or political information. Please do not upload it.

4.How we use information

  • To provide the platform — intake, document extraction, functional analysis, benchmarking, range computation, report and form generation.
  • To authenticate you, keep your session alive and enforce the access rules that apply to your role.
  • To maintain a defensible audit trail of the work, which is part of what makes a filing supportable.
  • To respond to enquiries and provide support.
  • To keep the service secure — detecting abuse, investigating incidents and preventing fraud.
  • To measure traffic on the public marketing pages so we know which pages are useful.
  • To meet our legal, tax, regulatory and professional obligations.

We do not sell personal data, we do not share it for cross-context behavioural advertising, we do not give it to data brokers or advertising networks, and we do not use engagement content to market to anyone.

6.AI processing and zero data retention

Document understanding and narrative drafting use Claude Enterprise. Every model call runs under a zero-data-retention agreement enforced at the provider’s organisation level: prompt and response content is not logged, not cached beyond the request, and never used to train a model.

Every service routes its model calls through a single internal gateway, which records only metadata — the model, the prompt version, an input hash, token counts, latency and status. Prompt text and model output are never written to our logs.

No figure a tax authority could recompute is produced by a model. Profit-level indicators, quartiles, the arm’s-length range and every adjustment are computed in code and tested against manual workpapers. The model reads documents and drafts narrative; it does not generate numbers that reach a filing.

7.Connected AI agents (Claude and other MCP clients)

A firm can let outside AI agents, such as Claude, read and work with its data through our connector at tpdocgenai.com/mcp. It is off until a Firm Admin turns it on in Settings → Connected agents, and each person then connects their own agent by signing in and choosing what it may do.

  • The agent acts as the person who connected it and can reach only what that person’s role can reach. Only a Firm Admin can let it make changes, and no agent can ever delete, submit or approve anything.
  • What an agent reads is sent to that agent’s own AI provider (for Claude, Anthropic) and handled under the agreement between you and that provider. It is outside our zero-data-retention terms with our own AI provider, and we do not control what that provider keeps.
  • We receive only the tool calls the agent makes and the inputs each one needs. We do not receive your conversations with the agent, its memory or your files on its side.
  • The connector keeps no copy of what it returns; each answer is read live from your firm’s data. Calls are recorded in your firm’s audit log, with the connection, the tool and the time.
  • Sign-in tokens last one hour and renew in the background. A connection unused for 30 days lapses, and every connection ends 180 days after it was allowed. We keep a record of each connection (who, which agent, what access, when) for audit.
  • You can disconnect at any time in Settings → Connected agents, and a Firm Admin can switch agent access off for the whole firm. Either stops access within a minute.

Setup, permissions and support for the connector are described at tpdocgenai.com/docs/claude-connector.

8.Automated decision-making

We do not make decisions about you based solely on automated processing that produce legal effects or similarly significantly affect you, and the platform is not designed to.

The platform scores and screens comparable companies, and drafts narrative, but those outputs are proposals for a qualified practitioner to accept, reject or change — every accept or reject carries a recorded human reason, and a named professional signs the filing. That is a design property of the product, not only a policy commitment.

9.Cookies, local storage and analytics

The platform uses browser storage and cookies for the purposes below. Most of it is what keeps you signed in rather than anything that tracks you.

PurposeWhat it doesWhere it lives
Sign-inKeeps you signed in and silently refreshes an expiring sessionBrowser local storage, plus an HttpOnly refresh cookie
Session stateShort-lived flags such as a required password resetBrowser session storage
Interface preferencesRemembers filters, column widths, collapsed panels and unsent draftsBrowser local storage
Cookie choiceRemembers whether you accepted or declined analyticsBrowser local storage
AnalyticsMeasures traffic to the public marketing pages, only if you acceptGoogle Tag Manager and Google Analytics cookies, production only

Analytics runs only with your consent. On your first visit to a public page we ask, and nothing is loaded until you choose “Accept analytics”. If you decline, or simply do not answer, no analytics tag or cookie is set, and we remember a decline so we do not ask again. You can change your choice at any time from “Cookie settings” at the foot of every public page; withdrawing consent stops the tag and removes its cookies.

Analytics is limited to the public marketing surfaces — the home page, the product, company, blog and legal pages. The tag is not loaded inside the signed-in application, so it never sees engagement URLs, tenant names or financial figures. It is also disabled entirely outside production builds.

You can block or delete cookies and clear local storage in your browser settings, and analytics can also be blocked that way or with a browser extension. Blocking the sign-in entries will stop you from staying signed in; blocking analytics has no effect on how the platform works.

10.Sub-processors

We keep the list of parties that can touch data deliberately short. These are the only ones.

ProviderWhat it doesWhere
Amazon Web ServicesHosting, database, document storage, audit trail and email deliveryMumbai, India — our primary region
Anthropic (Claude Enterprise)Document understanding and narrative drafting, under zero data retentionModel API; no content retained after the request
Auth0 by OktaSign-in and multi-factor authentication on staging and productionManaged identity service
Google Tag ManagerTraffic measurement on the public marketing pages, only with your consentNot loaded in the signed-in application

Each is bound by a written agreement covering confidentiality and security. There is no data broker, analytics reseller or advertising partner in the pipeline.

Before we add or replace a sub-processor that handles engagement data, we will give at least 30 days’ notice to the firms whose data it would touch. A firm may object on reasonable data-protection grounds within that period; if we cannot offer a workable alternative, it may terminate the affected part of the service without penalty.

11.Third-party comparable data

Benchmarking draws on licensed company databases, used under licence for our internal operations and never resold. These hold published company financial and industry information, not personal data about you.

The flow is one-way. We query those providers for company reference data; we do not send them your engagement content, your documents or your clients’ identities.

12.Data residency and international transfers

Application data, uploaded documents and derived records are held in our primary cloud region in Mumbai, India. Material classified strictly confidential does not leave that region in any form — not in application data, not in logs, not in a backup.

India is not covered by a European Commission adequacy decision. So where personal data moves from the EEA or the UK to us, the transfer is made under the European Commission’s standard contractual clauses (with the UK addendum where the UK GDPR applies), supported by an assessment of the destination and the technical measures set out in this policy. A copy of the clauses we rely on is available on request.

The platform serves India, the UAE and Europe, so a transfer across borders may also be necessary to support you. Where it is, we put an appropriate safeguard in place before the transfer.

13.Tenant isolation and internal access

Every table carries a tenant identifier and a database-level row-security policy keyed to the tenant on the connection. Isolation is enforced in Postgres, not by a filter in application code that someone could forget to write.

Access is further scoped by role, by perspective and by data classification. A practitioner cannot read through to the tax-authority view of the same case, and every document and knowledge record carries a classification from public through strictly confidential that is checked on each read.

Internally, access is limited to the people who need it to run and support the service, and their actions are subject to the same audit trail as everyone else’s.

14.How long we keep data

CategoryRetention
Engagement, entity and study recordsFor the life of the engagement, then as long as law or your agreement with us requires
Uploaded documentsThe same period as the engagement they belong to
Audit eventsRetained to meet regulatory and evidential requirements for the filing they support
Account and access recordsWhile the account is active, plus the period needed to resolve any dispute
Model prompts and responsesNot retained — zero data retention applies to every call
Enquiries and support correspondenceUntil the matter is closed and any follow-up period ends

There is no silent auto-delete running in the background. Retention beyond the life of an engagement is a conversation with your account team, not a default we impose on you.

Transfer-pricing documentation carries its own statutory retention period — broadly eight years in India and seven in the UAE — which continues to bind the firm after it stops using the platform. On request within 60 days of an engagement ending we provide an export of that data in a machine-readable format, and remove it from primary storage afterwards on instruction.

15.How we protect data

  • Encrypted at rest with customer-managed keys, and in transit with TLS throughout. Storage buckets enforce SSL and block public access.
  • Row-level security on every table, plus access checks combining tenant, role, perspective and data classification.
  • Short-lived sign-in tokens with silent refresh, an eight-hour sliding session, and multi-factor authentication required for administrator and tax-authority roles.
  • Services run as non-root containers inside a private network behind a web application firewall, with no direct public address.
  • Production databases run across multiple availability zones with automated backups retained for at least seven days.
  • Every state change emits an audited event. Logs carry metadata only — never personal data, financial figures or model content.

We are not SOC 2 or ISO 27001 certified today, and you should treat any claim otherwise as wrong. Formal certification is on our roadmap; the individual controls those audits test are already running, and the Trust Centre sets them out in full.

No method of transmission or storage is completely secure, so we cannot guarantee absolute security.

16.Security incidents

If a personal data breach affects data we hold, we will notify the affected firm without undue delay after becoming aware of it, and give them the information they need to meet their own notification duties — what happened, the categories and approximate volume of data involved, the likely consequences, and the steps we are taking.

Where we are the controller, we will notify the competent authority and, where the law requires it, the individuals affected — within 72 hours under the GDPR, and within the periods India’s Digital Personal Data Protection Act and the UAE’s personal data protection law set.

If you believe you have found a vulnerability, report it to info@sbcllp.in rather than disclosing it publicly. We will acknowledge the report, keep you updated, and will not pursue anyone who reports a genuine issue in good faith and does not access or alter other people’s data.

17.Your rights

Subject to the law that applies to you, you can ask us to do the following.

  • Access — get a copy of the personal data we hold about you.
  • Rectification — correct data that is wrong or incomplete.
  • Erasure — delete data, where we are not required to keep it.
  • Restriction — pause processing while a dispute is resolved.
  • Objection — object to processing we base on a legitimate interest.
  • Portability — receive your data in a structured, machine-readable format.
  • Withdraw consent — where we relied on consent, without affecting what came before.
  • Nominate — in India, nominate someone to exercise your rights if you die or become incapacitated.

You can also complain to a supervisory authority: in the EEA, the authority for the country where you live, work, or where the issue arose; in the UK, the Information Commissioner’s Office; in India, the Data Protection Board; in the UAE, the UAE Data Office. We would rather you raised it with us first so we can put it right.

18.Making a request

Write to info@sbcllp.in, marked for the attention of the privacy contact, who handles privacy questions and grievances for SBC. We will verify who you are before acting, and respond within the period the applicable law sets — one month under the GDPR unless the request is complex, in which case we will tell you.

If your request concerns engagement data that an advisory firm loaded, we will refer it to that firm as controller and help them respond. On a firm’s instruction we export a tenant’s data and then remove it from primary storage.

19.Children

This is a professional service for businesses and their advisers. It is not directed at children and we do not knowingly collect data from anyone under 18. If you believe a child has given us personal data, contact us and we will delete it.

20.Changes to this policy

We update this policy when the platform changes. The date at the top always reflects the current version. Where a change materially affects how we handle your data we will tell you directly rather than relying on you noticing this page. Previous versions are available from us on request.

21.Contact us

Privacy questions, requests and complaints go to info@sbcllp.in. For anything specific to a transfer-pricing engagement, tpdocteam@sbcllp.in reaches the team directly.

Steadfast Business Consulting LLP, Suite 5, Level 3, Reliance Cyber Ville, Vittal Rao Nagar, Madhapur, Hitech City, Hyderabad – 500081, Telangana, India. Telephone +91 9014 079 600.

We use Google Analytics to see which pages are useful. It only loads if you accept, and you can change your choice later from Cookie settings. Privacy policy