First-pass tax and transfer pricing analysis on an open-weight model in an Intel TDX enclave. Nothing goes to OpenAI, Anthropic or Google.
Transfer pricing files, provision workpapers, R&D credit studies and IRS audit responses contain pre-filing positions and often return information covered by §7216. This agent is an AI model following a review checklist, not a tax adviser. Documents are parsed by our servers and passed to an open-weight model in an Intel TDX enclave; never used for training. It checks arm’s-length pricing, models BEPS Pillar Two exposure and structures audit-defense analysis for review by tax counsel. Whether a given disclosure is permitted under §7216 remains your call.
Built for: VP Tax, Transfer Pricing Directors, Tax Provision Managers, Big Four international tax partners
Transfer pricing documentation, provision support and audit responses are slow, expensive advisory work, and Pillar Two adds jurisdiction-by-jurisdiction ETR calculations for in-scope groups. Pasting pre-filing positions into a general-purpose AI tool sends them to a third-party model provider, so the first pass is still done by hand.
Validate transfer pricing benchmarking, run Pillar Two GloBE ETR calculations and structure audit defense in hours, with inference in an Intel TDX enclave and nothing stored.
What the Tax & Transfer Pricing Analyst does on every document, with a TEE model that runs inside an Intel TDX hardware enclave.
Validates Local File, Master File and CbCR against OECD 2022 guidelines and IRC §482. Flags methods that would fail Best Method Rule, comparables with material adjustments and intercompany agreements that do not match actual conduct.
For each transaction: validates CUP / TNMM / profit split / resale price / cost plus selection. Checks PLI calculation, interquartile range, year-over-year consistency. Surfaces patterns like "tested party always at bottom quartile" a profit-shifting signal.
Calculates effective tax rate per jurisdiction, flags ETR below 15%, models top-up tax exposure under QDMTT / IIR / UTPR, applies transition rules and safe harbors. For groups with >€750M revenue.
Reviews uncertain tax positions under FIN 48 / IFRIC 23, valuation allowance assessment, rate reconciliation completeness, indefinite reinvestment assertion support, stock comp excess tax benefit treatment.
Walks each claimed activity through the IRC §41 four-part test (permitted purpose, technological uncertainty, process of experimentation, technological in nature). Flags activities the IRS aggressively challenges post-§174 capitalization.
Builds the defensibility scorecard for each position: authority level (statute > regs > court > rev rul > PLR), risk of adjustment, penalty exposure, reasonable-cause defense, adequate-disclosure analysis.
Four steps from upload to export. Your document is parsed by our servers and passed to a TEE model; never used for training.
TP Local File, Master File, CbCR, benchmarking study, intercompany agreement, ASC 740 workpaper, R&D credit study, IDR response. Files go over TLS, are parsed by our servers and passed to a TEE model; never used for training.
The model runs in an Intel TDX enclave, where memory is encrypted by the CPU during inference. The controller is French; compute sub-processors are US-based and listed.
Tax analysis summary, transaction matrix, method validation, BEPS Pillar Two table, provision review, R&D qualification table, audit defensibility scorecard. Positions cite the code section so you can check them.
Markdown output for partner review; copy tables into the provision workpaper binder, or call the OpenAI-compatible API (Pro tier) from your own tooling.
Every output cites the specific article, section or control. The Intel TDX hardware boundary handles the data-protection side; the agent handles the analytical side.
Real prompts from tax / transfer pricing teams.
Review this TP documentation, is our benchmarking defensible under IRC §482?
Assess our BEPS Pillar Two exposure. Which jurisdictions trigger top-up tax?
Analyze this R&D credit study, which activities actually qualify under the 4-part test?
We received an IRS IDR on our transfer pricing. Help structure the defense.
Prompt: Calculate our GloBE effective tax rate for each jurisdiction in this CbCR.
Output: Ireland: ETR 9.2%, below 15% GloBE threshold, top-up tax ~$4.2M (subject to QDMTT priority). Luxembourg: ETR 8.4%, below threshold, top-up tax ~$2.1M. Singapore: ETR 13.7%, below threshold, top-up tax ~$1.1M (Singapore QDMTT in effect from 2025). Total estimated Pillar Two exposure: ~$7.4M before transition safe harbors. Recommend transitional CbCR Safe Harbor analysis on Ireland and Luxembourg given simplified ETR computation may yield ≥15%.
Prompt: Is this 2% intercompany loan arm's length?
Output: NO, at 2% the rate is materially below arm’s length for an unsecured 5-year intercompany loan to a BBB-equivalent affiliate. Treas. Reg. §1.482-2(a)(2)(iii) safe-haven range (100-130% of the applicable AFR) for the period: roughly 4.8-6.2%. IRS §482 adjustment risk: imputed interest of ~$1.8M/year, plus §6662(e) transactional penalty (20% for a substantial valuation misstatement, 40% if gross under §6662(h)). Recommend retroactive credit memo to restate to arm’s-length rate before year-end, with contemporaneous benchmarking documentation.
Every tier runs inside Intel TDX hardware enclaves. Plans stay in sync with /pricing.
1,000 requests/month, 1 seat. For solo tax practitioners, fractional tax directors.
3 seats, 2,000 requests/month, 100 MB uploads, audit log. For mid-market tax teams.
10 seats, 5,000 requests/month, 500 MB uploads, API access for tax provision integration, 12-month audit log retention.
Unlimited seats, fine-tuning on your TP playbook and provision methodology, SSO/SAML, dedicated TDX capacity, signed DPA, EU-hosted customer database.
Honest comparison. Hardware-rooted confidentiality is what most alternatives are missing.
| Alternative | Pros | Cons vs VoltageGPU |
|---|---|---|
| Aibidia |
|
|
| ONESOURCE / Corptax |
|
|
| ChatGPT Enterprise |
|
|
Your documents are parsed by our servers and passed to an open-weight model running in an Intel TDX enclave; never used for training. Nothing goes to OpenAI, Anthropic or Google. This reduces third-party exposure, but it does not settle your §7216 / §6713 analysis: whether a disclosure to us is permitted, and whether privilege holds, is your call.
Big Four TP and provision support runs $300-800/hr fully loaded. The agent is an AI model following a review checklist, not a tax adviser: it does not replace partner sign-off or opinion-level work. It speeds up the first-pass reading and structuring that otherwise takes analyst hours.
Coverage includes GloBE jurisdiction-level ETR, top-up tax computation under QDMTT / IIR / UTPR, transition safe harbors (CbCR Safe Harbor, Simplified Calculations), substance-based income exclusion and BEPS Pillar One Amount A/B context. The agent will flag where simplified calculations differ from full GloBE methodology. Check computations before relying on them.
The agent builds the defensibility framework, authority hierarchy, penalty exposure, reasonable cause and adequate disclosure analysis. Final positions and opinion letters remain attorney/CPA work product. Output is framed as "analysis of technical merits and risks for review by tax counsel."
Not natively. On the Pro tier, the OpenAI-compatible API lets your team script its own uses, such as reviewing intercompany agreements before signature, screening Local Files for benchmarking gaps or pre-checking provision workpapers before quarterly close. Requests pass through our relay, which sees them in clear (TLS ends at our host). To keep plaintext off our servers, run your own inference server in a Confidential VM and terminate TLS inside its trust domain.
The agent screens cross-border arrangements against DAC6 hallmarks (A1-A4, B1-B3, C1-C4, D1-D2, E1-E3) and flags reportable transactions. For DAC7 platform reporting, it can review platform terms against the reporting obligations.
Yes. The agent handles both US (IRC §482) and OECD methodologies on cross-border transactions, including the France-US tax treaty articles, French CFC (Art. 209B), the DGFiP audit posture and 2024-2026 French transfer pricing penalty regime updates.
The Pro tier is $1,199/month ($14,388/year) for 10 seats. It supplements an adviser engagement rather than replacing it; it helps most on first-pass work such as benchmarking refreshes and routine provision support.
Model in an Intel TDX enclave, with an attestation of its serving instances on your nonce. French company (VOLTAGE EI), compute at US sub-processors under SCCs. Cancel anytime.
