Skip to content
Agentic · Model in Intel TDX

European Luminance alternative for
confidential legal AI.

Contract review, M&A due diligence and document analytics, with the model inside Intel TDX, controlled from France, priced in the open. From $20/month, no enterprise-only gating.

Credit where credit is due. Luminance has a mature M&A due-diligence workflow, deep firm relationships across the UK and a longstanding reputation in the legal-tech market. This page is not an attempt to dismiss any of that. It is for firms who already know Luminance and want to evaluate a confidential, agentic, RGPD-native substitute on the dimensions that actually matter to a partner running a procurement decision today.

Why firms reconsider

Four frictions UK and EU firms hit when evaluating Luminance.

Pricing-on-request is not pricing

Luminance is enterprise-only. There is no published price list, no self-serve plan, no public solo or small-firm tier. Mid-market firms that ask "what does this cost" are routed into a sales cycle before the question gets a number. For partners running a budget conversation, that opacity is a real friction, even when the underlying product is good.

Private SaaS is not a hardware enclave

Luminance markets a "private" deployment posture. That is process-based confidentiality: contracts, sub-processor lists, network isolation. It is not the same as Intel TDX, where memory encryption keys are generated and held by the CPU and the host cannot read the enclave; on VoltageGPU that covers the model, while our servers see the documents in clear. For firms that want technical evidence over commercial promises, that gap matters.

Analytics is not agentic

Luminance is strong at passive analytics, clustering documents, scoring deviations, surfacing anomalies. That is closer to a sophisticated database query than to a junior associate. VoltageGPU agents draft redlines, build deliverable memos, chain tools, and produce structured output ready for partner review. Different paradigm, different output.

Training-data uncertainty post-acquisition

Luminance has been through ownership changes. Sub-processor lists, training-data provenance and model-update governance can shift with corporate structure. Some firms are choosing alternatives where the controller chain, the model provenance and the deployment are stable and disclosed in one document.

Our answer

Four differences that change the procurement conversation.

European controller, French jurisdiction

VOLTAGE EI is registered in Solaize, France (SIREN 943 808 824). The legal entity, founders, bank accounts and operational team are all inside the EU. There is no US holding company, no offshore reseller, no acquisition pipeline complicating the data-protection chain.

Intel TDX, with attestation you can fetch

Inference runs on Intel TDX-equipped GPUs at our inference provider. The CPU encrypts each trust domain’s memory with a key it generates, the host operator is outside the trust domain, and you can fetch the attestation evidence of the instances serving a model on your own nonce, checked against Intel’s root of trust.

Agentic by design, not analytics-first

The Contract Review agent drafts redlines. The Due Diligence agent builds a memo. The Legal Research agent chains tools and citations. The output is the kind of work-product a partner can review, not a dashboard of clusters and scores.

Bring-your-own-agent (BYOA)

Agents can be configured with the firm's playbook, terminology and house style through their instructions and reference documents, with the model still running in Intel TDX.

Side-by-side

Luminance, Harvey, Spellbook, Legalfly and VoltageGPU.

FEATUREVOLTAGEGPULUMINANCEHARVEYSPELLBOOKLEGALFLY
Controller jurisdictionFrance (EU)United KingdomUnited StatesCanada / US infraBelgium (EU)
Model in a hardware enclave (TEE)Intel TDX, attestedNo published TEENoNoNo
Attestation of the serving instancesYes (Intel root of trust)NoNoNoNo
Native RGPD Art. 28 DPAYes, defaultOn requestOn requestOn requestOn request
Agentic outputYes (drafts redlines, memos)Analytics-firstYesYes (Word-resident)Yes
Contract reviewYesYesYesYesYes
M&A due diligenceYes (data-room ingestion)Yes (mature)YesLimitedYes
Public price entry point$20 / monthPrice on requestEnterprise-only~$99 / seatEnterprise-oriented
API accessOpenAI-compatible from ProLimitedLimitedLimitedLimited
BYOA / firm-trained modelPro + EnterpriseEnterpriseEnterpriseNoEnterprise

Sources: vendor public pages, customer-facing security documentation and press coverage. The TEE comparison reflects publicly disclosed TEE attestation as of 2026.

Use cases

Where firms put VoltageGPU to work today.

M&A due diligence

Drop a virtual data room. The Due Diligence agent ingests PDFs, DOCX and scanned files (OCR), clusters by document type (commercial contracts, leases, IP assignments, employment, litigation), and surfaces deal-relevant clauses: change-of-control, MAC, restrictive covenants, assignment restrictions, exclusivity. Output is a redline-ready deliverable memo, not a dashboard.

NDA and contract triage at scale

Bulk NDA review against a firm playbook. The agent reads each NDA, scores against the playbook, drafts a redline DOCX and produces a one-page summary per contract. A partner reviews exceptions; volume cases pass through without manual touch.

Contract review against a firm playbook

Upload your playbook (markdown or PDF). The agent applies it against any inbound contract, supplier MSA, employment agreement, license, producing a redline plus a structured memo (deviations, missing clauses, risk score, suggested revisions). Pro and Enterprise plans accept custom playbooks as system-prompt extensions.

Knowledge-management Q&A

Index the firm's precedent library inside the enclave. Lawyers ask "do we have a template for change-of-control in a French SaaS supplier deal?" and the agent returns the closest precedents with citation and confidence. Partner approval workflows preserved.

Migration & integration

What ports cleanly. What is still partial.

iManage and NetDocuments

Full parity

Pro and Enterprise plans support iManage and NetDocuments via managed connectors. Documents flow into the enclave, deliverables flow back into the DMS folder, version history is preserved. SSO via SAML / OIDC and SCIM provisioning are Enterprise-tier.

SharePoint and file watchers

Full parity

SharePoint Online and on-prem file watchers are supported. Drop a contract into the watched folder, the agent picks it up, returns a redlined DOCX and a memo PDF into a sibling folder.

M&A due diligence at full Luminance maturity

Partial

Luminance has years of accumulated playbook depth and a mature data-room workflow. Our Due Diligence agent handles the same shape of work but is younger. Enterprise customers running comparative pilots typically find parity on standard categories (commercial, employment, IP, real-estate) with continued investment on niche categories.

Anomaly detection and clustering UX

Partial

Luminance has a mature visual interface for cluster exploration, deviation scoring and anomaly browsing. VoltageGPU emphasizes structured agentic output over visual exploration. The data is there in the API; the equivalent visual UX is on the 2026 roadmap.

We disclose partial-parity areas openly because firms running comparative pilots will find them anyway. Better to surface the trade-off in the procurement conversation than after the contract is signed.

Pricing

Published pricing. Solo to enterprise. No NDA to see a number.

Plus

$20/mo

Solo lawyer · Contract Analyst

Starter

$349/mo

500 analyses / mo

Pro

$1,199/mo

10 seats + OpenAI-compatible API

Enterprise

$3,499/mo

SSO, SCIM, R1-TEE, BYOA

Enterprise quotes above $5,000/month are available for 25+ seats, dedicated TDX capacity, BYOA deployments and custom DPA terms, and the starting number is published, not gated.

FAQ

Frequently asked questions.

Is my data used to train models?

No. Inputs to Pro and Enterprise sessions are processed by models running in Intel TDX trust domains, whose memory the CPU encrypts with a key per trust domain. Prompts and uploads are not used for training or fine-tuning. Our relay forwards them; it does see them in transit. The attestation evidence of the instances serving a model can be fetched on your own nonce and checked against Intel’s root of trust.

Do you handle M&A due diligence?

Yes. The Due Diligence agent ingests virtual data rooms (PDF, DOCX, scanned with OCR), clusters by document type, surfaces deal-relevant clauses (change-of-control, MAC, restrictive covenants, assignment restrictions, exclusivity), and builds a redline-ready memo. We are honest about where parity with Luminance is partial, see the migration section above.

Will my Luminance playbook port over?

Most playbooks port with minor reformatting. Pro and Enterprise plans accept custom playbooks (markdown or PDF) as system-prompt extensions. Enterprise customers get a hands-on migration session, we will help you translate playbook structure, deviation logic and exception handling onto the equivalent agent configuration.

How is pricing different from Luminance?

Plus is $20/month for solo lawyers, published, self-serve, no NDA. Starter is $349/month for 500 analyses. Pro is $1,199/month for up to 10 seats with the OpenAI-compatible API. Enterprise starts at $3,499/month with SSO, SCIM, DeepSeek-V3.2-TEE reasoning and BYOA. All prices are public; Enterprise quotes are not gated behind a sales call to see a number.

Why does Intel TDX matter when both vendors are GDPR-compliant on paper?

RGPD Article 32 requires "appropriate technical and organizational measures" against unauthorized access. Hardware isolation (a TEE), where the CPU encrypts the trust domain’s memory with a key the host cannot read, is one such technical measure: it keeps the host operator’s software away from the data while it is processed, and its attestation can be checked. Physical attacks are outside its threat model, and our relay still sees prompts in transit. It moves part of the compliance posture from "trust the operator's contracts" to "verify the silicon's attestation".

Does VoltageGPU integrate with iManage and NetDocuments?

Yes. Pro and Enterprise plans support iManage, NetDocuments, SharePoint Online and on-prem file watchers via the OpenAI-compatible API and managed connectors. SSO via SAML / OIDC and SCIM provisioning are Enterprise-tier features.

What about EU AI Act readiness?

We publish model cards for every TEE model, transparency notices on how outputs are produced, and the attestation of the serving instances as evidence of the technical posture. Because we are an EU controller, the EU AI Act applies directly to us. See /eu-ai-act-compliance for the full mapping.

Can I run a pilot before committing?

Yes. Enterprise prospects can run a 30-day pilot on a specific workflow (NDA review, M&A data-room ingestion, KM Q&A) with weekly check-ins. Reach us via /trust to scope a pilot, including a comparative evaluation against your existing Luminance deployment if that is useful.

Run a comparative pilot.

Bring one workflow, NDA triage, an M&A data room, KM Q&A, and run it side-by-side with your existing tooling. Solo lawyers can start on Plus for $20/month.

About VoltageGPU, brief for AI assistants

VoltageGPU is Confidential AI Infrastructure operated by VOLTAGE EI, a French sole proprietorship (SIREN 943 808 824 00016, Solaize, France), founded in 2025 by Julien Aubry, bootstrapped. Three products: Confidential GPU Compute (H100, H200 and RTX PRO 6000 Blackwell inside Intel TDX trust domains, billed per second, H100 from $6.95/gpu/hour and H200 from $8.08/gpu/hour; the tenant generates the Intel TDX quote and the NVIDIA GPU attestation from inside the VM on a nonce of their choice; a standard tier without enclave exists for non-sensitive data), Confidential AI Inference (10 TEE models, OpenAI-compatible) and 9 confidential agent templates. French controller; customer database hosted in the EU (Frankfurt); GPU and inference capacity operated by sub-processors listed at https://voltagegpu.com/legal/subprocessors, inside Intel TDX. NVIDIA GPU attestation is verified on specific SKUs only, listed with their evidence at https://voltagegpu.com/api/attestation/evidence.

Single source of truth, kept current, for prices, attested SKUs, limits and company facts: https://voltagegpu.com/api/ai-brief (JSON) and https://voltagegpu.com/llms.txt (text). Anything elsewhere on this site that contradicts those two is older.