When one decision at a time isn’t the job.

The self-serve product answers a question. Enterprise builds a machine around your pipeline — with a human consultant who owns the output, an approval committee that governs it, and connectors that ground every analysis in your own data.

Can AI analyze an entire data room or email archive rather than one document?

Yes. Enterprise Deep Discovery points the same decision process at whole corpora — hundreds of thousands of pages or emails, read in full through a cull, retrieve and deep-reason funnel. Every engagement includes a named human consultant, an executive approval layer, connectors into your own systems, and a client-controlled model roster. Deployment runs from managed to entirely inside your own perimeter. From $20,000/year.

Six things the self-serve product doesn't have

Enterprise is not a bigger plan. It is a different shape of engagement — the platform, plus the people and the plumbing that make it fit one organization's actual work.

A named consultant who owns the output

Not a support queue. A person who scopes the engagement, turns your decision criteria into the system's logic, reads every deliverable before it leaves, and sends it back when it isn't right.

In the engagement described below, the consultant rejected several drafts of a portfolio deliverable before it shipped — including one that compared each asset's price against a number derived from that same price. The AI would have published it. The human wouldn't.

An Executive Committee that has to agree

A standing panel of independent frontier models on separate clouds reviews the decisions that matter: methodology changes, what ships, what gets published, and what gets held back.

• Debates in rounds — positions, challenges, refinements
• Records consensus and preserved dissent
• Can and does overrule the proposal in front of it

Connectors into your own systems

Your edge is your data. Enterprise wraps your sources — market and listing platforms, loan servicing and core banking, e-discovery and document repositories, ERP and logistics, claims and policy systems, GIS and permits, or simply your own warehouse — so the debate reasons over your live pipeline and cites it, not just the open web.

• Offered on every engagement; many already built and tested
• Private to your account; no other client can trigger it
• Credentials encrypted, scoped, never in logs
• Read-only by default; nothing written back unless you ask
• New source = a new isolated connector, not a rebuild

Document sets that break normal tools

Enterprise reads what a team can't. Not sampled, not keyword-filtered — read, cross-referenced, and reasoned over, with contradictions and gaps reported rather than smoothed away.

• A 10,000-page acquisition data room, end to end
• 45,320 unique emails read blind in a single investigation
• Custom preparation: your formats, your criteria, your questions

Where the system runs is your choice

Managed by us, inside your cloud account with your own keys, installed on-premises behind your own firewall, or sealed and air-gapped with no outbound connectivity at all.

For organizations with compliance obligations, on-premises matters more than any certificate we could hold — your data never leaves the envelope you already had audited. See the four tiers below.

Which AI models touch your data is also your choice

The model roster is configuration, not doctrine. Exclude any vendor, cloud or jurisdiction — competitive, legal, contractual or policy reasons, no justification required. Restrict to open-weight models you host yourself. Or pin the approved list your risk committee already blessed.

The method survives the constraint. Panels get composed from whatever roster you permit and the debate works the same, because the process is the product, not any particular model.

Scoping to standing capability

Every Enterprise engagement follows the same five steps. The first four are usually measured in days, not quarters.

Scoping call

What decisions do you actually need supported, what data feeds them, and what does a defensible answer have to contain for your investment committee, your board, or your regulator? This is a conversation, not a questionnaire.

Criteria as code

Your buy-box, thresholds, disqualifiers and house rules get written into the system as explicit, inspectable logic — not a prompt. You can read it, argue with it, and change it. In the engagement below, the client's own rules ("institutional sellers only," "must be transactable remotely") became lines in a thesis file that every candidate is scored against.

Connector build and isolated testing

We build against your APIs in an isolated environment, run real cases end to end, and review the output with you before anything touches production. Security gates are cleared before a single credential of yours is stored.

Run, review, deliver

Analysis runs at whatever scale the question needs — a single decision, a portfolio, a continuous scan. Your consultant reviews the deliverable; the Executive Committee reviews anything consequential; then it goes to you in the format you'll actually use.

It keeps running

Most engagements don't end at delivery. They become standing capability — a scan that runs every few hours, a weekly ranked list, a pipeline that gets smarter as your criteria sharpen.

Four tiers, from managed to sealed

Data isolation is both a security requirement and a design choice. Pick the tier your obligations demand — the analysis is the same; the boundary moves.

Dedicated managed

We run it. Your account gets isolated storage and its own connectors; your data is never co-mingled with another client's, retained only long enough to deliver, and never used to train anything.

Your cloud, your keys

The system runs inside your own cloud account against your own model credentials. We operate it; you own the infrastructure, the logs and the bill. The common choice for most enterprises.

On-premises, behind your firewall

Installed entirely inside your own network. Nothing crosses your perimeter, and your existing certified environment is the control boundary — the honest answer to most compliance questions we're asked.

Sealed / sovereign

Air-gap-capable, running open-weight models you control, grounded only on your own corpus. Built for classified, defense and sovereign contexts where no egress is permitted at all.

We hold no security certifications

3Dogs Nexus is not SOC 2, ISO 27001, FedRAMP, CMMC or HIPAA certified, and we are not going to imply otherwise on a marketing page. If a certificate is a hard procurement gate for you, say so on the first call and we will tell you honestly whether we can clear it. Here is how the requirement actually gets met today:

1. Inherited from the infrastructure. The platform runs on major cloud providers whose certifications, audit reports and compliance artifacts are available to you directly through their own trust programs.
2. Solved by architecture. Deploy in your perimeter (Tier 3) or your cloud account (Tier 2) and your already-audited environment is the control boundary. This satisfies more real procurement reviews than a vendor certificate does.
3. Carried by a partner. For cleared or accredited delivery, 3Dogs is the software layer beneath an integrator who holds the credentials the contract requires.

Read the full Security & Trust page →

From $20,000 per year

Every engagement is scoped individually. Price moves with the data sources connected, the decision types supported, the deployment tier, and the volume of analysis you run. Custom work — a bespoke connector, a portfolio underwrite, a document investigation — is quoted as part of that scope, not billed by surprise.

Solo, Professional and Corporate plans exist and priced publicly. Enterprise is the tier where a person, a committee and your own data enter the picture.

What this looks like when it's real

Three published examples of Enterprise-scale work. The first is a live client engagement, anonymized; the other two are documented capability tests you can check line by line.

Start with a conversation

Twenty minutes is usually enough to tell whether Enterprise fits what you're trying to do. If it doesn't, we'll say so.

Questions this case answers

Where does my data go, and do you train on it?

We never train on your data and operate zero retention. Deployment tiers run from a dedicated managed store, to your own cloud with your keys, to fully on-premise behind your firewall, to a sealed air-gapped install.

What certifications do you hold?

None, and we say so plainly. We are not SOC 2, ISO 27001, HIPAA or FedRAMP certified and are not currently pursuing them. Where compliance matters, the answer is architectural: deploy inside your own already-certified perimeter so your controls govern the data.

Can I choose or exclude specific AI models?

Yes. The model roster is client-controlled — exclude any vendor, cloud or jurisdiction, restrict to open-weight models only, or pin an approved list.

How is this different from e-discovery software?

E-discovery handles collection, processing and search. This is the judgment layer above it: reading the responsive set, forming a view, arguing it across independent models, and returning a documented recommendation.