Noema for organizations
Deploy Noema Teams with your company’s knowledge and access controls. Work with NoemaAI on a custom model pilot for specialized workflows and private deployment on supported devices.
Ways to work with NoemaAI
Each offering builds on the same private Noema app your employees use on iPhone, iPad, and Mac. Start with one and add the others when a workflow calls for them.
Prepare a protected workspace, manage console access and distribute encrypted knowledge. Your company approves device access and signs its policies through a connector you control.
Protected employee workflows arrive with Noema 4.2.
Use encrypted offline datasets or retrieve authorized passages directly from your company connector. Original documents, private indexes and decryption keys stay in your environment and on approved devices.
Connecting documents is retrieval. It does not train the model.
A guided engagement to adapt a model with data you approve, for your terminology, output formats, and workflows, then evaluate it on your task and your hardware.
Scoped pilots with defined deliverables and acceptance criteria.
Example workflows
Illustrations of work we can scope with you, not published customer results. Each one starts with a named task and a measure agreed before any change is made.
Classify incoming internal requests and produce a structured handover for the right team on a company Mac.
Answer new employees’ questions from the handbook and current procedures, with the sources shown.
Pull the fields your systems need from forms, reports, or contracts into your schema.
Draft replies or reports in approved terminology and structure, grounded in current material.
Retrieval or adaptation
Retrieval supplies current facts but can miss the right source. Fine-tuning can improve behavior but can also introduce regressions. We choose from an observed need and the data available, then evaluate the complete workflow.
| Need | Initial approach | Evidence before expanding |
|---|---|---|
| Answer using changing internal information | Retrieval with permissions and source references | Answer correctness, citation quality, and access tests |
| Follow a consistent style, schema, or procedure | Configured prompts; fine-tuning on approved examples if needed | Improvement on unseen cases against the configured baseline |
| Improve a specialized task or domain vocabulary | Task fine-tuning or domain adaptation | Representative data, a defined task metric, and retained general behavior |
| Reduce latency or memory for a specific workflow | Model selection, quantization, and possibly distillation | Quality and performance measured on the target device |
| Use current knowledge with specialized behavior | Retrieval combined with model adaptation | Evaluation of each component and the combined result |
How a pilot runs
Pilots start from an appropriate pretrained model, including a Noema model when it fits. We record the base model, the method, and the resulting artifact at every step.
Identify the task, the current process, the available data, and your hardware.
A specific problem suitable for evaluationAgree deliverables, acceptance criteria, processing location, timeline, and budget.
An agreed pilot scopeEvaluate the configured existing model and workflow before changing anything.
A measured reference and its limitationsPrepare the approved data and perform only the customization the baseline justifies.
A reproducible candidate and run recordTest held-out cases and measure size, memory, and latency on your target devices.
A comparison against the agreed criteriaPackage the approved model and deliver it to the permitted pilot audience.
Use by the pilot groupHandle updates, regressions, and compatibility for the deployed model.
A defined maintenance planIf adaptation does not improve the agreed outcome, we say so. You receive the measurements and a recommendation for the configured baseline or another approach.
Deployment and data handling
In a local deployment, inference runs on your supported devices. If you commission model adaptation, we agree which data may be processed, where training takes place, who can access it, and how long it is retained. Client data is not used for unrelated models or other customers without explicit permission.
Normal local use
With Noema 4.2, approved devices can retrieve and generate locally. Direct retrieval contacts your company connector; remote inference requires a company-approved endpoint. Protected content is not processed by Noema servers.
Optional adaptation
Only data you explicitly approve is processed, in the training environment agreed for the engagement. Access, retention, and deletion of files, checkpoints, and logs are set in advance.
Administration
Workspace accounts, billing, console roles, device-delivery registrations, audit records and encrypted artifacts. Noema does not hold the customer signing or decryption keys. Optional model adaptation has a separate agreed data scope.
Permissions apply before training. A model can retain what it was trained on, and role-filtered retrieval cannot remove it. Restricted information stays in a separate model for its permitted audience or in controlled retrieval.
Noema Teams
The production console manages workspace accounts and encrypted delivery. Your customer connector remains the authority for protected device access, sources and inference policy. Employee workflows ship with the upcoming Noema 4.2 app release.
In Noema 4.2, employees import a device profile approved by your company authority. The app verifies its customer signature and enforces the approved datasets, tools, models, endpoints and offline lease.
Encrypt documents with the customer publisher before uploading an artifact to Noema. Only approved recipient devices can decrypt it. Direct retrieval keeps source files and indexes on your company server.
Console roles govern administration and encrypted delivery. Customer-signed policies separately govern source access, models, tools, approved inference endpoints and off-grid operation.
Noema 4.2 checks company policy where tools and inference run. An expired lease or device reboot locks protected access until a fresh policy is obtained, while retaining encrypted content for renewal.
Use verified email and DNS domain ownership for workspace enrollment. Customer connectors can integrate configured OIDC and SCIM, while customer administrators separately approve and revoke content-access devices.
The console records administrative actions such as invites, role changes, and device registrations. Employees’ local prompts and answers are not sent to the console.
For companies, schools, research groups, clinics, law practices, campaigns, and nonprofits that need useful AI without surrendering internal knowledge.
Evidence
We describe performance with a named task, a baseline, the hardware, and the measurement. Customer results appear here only from real engagements, with permission.
Noema answering from an on-device dataset, with the evidence it used.
Published model work
Trained, evaluated, and released with its methodology and measured limits.
Read the model cardPilot results
Measured results from customer pilots will appear here once an engagement is complete and the customer approves publication.
Commercial approach
Pricing depends on the task, data, devices, and support you need. We propose it once the scope is clear.
Yes, as a guided pilot service. Each pilot has a defined scope, deliverables, and acceptance criteria, and the work is hands-on with our team. There is no self-service training dashboard.
No. Company knowledge uses retrieval: the model reads relevant passages when it answers, and its weights do not change. Training happens only in an adaptation project you commission, on data you approve.
We agree in writing which data may be processed, where training takes place, who can access it, and how long source files, checkpoints, and logs are kept before deletion. Client data is not used for unrelated models or other customers without explicit permission.
A model can retain information from its training data, so access is defined before training. A broadly distributed model is trained only on data approved for that audience; restricted information stays in a separate model or in controlled retrieval.
Usage rights, export, and retention for each artifact are agreed in the engagement’s scope. The base model’s license still applies, and we record the base model and method for every artifact we deliver.
Pilots start at the organization and task level: your vocabulary, templates, procedures, and approved examples. Role-specific prompts, knowledge access, and model assignments add further personalization. Per-employee training and continuous learning from conversations are not offered.
We report it. If the adapted model does not meet the agreed criteria, you receive the measurements and a recommendation for the configured baseline or another approach. An evaluation can end in a clear decision without a new model.
Pricing depends on the task, data, devices, and support you need. We propose it once the scope is clear, with the deliverables and responsibilities written down.
Offline datasets are decrypted on approved devices. Direct retrieval sends a query to your company connector, and optional remote inference sends context only to an endpoint your company approves. The protected path keeps plaintext content and decryption keys outside Noema infrastructure.
Yes, within the customer-signed offline lease and with the required data already installed. Off-grid policy blocks online retrieval and remote inference. Lease expiry or reboot requires policy renewal before protected work can continue.
Revoke customer content access in your company connector. Online requests then fail and a device erases governed data when it receives revocation; an offline device remains bounded by its signed lease. Revoking a Noema delivery credential separately stops ciphertext downloads.
The customer connector includes configured OIDC JWT exchange and SCIM users, groups and deactivation. It requires your identity-provider setup and validation, plus independently approved device keys. SAML is not included. Native protected access arrives in Noema 4.2.
Start a conversation
Email us a short description and we’ll follow up with a practical next step. Please do not send confidential datasets; data transfer follows an agreed scope and handling process.
Helpful to include
