Run AI on infrastructure you control.
Start with local AI apps, add AI Server when a team needs shared models or APIs, and use AI Admin Console when IT needs central identity, entitlements, policy and audit. Buy only the layers your deployment requires.
Reviewed 22 August 2026
A product appears here as public only after customers have a verified acquisition path. Source code, packages and reserved store identities do not count as a release.
A public platform, with one desktop component still in validation
Software Tailor has 14 public current-generation products: twelve consumer products, AI Server and AI Admin Console. AI Client v2 is implemented but is not counted as public until its release gates and acquisition path are complete.
Twelve consumer products
AI Suite and eleven specialist apps cover chat, reasoning, translation, writing, documents, images and speech. Every consumer product has a free local path.
Compare the apps →AI Server and Admin Console
AI Server supplies shared private inference and OpenAI-compatible APIs. AI Admin Console manages the organisation around apps and servers.
See AI Server → See the Console →AI Client v2
AI Client v2 is the remaining desktop dependency for the complete Server–Admin–Client workflow. It has no public v2 acquisition path or promised release date yet. The previous-generation AI Client v1 listing remains separate.
Get release updates →One person, one team, or a managed organisation?
One person or one workstation
Use a standalone app or AI Suite with its built-in local model. You do not need AI Server or AI Admin Console.
Find the right app →A team sharing models or APIs
Deploy AI Server on hardware you control. Existing OpenAI-compatible clients and developer tools can use its API without sending inference to Software Tailor.
Explore AI Server →An organisation needing central control
Add AI Admin Console for roles, entitlement allocation, offline keys, policy, server enrollment, install inventory, usage and content-free audit.
Evaluate the Console →Where content goes depends on the path you choose
“Private AI” should describe a data path, not a slogan. These are the boundaries for the currently published products.
| Path | Prompt and document content | What Software Tailor receives |
|---|---|---|
| Local app + local model | Stays on the user's device. | No prompt, document or generated content. Optional telemetry is consent-gated and content-free. |
| App or API client + AI Server | Travels between the approved client and the customer-controlled server. | No inference content. The customer operates the runtime, storage, network and keys. |
| AI Admin Console | The Console does not run inference or process prompts and documents. | Work identity and authorised organisation-management requests: members, entitlements, policy, install inventory, usage totals and audit metadata. |
| Optional third-party cloud model | Sent to the provider the user deliberately selects, under that provider's terms. | Software Tailor does not become the inference provider. Cloud use is optional and separate from the local path. |
For the current control matrix and known limitations, read the Trust Centre. Product-specific collection, retention and account details are in the Privacy Policy; AI Admin Console identity scopes are documented on its product page.
From one machine to a governed server fleet
Start with a functional pilot, benchmark the models on your workload, and add availability or governance only when the operating requirement is clear.
On-premises or private cloud
Run AI Server on Windows, macOS or Linux infrastructure you administer. Keep network, storage and model placement inside your chosen boundary.
Docker or Kubernetes
Use container and Kubernetes paths for repeatable deployment, worker pools, health routing and controlled scale-out.
Offline or air-gapped
Use documented offline licensing and pre-positioned models where the environment cannot make normal internet requests. Validate the complete update and recovery procedure before production.
Four decisions make the evaluation useful
- Choose one real workload. Name the users, data types, acceptable outputs and the decision the AI will support.
- Draw the data boundary. Decide which content may remain on a device, move to an internal server, or reach an optional third-party provider.
- Benchmark before sizing. Test the chosen model for quality, latency, concurrency, memory and GPU capacity using representative inputs.
- Assign operations ownership. Define who manages identity, keys, models, updates, backups, audit review and incident response.
What IT and security teams usually ask first
Do we need AI Admin Console?
Not for individual apps or a simple server evaluation. Add it when IT needs organisation identity, central entitlements, policy, fleet visibility, install inventory, usage or audit.
Can our existing AI tools use AI Server?
Often, yes. AI Server exposes OpenAI-compatible APIs for chat and other supported capabilities. Validate the exact endpoint, model and authentication requirements during the pilot.
Does the software make us compliant?
No software product makes an organisation compliant by itself. The platform provides deployment, identity, policy and audit controls; your organisation remains responsible for its legal assessment, configuration, operating procedures and evidence.
When will AI Client v2 be available?
It is in release validation. We will not publish a date until enrollment, identity, policy, recovery, update and support gates pass. Subscribe for the release notice rather than planning against an unverified date.
How is it priced?
Consumer apps have a permanent free path and optional Pro subscriptions. AI Server has published direct licensing; Enterprise licensing, volume terms and support are quoted for the deployment.
What can we verify before contacting sales?
Start with the Trust Centre, public product catalogue, AI Server operations guide, deployment planner, Admin Console data scope, licensing terms and Privacy Policy. We separate current capabilities, customer responsibilities and roadmap items.
Prove one workload before you scale the estate
Bring the workload, data boundary and operating constraints. We can review the architecture or scope a free one-week pilot without pretending an unreleased component is already available.