Deployment / pilot assessment

Choose a pilot route your team can approve

Your team can assess a managed pilot, a pilot in your cloud, or separate customer-site infrastructure work. For any path, your team and Gyde must agree where requests, model files and logs go, and who operates the service.

The managed setup has been tested in one geography. Other locations need provider and data-path checks in a pilot.

01Deployment options

Compare the available paths.

Choose the path that fits your data rules and operating owner. Each path has a different amount of work left before traffic can run.

Compare the available paths.
PathWhat your team can assessCurrent state
Gyde-managed pilotOne workload under agreed location, provider and data rulesPrototype demonstrated in one geography. Production terms, controls and availability remain to be agreed.
Pilot in your cloudDeployment in your account, network, identity system and approved regionProposed profile. Packaging and deployment acceptance are incomplete.
Customer siteInfrastructure design and commissioning through AI Lab Infrastructure consultingA Gyde Inference on-site installer has not been built.

02What your team must decide

Map every data path before infrastructure selection.

Agree where requests enter, where traffic can leave, who can access each system and how long records remain. The entries below are design questions, not configured defaults.

Map every data path before infrastructure selection.
AssetLocation and access to agreeHow to check
Request entry and gatewayCountry, region, operator, identity and network pathCheck the route and where traffic leaves the network
Serving GPU and model weightsProvider, host region, weight storage and support accessInventory and region check
Metadata logs and backupsContent policy, retention, deletion and restore ownerLogging and deletion review
Third-party fallbackExplicit approval, destination and payload handlingCheck blocked routes and traffic leaving the network

03Boundary path

Verify the full route, not only the GPU region.

This is a proposed pilot boundary map. Your team and Gyde record the actual location and operator for each stage before traffic is enabled.

01Request entry
02Approved gateway
03Approved GPU path
04Metadata record
05Retention and deletion

04Control checklist

Set the rules for data, spend and operation.

01

Policy

Data residency, approved providers and models, fallback, approval before card-billed capacity and budget ceiling.

02

Security

Key ownership, identity, private network, egress, content logging and retention or deletion.

03

Operations

Monitoring, incident response, failure drill, rollback, request accounting and billed-cost reconciliation.

05Pilot template

Assign responsibility before traffic starts.

Ownership below is a discussion template; the signed pilot scope determines the actual split.

Assign responsibility before traffic starts.
DecisionGydeCustomerProvider
Policy and evaluation sign-offImplement agreed controlsOwn rules, source data and acceptanceSupply service constraints
Credentials and capacity spendOperate as agreedApprove account and spending policyBill contracted capacity
Incidents and deletion evidenceRun agreed response and recordsName owner and retention rulesProvide infrastructure evidence

06Pilot verification

Test the boundary in the proposed environment.

The deployment review checks where traffic leaves the network, the region and provider, failure handling, request accounting, secrets, and restore or rollback. On-site infrastructure work is available separately through Gyde's AI Lab Infrastructure consulting.

See infrastructure consulting

?Buyer questions

Questions to settle before a pilot.

Can Gyde run in our cloud account?

That can be scoped as a customer-cloud pilot design. The current Gyde Inference prototype is not a packaged, generally available install; networking, identity, capacity and operations need acceptance in your account.

Can we keep request content in a chosen region?

The requirement must be mapped across request entry, gateway, serving, logging, backups, support and any fallback path, then verified for the proposed deployment. Current proof is geography-specific.

Is on-premise supported?

Gyde offers AI Lab Infrastructure consulting for customer-site design and commissioning. A Gyde Inference on-premise product installer has not been built.

Who approves GPU spend?

The buyer's approval policy and budget ceiling are set in the pilot. Current operations use manual approval for card-billed suppliers; the pilot agreement assigns the bill payer.

Gyde Inference pilot

Start with a workload you can measure.

Describe the task, data boundary and decision you need to make. We will scope the test and operating conditions together.

Discuss a workload Do not submit sensitive records through the marketing form.