Skip to content
scalablelab
Back to homepage

OWNERSHIP & WORKING BOUNDARIES

Your system.
Your control.

Built for you. Owned by you. Managed by Scalable Lab.

Your own AI team should make your business more capable, not more dependent. Here is how we scope the build, agree access and prepare you to take it forward.

01 / THE BUILDA working system for your business.

02 / THE ACCOUNTSYour providers. Your billing. Your access.

03 / THE CHOICEOngoing care, only if you want it.

AGREE IT. BUILD IT. CHECK IT.

The controls below are agreed, configured and tested during each build. They are acceptance criteria for the implementation, not a claim that every control is already live in every system.

01 / OWNERSHIP

The work stays with you.

You own the agreed system delivered under your proposal: custom workflows, configurations, company knowledge, generated assets and code within that scope. Your business controls its provider accounts, billing and access.

The build is the core offer; monthly maintenance is optional. You do not need to keep Scalable Lab as your manager to retain the agreed system.

A clear boundary, before we build

The proposal identifies what we deliver and any third-party software, licenses or provider terms it depends on. Ownership of the delivered work does not transfer ownership of Claude, OpenAI or their underlying models.

02 / ACCESS

Relevant context.
Not unrestricted access.

We agree which people and agents need which sources and actions. Each agent should see only approved, relevant sources, not broad shared read access to sensitive records.

  • Limit the permissions. Agree role-based access, minimize credentials and document who can grant or revoke access.
  • Keep the source visible. Agree source attribution so your team can trace the knowledge used and review the work.
  • Keep consequential actions with people. Configure human approvals for publishing, budgets, financial changes, account access and irreversible actions.

These boundaries are tested against the agreed workflows before acceptance. They are implementation checks, not a universal security guarantee.

03 / ACCOUNTS & MODELS

Your provider relationship.

You use your own Claude or OpenAI accounts, as appropriate to the build, rather than a bundled Scalable Lab account. You control provider billing and access. We document the required services and model choices with you.

A chat subscription is not automatically API credit. Before implementation, we verify each required provider's API eligibility, billing setup and usage limits for the actual integration.

No unapproved detours

Alternate models must be pre-approved and stay within the same data boundary. If a model is unavailable or a limit is reached, use only a pre-approved alternate that meets the same permissions and privacy rules. If no permitted route remains, stop and escalate. Never quietly change providers or move private-only work to a cloud model.

04 / PRIVATE DEPLOYMENT

Privacy is a whole-system decision.

A local language model alone does not make an entire workflow private. Before making a privacy claim, we need a tested data map of where information goes, where it is stored and which services can receive it.

What the data map must cover

Inference, embeddings, parsers, memory, logs, backups, speech, analytics, integrations and fallbacks.

We agree the deployment boundary, permitted connections, retention and fallback behavior for the scoped workflow. Private tasks must never silently fall back to cloud. If the approved private route is unavailable, stop and escalate.

Local, private-cloud or mixed deployment is a design choice to assess and test, not a blanket promise of privacy.

05 / COSTS & LIMITS

Know what keeps it running.

Your proposal separates the build from optional care. Continuing use may also require hosting, provider usage, software licenses and integrations, paid through the agreed accounts.

  • Name the owner. Agree the billing owner for each service and the person who receives alerts or escalation.
  • Set the boundaries. Configure alerts and spend caps where supported. Document provider limits and what happens when a limit is reached.
  • Check readiness. Have the payment method checked by an authorized person. Do not assume a subscription funds the API.

We do not promise automatic top-up or automatic extra spending. Additional spending needs the agreed authorization; without a permitted route, stop and escalate.

06 / HANDOVER & CONTINUITY

Ready for the next person.
Even if that isn't us.

Handover is part of the build, not a reward for staying on a monthly plan. We agree the acceptance checks and document the working system so your team or a replacement operator can take over.

Your handover pack

  • Documented working system and current configuration
  • Dependencies and source/config export for the agreed scope
  • Account/access register with owners and permissions
  • Backup/restore instructions and agreed backup checks
  • Training and acceptance checks for your team

Transfer credentials through a secure, approved process, never in public documents. Agree how Scalable Lab's access will be removed or changed during the transition.

Ending care does not erase the work. Continued operation still depends on hosting and provider costs, account permissions and API access, and the services the system uses. Changes after handover are not automatically included in the original build.

We plan for transition continuity. That is not an indefinite reliability guarantee or a promise that external services will never change.

MAKE THE BOUNDARIES PART OF THE BRIEF.

Start with the work.
Agree the foundations.

Bring a workflow, the systems you use and the information that needs to stay private. We can scope what to build, what it needs and what you will receive.

Talk through your build See how we deliver