MANAGED AI / HUMAN-GUIDED FROM DAY ONE
Your own AI assistant.
A dedicated assistant for the independent professional or business owner who wants everyday work organised, researched and prepared.
Tell us your requirements. We confirm scope, service allowances and a tailored proposal before onboarding. No purchase or immediate deployment is created by an enquiry.
Configure this pathwayWho it is for
Independent professionals and business owners organising everyday work.
When another route may fit
Unbounded execution, a large multi-user operation or complex regulated workflows.
Example work
- Daily briefings and task preparation.
- Research, drafts and document workflows.
- Connections to selected everyday tools, after feasibility checks.
Examples require agreed scope and integration feasibility; these are not guaranteed ready-made connections.
Onboarding and approvals
Tell us what you need.
Choose outcomes, tools and how much authority your agents should have.
We configure your environment.
Confirm scope, connect approved systems and test workflows.
Your agents get to work.
Start with supervised execution and clear approval points.
We keep it running.
Monitor, maintain and improve within the agreed service plan.
Scope agreed before onboarding
- Runtime
- An isolated managed runtime is the proposed starting model. A dedicated VM is not automatically included as standard.
- Workflow scope
- One primary user and a small, agreed set of starter workflows. Integration counts and execution allowances are confirmed before onboarding.
- Integration count
- A defined count, supported capabilities and access scopes are confirmed in the proposal.
- AI usage and execution
- Model usage, schedules, concurrency, spend limits and overage handling are defined before activation.
- Monitoring and support
- Coverage, response targets, recovery expectations and maintenance boundaries are agreed in writing.
Not included by implication
- Unlimited development, tokens or autonomous execution.
- Unsupported integrations or unrestricted access to customer systems.
- Unverified certifications, residency guarantees or an assumed enterprise SLA.
Control and ownership
- You choose the tools and data to connect; we confirm feasibility and permitted access.
- Explicit permissions restrict actions. Sensitive or high-impact work needs approval unless specifically authorised under a bounded policy.
- Activity and usage must be recorded and reviewable within the agreed plan.
- Access must be revocable and workflows stoppable. These controls are validated before enabling execution.
- Customer configuration, secrets and data must be separated between environments; architecture and isolation are checked during onboarding.
Changing your plan
Review Business when you need several agent roles, more users or a dedicated environment.
Upgrade, cancellation, offboarding and data/configuration export are confirmed in your agreement before onboarding.
Questions before choosing
What am I actually buying?
A human-configured managed agent engagement: agreed workflows, approved connections and ongoing operation within a written plan. It is not a licence to a generally available self-service control plane.
Is this different from a chatbot or automation tool?
The engagement adds workflow specification, configuration, integration and management around agreed work. Tools and models are chosen for the task; their separate licences and limitations are reviewed during scoping.
Will I have my own server?
Personal starts with an isolated managed-runtime design, not an automatic dedicated VM. Business defaults to a dedicated VM subject to validation. Enterprise infrastructure is designed after discovery. The final architecture is confirmed before onboarding.
What does setup include?
The agreed discovery, configuration, connection checks and workflow testing needed for your initial scope. Setup effort, revisions and deliverables are specified in writing; additional engineering is separately measured.
Which systems can you connect?
Tell us your project, CRM, document, email, support or development tools. We assess available APIs, authentication and permitted actions. Listed use cases are examples, not a verified out-of-the-box integration catalogue.
What happens when the AI makes a mistake?
We agree review points, stop conditions and recovery actions for each workflow. Start under supervision and investigate recorded failures before increasing authority. Outputs and actions still require checks appropriate to their consequences.
Can agents act without approval?
Only specifically authorised actions inside agreed bounds should run without a new approval. Sensitive or high-impact actions require review unless explicitly authorised under a bounded policy. Unrestricted autonomy is not a package entitlement.
Are AI model costs included?
Each final proposal defines its AI allowance, usage measurement, spending limits and overage mechanism. Unlimited tokens or unlimited execution are not included. These numerical allowances are not yet published as firm entitlements.
Can I upgrade or cancel?
Changes, contract term, cancellation notice, export/deletion and offboarding are agreed before onboarding. No self-service cancellation terms or refund promises are assumed from these indicative prices.
Who owns my data and workflow configurations?
Data access, configuration ownership, portability, subprocessors and model-provider policies are recorded in the agreement. Secrets and customer configurations must be separated from reusable runtime code; no rights are inferred from submitting this form.