Managed Automation
Make your recurring workflow reliable without turning your team into automation engineers.
Memory(One) maps your process, configures the automated workflow, gives the business owner a clear operating view and provides the agreed monitoring, maintenance or incident response.
Start with one recurring workflow, a named business owner, manageable exceptions and a workable fallback. The applicable proposal and agreement determine whether the selected arrangement provides continuous Managed Care, bounded Managed Support or another route.
Teams often know exactly which administrative process consumes time: information is copied between systems, missing details trigger follow-up, documents need sorting, or someone has to check that the same routine completed again.
Managed Automation turns one suitable workflow into a bounded, monitored service without asking the client’s team to maintain the automation platform. Memory(One) maps the process, configures and tests the workflow, gives the owner a clear operating view and runs the covered service while Managed Care is active.
What Managed Automation is
Managed Automation combines a fixed launch with ongoing care:
- Free workflow discovery confirms fit and identifies the smallest useful solution.
- Managed Automation Launch configures and puts one bounded workflow into operation.
- Managed Care provides the covered hosting, monitoring, maintenance, support and agreed small changes needed to keep operating it.
This is a managed service on the Memory(One) platform. It is distinct from a Custom Build, Foundation or Operations project delivered into a client-owned or client-approved environment.
When is Managed Automation the right route?
Managed Automation fits one bounded recurring workflow with a named owner, manageable exceptions, human review where needed and a workable manual fallback. Broader data, reporting, interface or multi-workflow needs belong on another route.
A standard first workflow normally has:
- one clearly bounded process;
- one named business owner and a backup contact;
- two or three supported integrations at most;
- low or moderate operational risk;
- manageable exceptions;
- human approval before consequential actions where appropriate;
- a workable manual fallback;
- no bespoke infrastructure, broad procurement or custom SLA requirement.
It is probably not the right route when the requirement includes several workflows, a shared data and reporting foundation, a bespoke interface, independent deployment, substantial code ownership, high-impact autonomous decisions, broad data cleanup or custom security and procurement work.
Don't worry—we'll help you figure out what automations could help your company.
Suitable workflow patterns
Subject to discovery and access, a first workflow can include:
- enquiry intake and routing;
- moving submitted information into a work system;
- missing-information reminders;
- document intake and classification;
- recurring internal summaries;
- service triage and preparation for human review;
- document routing;
- status notifications;
- a bounded AI-assisted extraction or drafting step with human review.
Suitability depends on the real trigger, systems, data, exceptions, consequences and fallback — not the label attached to the workflow.
Process
- Understand — walk through a recent real example, including manual decisions and exceptions.
- Select — choose the smallest workflow that is useful, supportable and safe enough to operate.
- Configure — connect the agreed systems and set the workflow, owner, review and fallback behaviour.
- Test — cover the happy path and agreed failure paths with representative examples.
- Launch — confirm acceptance, onboard the owner and backup contact, and make the operating view available.
- Operate and improve — monitor the covered outcome, investigate issues and make eligible small changes under the active care plan.
Free workflow discovery
Bring the person who knows the process and walk through a recent real example.
The discovery includes:
- a suitability conversation;
- the trigger, steps, systems, exceptions and manual decisions;
- an initial risk and fallback review;
- a recommendation for the smallest useful solution;
- a fixed initial proposal;
- an estimate of third-party licences and material usage.
It does not include a reusable technical blueprint, connector specification, detailed data model, node-by-node workflow design, production access review or final implementation instructions.
Managed Automation Launch
Launch puts one scoped workflow into operation through supported Memory(One) patterns and the managed platform.
Included:
- focused discovery and a plain-language process map;
- one bounded automation, integration or AI-assisted workflow;
- agreed supported integrations;
- human-review and exception paths;
- happy-path and agreed failure-path testing;
- a documented manual fallback;
- client portal access;
- monitoring and support configuration;
- onboarding for the owner and one backup contact;
- fixed scope, price and timeline.
Launch excludes additional workflows, bespoke interfaces, a data warehouse or broad reporting foundation, historical migration, broad data cleanup, custom SLAs, unlimited revisions, material third-party costs and custom security, procurement or legal work.
Managed Care
The workflow remains hosted and operated only while an active care subscription is in place.
Managed Care Basis
Best for deterministic automations and integrations that should remain operational without the client maintaining them.
Included:
- covered hosting and health monitoring;
- platform and security updates for covered Memory(One)-managed components;
- backup and restoration for covered configuration and platform data;
- client portal access;
- support and diagnostic triage;
- plain-language exception and incident notifications;
- approved safe retries and recovery actions;
- connection-status checks where supported;
- human investigation when automated triage is insufficient;
- up to two hours of eligible small changes each month.
Eligible small changes can include a notification recipient, simple wording, an existing threshold or schedule, a simple field mapping or a minor covered configuration correction. The allowance does not accumulate and does not include new workflows, integrations, interfaces, infrastructure or material process changes.
Managed Care AI
Managed Care AI includes the Basis features plus recurring quality management where an AI component creates a real operating burden. It can include output review, quality checks, covered prompt and operating-instruction updates, covered knowledge-base updates, model-upgrade evaluation, usage and cost monitoring, exception review and adjustment of agreed quality or human-review thresholds.
A model call alone does not require Managed Care AI. Plan selection follows the actual quality-review, tuning and operating responsibility.
Managed Support
999 DKK/month, billed monthly in advance.
Managed Support is a low-cost incident-response option for a covered workflow. It includes one customer-raised incident per billing month and up to 60 minutes of investigation or correction for that incident. Memory(One) provides an initial response within two working days during the service window: Monday–Friday, 09:00–16:00 Copenhagen time, excluding Danish public holidays. This is an initial-response commitment, not a resolution SLA.
The allowance does not accumulate. Additional time or work requires approval and a fixed quote. Managed Support does not include continuous monitoring, proactive maintenance, routine health checks, a monthly small-change allowance, new functionality, new integrations, redesign, data cleanup or custom SLA coverage. Third-party failures, customer-owned system changes and new requirements are outside the included incident boundary.
During the warranty period defined in the written scope, Memory(One) fixes issues caused by its own implementation or configuration at no additional charge.
Warranty corrections do not consume the monthly incident allowance. Cancellation is the current month plus 30 days; service continues through the paid notice period and then stops. Migration Assistance is separately scoped and priced.
Baseline and success measures
Before launch, the written scope records the current workflow and the evidence needed to judge the managed outcome. Depending on the workflow, the baseline and measures may include:
- elapsed time and hands-on effort from trigger to useful outcome;
- volume completed, waiting or routed to an exception;
- exception rate, recovery time and repeated failure patterns;
- manual follow-up, rework or missed handoffs;
- human-review turnaround and rejection or override patterns;
- last successful business outcome, not only the last technical run;
- named owner readiness, backup coverage and manual-fallback use;
- material usage or operating cost where relevant.
Each selected measure states its source, period, owner and review point. Targets are engagement-specific and do not create a universal savings, uptime, ROI, payback or performance promise.
Monitoring and support
The operating view is organised around the business capability, not only technical executions:
| Visible state | What the owner needs to know |
|---|---|
| Last successful outcome | When the most recent useful result completed and what evidence confirms it. |
| Waiting items | What is waiting for input, approval, an external system or a scheduled retry. |
| Exceptions | What needs attention, why it stopped and who owns the next action. |
| Recovery | Whether an approved retry or intervention restored the expected outcome. |
| Accountable owner | Which client owner or backup receives the decision, exception or escalation. |
Monitoring can ask:
- Did the expected record or document arrive?
- Was the last useful outcome recent enough?
- Are items accumulating in an exception queue?
- Are credentials or connections failing?
- Did a schema or external response change?
- Are retries hiding a persistent problem?
- Did the workflow report success without producing the expected result?
When an issue appears, Memory(One) detects and classifies it, attempts only approved safe recovery, confirms whether the business outcome recovered and escalates unresolved issues to a human with context.
No uninterrupted-operation promise is made. A manual fallback remains part of the operating design.
What happens when a managed workflow fails?
The operating view makes waiting items and exceptions visible. Memory(One) classifies the issue, attempts only approved safe recovery, checks whether the expected business outcome recovered and escalates unresolved cases with context. The named owner retains a documented manual fallback. The applicable Care or Managed Support agreement defines the monitoring and response boundary; there is no universal uptime or resolution promise.
What we need from the client
- One recent real example of the workflow from trigger to outcome.
- A named owner and backup contact who can answer process questions and receive exceptions.
- Access to the agreed systems through an approved method.
- The business rules, exceptions and manual decisions the workflow must respect.
- Representative test cases, including failure and edge cases.
- A workable manual fallback for interruptions or unresolved exceptions.
- Timely acceptance, approval and notification decisions.
- Payment for agreed third-party licences and material usage.
Illustrative example
A submitted enquiry arrives by form. The workflow checks whether the required information is present, creates or updates the record in the agreed work system and sends a plain-language missing-information request when needed. Uncertain matches or consequential routing decisions pause for human review. The owner can see whether items completed, are waiting for input or need attention.
The actual systems, fields, messages, review rules and service boundary are set during discovery and written scope.
Cancellation, access and exit
- The service runs monthly after launch.
- Cancellation is the current month plus 30 days.
- Service remains active through the paid notice period.
- Portal access, hosted workflows, monitoring, support and updates depend on the active subscription.
- When the subscription ends, hosted workflows stop and portal access ends.
- Outputs already written into client-owned systems remain there.
- Client-owned third-party accounts remain the client’s.
The Launch fee covers discovery, configuration, testing and launch on the Memory(One)-managed platform. It does not purchase a standalone copy of the platform or a perpetual right to operate the hosted workflow independently.
Ownership
| Component | Ownership or right |
|---|---|
| Client data and documents | Client |
| Outputs written into client systems | Client |
| Client business rules and process knowledge | Client |
| Client-owned accounts | Client |
| Memory(One) platform | Memory(One) |
| Shared connectors and monitoring components | Memory(One) |
| Reusable workflow templates | Memory(One) |
| Client-specific managed-platform configuration | Licensed while subscribed |
This model is different from Foundation, Operations and Custom Build projects, which generally use client-owned or client-approved environments and project-specific ownership terms.
Exit archive and Migration Assistance
The included exit archive can contain a plain-language process description, client-supplied business rules, connected-system list, client-data export where applicable, recent operating and incident history, fallback description and credential-revocation confirmation. It is not an independently executable automation.
Migration is fixed-quoted after the destination is reviewed, and pricing varies depending on project scope. It may recreate the workflow in a client-owned account, transfer client-owned credentials, configure appropriate monitoring, test the independent workflow and provide limited documentation and onboarding.
Third-party costs
The monthly service price includes Memory(One)'s infrastructure, hosting and internal platform.
Genuine consumption-based external costs are charged separately. These include AI or token usage, automation-platform execution costs, SMS or messaging costs, advertising spend and other customer-specific external consumption. Memory(One) estimates these costs before launch and agrees how they will be billed.
All prices are excl. VAT and third-party costs.
Related route: Custom & Advisory Work
Managed Automation not the right choice for you? Are Foundation and Operations also not the right fit? We also provide Custom & Advisory Work. Reach out and we'll see whether we can help.
Let's automate the work you're tired of doing!
Bring one real example, the systems involved, the person who owns it and the manual fallback.