Written for COOs, heads of operations and shared services, and the transformation leads who build the automation program.
What's in this playbook
This 34-page playbook is a method for automating work that holds as the organization grows. Most automation programs, it argues, do not fail because the tools are weak. They fail because the work was never truly seen, the process was never redesigned, and nobody owned the result after go-live. The playbook is the method we use to avoid that, written down so you can apply it with or without us.
It is written for the people who decide what gets automated and who live with the result: chief operating officers, heads of operations, shared services and transformation, and the process owners who will be asked to change how their teams work. It assumes familiarity with how work moves through your organization, not with any particular platform.
The method has six stages — Discover, Qualify, Prioritize, Redesign, Build, and Operate — and a companion register in which the work is held. Sections 1 and 2 explain why programs stall and set out conventions for how numbers and examples are used. Every figure in the playbook is either a method, an illustrative example labeled as such, or a question you can answer from your own file. It will not invent a sector rate, a typical payback, or a ranking of tools.
Who it is for, and how to use it
Read the opening once, then work from the register. The stages are a sequence, not a buffet: skipping Discover and Qualify to “just automate the obvious one” is how a program starts with the wrong work. The companion file, asymilink-workflow-automation-register.xlsx, is the working document — a pipeline of candidates, scores, owners, baselines and operating reviews.
A first case is real work, on a named process, with a process owner who will sign a baseline and live with the result. The playbook’s view is that a reliable workflow is often the right first mechanism, and that a goal-directed agent is a later move, when a simpler path cannot do the job. The agentic-workflows paper in this library is the companion for that question. The financial method for any case you then fund is the method in The Complete Guide to AI Automation ROI, and in the ROI calculator on this site.
Production work is handed over documented, so your team can run it. An operator, a log, a sampling plan and a switch-off are part of Operate, not a later hardening. The playbook will not promise that a workflow heals itself, and it will not treat a pilot that only the builder can run as a finished case.
Six stages, in order
The playbook moves through Discover, Qualify, Prioritize, Redesign, Build and Operate. Each stage has an output that the next stage consumes. Skipping a stage is how a convenient program quietly fails.
Discover
See the work as it actually runs, including exceptions, unofficial paths and the days that go wrong. A process that exists only in a slide is not discovered.
Qualify
Put every candidate through a standard rubric, with knockouts. Two people, with the same evidence, should reach a similar answer.
Prioritize
Sequence the work in waves, given the people you actually have. A first wave that contains every idea is not a plan.
Redesign
Draw the future-state — including exceptions, ownership and hand-offs — before anyone builds. Automating a poor process at speed is still a poor process.
Build
Build behind gates, against the redesigned process, on a signed baseline. A demo is not an outcome.
Operate
An operator who is not only the builder, a log, a sample, a value review, and a way to stop or change the workflow without a project.
Six questions for every automation proposal
Before a candidate is funded, the playbook asks a short set of questions that any sponsor can put in writing. A missing answer is a reason to pause, not a reason to start.
The work
What is the process, as it actually runs, and who owns it? If that cannot be said in a sentence, Discover is not finished.
The result
What named result would be better, for whom, and how would you know? A platform is not a result.
The mechanism
Why this mechanism, and not a simpler one — a rule, a redesigned form, a reliable workflow, a person? “We have the platform” is not a reason.
The baseline
Volume, time, cost, quality and a realization factor, agreed with the process owner, dated before design is funded.
The operator
Who will run it next month, and how do you stop it today, without a project?
The cost of leaving
What is in the three-year view, including the cost of leaving? If that line is “to be confirmed,” the case is not yet a case.
Get the guide and workbook
Enter your name, work email and company. The files appear on this page immediately. A backup download link is also emailed when email delivery is available.