Your new workforce won't be all human. Join the
conference to learn the new workforce OS.
request invite
TL;DR: The seven ITIL 4 guiding principles are a mindset, not a checklist: focus on value, start where you are, progress iteratively with feedback, collaborate and promote visibility, think and work holistically, keep it simple and practical, and optimize and automate. Most tools quote them and then do the opposite. This is a concrete position on each, with a real example. The through-line: meet people where they already are, keep humans on the decisions, and let generative AI and agentic AI, running as a governed AI Coworker, do the connective work.
You will see that for each Guiding principle, links to related blogs with video demos on how they align in the Atomicwork AI Workforce ecosystem are shared below.
The ITIL 4 guiding principles are the part of the framework that survives contact with reality. Practices change, org charts change, but focus on value and keep it simple hold up. The catch is that they are easy to nod along to and hard to actually build to. A tool can claim focus on value while making an employee fill in a ten-field form to ask a simple question.
So rather than restate the principles, here is where we plant a flag on each one, and the proof. For a neutral primer on the framework itself, Atomicwork keeps an ITIL 4 guide.
The principle: every activity should trace back to value for the customer, the user, and the business, and value is the outcome, not the activity.
Our position: an employee does not want a ticket, a queue position, or a deflection. They want the thing resolved. A high chatbot deflection rate is not value if the person still ends up messaging IT ten minutes later, which is exactly why employees ignore the chatbot. Value is the answer, delivered. For leaders, value also means seeing the truth of the operation without a six-week reporting project, which is what the CIO dashboard run is about.
The principle: do not rip everything out and rebuild. Assess what you already have and build on it.
Our position, two ways. For employees: they already live in Slack, Teams, and the browser, so that is where the help should be. You do not move people into a portal; you meet them where they are, and Atom works inside the browser and the chat tools they already keep open. For IT: your data already exists, scattered across cloud, identity, MDM, and the CMDB. You do not need a new system of record, you need to read the ones you have. That is the whole idea behind a self-updating asset inventory that reconciles from the real sources of truth, and behind auditing the CMDB you already have instead of trusting it.
The principle: deliver in small increments, gather feedback, and adjust, rather than attempting everything at once.
Our position: you do not deploy an all-knowing agent on day one. You start with one Coworker doing one useful thing, see the result, and expand. And feedback is built into the shape of the work: every Coworker run leaves its reasoning and evidence on the ticket, so a human can review it and tune what happens next. The outcome compounds, like closing the loop between monitoring, ITSM, and chat to take incident MTTR from hours to minutes, one iteration at a time.
The principle: work across teams and stakeholders, and make information visible so decisions are faster and better.
Our position: a Coworker should collaborate in the open, not act as a black box. When a major incident hits, it opens a triaged war room in Teams and brings people together with context already in place, the same way an on-call AI buddy briefs the responder before they are even online. And visibility is not a slogan here: the Coworker writes every step it took as a note on the ticket, so the reasoning is auditable, not hidden.
The principle: a service is a system. A change in one place affects another, so understand the whole before you act.
Our position: single-tool automation cannot reason holistically, because the truth is spread across systems. Judging a change means seeing its blast radius across the CMDB, which is what change intelligence does. Debugging a broken build means correlating monitoring, logs, and the code repository in one pass, as in the guardrailed infra-debugging run. Finding shadow AI means reading identity grants across every provider, not one, which is the point of shadow AI detection. Holistic is not a nice-to-have; it is the only way the answer is correct.
The principle: complexity is not sophistication. Remove steps that do not add value and keep it usable.
Our position: the least simple thing in most helpdesks is the rule tree, hundreds of categories, triggers, and routing rules that still misfire. That is how every helpdesk becomes a workflow-automation black hole. Reasoning over the actual request is simpler than maintaining the maze. Simple for the employee, too: ask for a cloud server in Slack and it gets provisioned inside policy, no form, no five-day wait.
The principle: optimize first, then automate. Do not automate a broken process, and be deliberate about what should stay human.
Our position: this is the one we build the whole product around, and the order matters. Automate the connective work that carries no downside (read, correlate, draft, recommend), and keep humans on the decisions that carry risk (changing production, revoking access, merging code). That is why the Coworker that finds the exact broken line will still refuse to ship the fix without approval, and why the shadow AI scanner recommends revocation but never revokes. Automation up to the line, a human at the line. That is how you get the speed without handing over judgment.
The seven guiding principles are still the best short answer to how should we run modern ITSM. The test is whether your tools actually build to them or just quote them. Ours come down to a simple stance: meet people where they already work, reason across the whole system instead of a single tool, keep it simple for the human, automate the glue, and stop at every decision that deserves a person. Focus on value, and let the rest serve it.
See how AI Coworkers put these principles to work at atomicwork.com.
Related: Your Shadow AI Isn't Hiding | Every Helpdesk Becomes a Workflow-Automation Black Hole | Your CMDB Is Lying to You
Sources and further reading: Atomicwork ITIL 4 guide; PeopleCert ITIL 4 Foundation; AXELOS ITIL service management.
