The five rungs
Rung 1 — Personal chatbot use
Somebody on the team pastes an angry email into a chatbot and asks for a calmer reply. It works. They do it again the next week. The wins are real — and private, inconsistent, and invisible to the business. The quiet risk lives here too: customer information flowing into personal accounts nobody governs, on terms nobody read.
Rung 2 — Shared practice
The team compares notes. Good prompts become shared templates. Someone writes the two rules that matter: what may be pasted, and what may never be. Accounts move to business terms. This rung is cheap and worth doing this month regardless of what you ever buy — it converts private wins into team habits and closes the data leak.
The paste policy doesn’t need to be a binder. One page does it: customer names and financials never leave governed accounts; anything already public is fair game; when unsure, ask. The point is that a rule exists and everyone has read it.
Rung 3 — The first automated workflow
One trigger, one defined output, and a controlled launch. Pick visible, reversible work from the first-workflow list. Start in review or shadow mode, then allow routine cases to run under policy while exceptions remain routed.
Rung 4 — Connected workflows
Systems start talking: inbox to CRM to books. Value jumps, because whole processes finish without a human ferrying data between tabs. Surface area jumps with it — every connection is a place drift can enter and a place failure can hide, and a chain that half-completes leaves worse data than no chain at all. Logging and exception queues stop being nice-to-haves here.
Rung 5 — Operated systems
Automation becomes an operation: health checks, exception ownership, clear scope, change control, risk-based authorizations, and usable logs. Shared operating patterns make later workflows easier to design and maintain because the team has established how live systems are governed.
| Rung | What it looks like | What breaks if you stay too long |
|---|---|---|
| 1 — Personal use | Individuals pasting into chatbots | Private wins, ungoverned data risk |
| 2 — Shared practice | Team prompts and a paste policy | Savings stay per-person; workflows still manual |
| 3 — First workflow | One trigger, one output, controlled launch | Exceptions and changes need an operating path |
| 4 — Connected | Systems talk to each other | Silent breakage multiplies with each connection |
| 5 — Operated | Monitoring, clear scope, controls, and logs | Maintenance and change management continue |
Tools don’t climb ladders. Operations do — on purpose.
How to climb on purpose
Moving directly from individual use to a connected platform can create more integration surface than the team is prepared to operate. Establish exception handling, ownership, logs, and change practices with a bounded workflow before connecting a larger process.
The same tool can serve several rungs. Maturity comes from the operating discipline around it: what runs automatically, what routes to an exception, who can pause it, how completion is verified, and how changes are introduced.
Each rung generates the evidence the next one needs. Rung two tells you which tasks people actually reach for help with. Rung three teaches you what reviewing automation costs. Rung four shows you where your data seams really are. Skip the lesson and you pay for it one rung up.
Signs you’re ready to move up a rung:
- 1 → 2: more than one person on the team is quietly using a chatbot for work. The practice exists; it just isn’t governed yet.
- 2 → 3: the same task keeps showing up in everyone’s chatbot use. If three people paste the same kind of email every week, that’s a workflow announcing itself.
- 3 → 4: the first workflow runs reliably, exceptions are understood, and the adjacent manual handoff is now the clearest constraint.
- 4 → 5: connected workflows need formal monitoring, maintenance, incident handling, and change ownership to remain reliable.
Climbing deliberately just means someone maps the current rung honestly, picks the next workflow on evidence, and sequences the steps. That mapping is precisely what our AI Opportunity Audit does — it places you on the ladder and hands you the climb order.
Your rung is set by behavior, not software
An office can own an expensive platform and still be on rung one if every useful result begins with one person pasting text into a chat window. Another office can use ordinary tools and operate at rung four because triggers, handoffs, logs, and exception owners are explicit. Inventorying subscriptions will not tell you where you stand. Following one piece of work from arrival to completion will.
Ask practical questions along that path. Who notices a miss? Where does the exception wait? Which routine actions run under policy, and which high-risk actions require authorization? Can another qualified team member explain the workflow when its builder is away? An answer that depends on memory rather than a shared process marks the current operating ceiling.