Know what's running.
3 SIGNALS · 1 LESSON · 1 ACTION · ABOUT 5 MINUTES
A post by Jake Serval got me thinking about how much AI adoption comes down to automation engineering: mapping processes, connecting systems, handling exceptions, and monitoring results.
This week, I'm looking at that work, three lessons from Amazon's research, and how to decide which AI developments deserve your attention.
SIGNAL 01
AI adoption involves a lot of automation engineering
What happened. In “The Rise of the Automation Engineer”, Jake Serval argues that building automation is getting easier, while knowing what to automate and making it work inside a company remains difficult. He describes an engineer who works alongside a team, learns what it needs, and understands how its work crosses departments.
Why it matters. That describes much of what I think AI adoption requires. An AI tool can draft a customer quote, but someone must connect it to current pricing, define discount rules, handle missing information, and get the approved quote into the customer's hands and the company's records.
Because AI can give a wrong answer that sounds reasonable, build in a check from the beginning. Decide what the tool can change, when a person takes over, and who notices a failure. Involve employees who know the exceptions in designing and testing it.
What to do. Trace one recurring workflow from request to delivery with the people doing it. Record the systems, decisions, and exceptions. Choose a step where AI could help, then define how you will check its work. Give the person building it access to the people who understand the process.
SIGNAL 02
Three lessons from AWS on turning AI into value
What happened. My three posts this week drew on AWS's Reimagine: Turning AI into Value. The report describes patterns from interviews with business leaders. My takeaways:
Give the saved time a job. Decide where recovered time should go before the inbox absorbs it. Faster proposal drafting might make room for better discovery calls.
Capture operating knowledge. Your CRM may hold a customer record while the reason for an exception lives in someone's head. Write down that reasoning and name someone to keep it current.
Redesign the path to the customer. A quote drafted in ten minutes can still wait four days for reviews. Examine the handoffs and approvals around the faster task.
Why it matters. Faster individual tasks can leave business results unchanged. The process owner needs to decide what the freed time should accomplish, what knowledge the work requires, and which waits to reduce.
What to do. Choose one change to test in a workflow you own. Track how long it takes to deliver a usable result and how often it needs corrections. Check whether the recovered time goes where you intended.
SIGNAL 03
Keep your organization's work in focus
What happened. AI models, tools, and capabilities change so quickly that keeping up can feel like a second job. A demo makes your current tools look dated. A prediction makes you wonder whether you're already behind. I think it's worth stepping back and asking what your organization needs to do well, and where AI could help.
Why it matters. Your customers still need accurate answers, reliable service, and work delivered when you promised. Those needs give you a basis for deciding what deserves attention. If onboarding is slow because employees cannot find instructions, look for ways AI could help them use the knowledge you already have. Keep that goal in view while evaluating tools.
You can stay informed without reopening your AI strategy every time something launches.
What to do. Write down one business priority for the next 90 days and one way AI might support it. Check the tools you already use: your Microsoft or Google environment may already include AI capabilities that fit the need. Use that priority to decide which developments to investigate. Set aside time to review the news, then return to the work. Check progress against the business result you chose.
THE LESSON
Understand the work well enough to change it
The process owner names the outcome and assigns responsibility for the workflow. The builder works with employees who know the exceptions. Employees report mistakes, and the process owner checks whether results improve. Keep those people involved after the first automation goes live so they can correct problems as they appear.
DO THIS WEEK
Review one workflow in an hour
Bring the person who does the work and the person who manages the systems. Use one recent example.
Map it. Trace the systems, decisions, waits, and handoffs from request to delivery.
Capture an exception. Record what someone checked and why.
Choose a test. Define where AI helps and when a person takes over.
Assign follow-through. Name a process owner, review date, and measures for time and rework. Decide where the saved time goes.
WHEN YOU WANT HELP WITH THE WORKFLOW
Hackademic Solutions helps small and medium-sized organizations map a workflow, capture the knowledge it depends on, and test where AI can improve the result. Book a free 20-minute assessment to talk through a quote, proposal, or onboarding process in your organization.
See you next week.
— Chris Simpson
Hackademic Solutions
The Hackademic Briefing · Know what's running.