Isaac Bonney

I’m working on real engineering problems at the intersection of geotech, AI, energy, and infrastructure. My focus is simple: learn fast, build real prototypes, and turn classroom knowledge into real-world progress. I’m an engineering student by training and a builder by instinct. I’m especially interested in what happens below the surface, where many of the hardest and most impactful constraints live. This account is where I share key lessons from school, document active build work, and pressure-test ideas in public as they move from concept to reality. Longer-form breakdowns, walkthroughs, and build updates will also be shared on YouTube and LinkedIn. If you’re a student, engineer, or builder interested in serious systems and meaningful infrastructure, follow along and join the conversation. I’m always open to learning and collaborating. #Engineering #Geotech #Infrastructure #AI #Energy #STEM #EngineeringStudents #Builders #LearningInPublic #YouTube #Linked

Could 30 minutes a month with Bob prevent your next $100,000 mistake? Maybe. But don’t ask Bob to explain his whole job. Ask: “What mistake gets expensive fast, and how do you know it’s coming?” That’s the knowledge I’d capture first.

$13,000 a year in senior engineering time. That’s what five repeat questions a week can consume. The problem isn’t that experts are needed. It’s that expensive expert time gets spent on questions that shouldn’t require expert judgment every time. Capture what can be reused. Keep the expert focused on technical decisions, clients, mentoring, and the hard problems that actually need their experience. That’s where knowledge capture starts creating business value. #EverGX #KnowledgeManagement #AIReadiness

One thing I got wrong in my first AI knowledge-capture test: I treated SME validation like the finish line. It isn’t. A better path is: SME validation → peer review → usability test → approval The goal isn’t to capture more knowledge. It’s to create knowledge other people can actually trust and use.

What if your search worked perfectly, but you still couldn’t answer the client? You find: → Proposal_Final.pdf → Proposal_Final_Rev1.pdf → Proposal_Client.pdf → Proposal_Updated.docx The files are there. The problem is knowing which one is actually right. #EverGX

Can 5 avoidable minutes really add up to $30,000 a year? Illustrative example: 50 employees × 3 FIND events/week × 5 minutes × $50/hour × 48 weeks = 600 employee-hours = $30,000 modeled labor cost Not measured ROI. Replace the assumptions with your own.

Yesterday I ran a simple 60-second test on my own files. I found plenty of resumes. The harder question was: Which one was the correct, current version I should actually use? Finding a file isn't the same as finding the right information. #EverGX

Can your employees find the correct, current file they need in 60 seconds? I tested it on myself today. Most files were easy. My resume failed badly. Too many versions. Poor folder placement. One filename mistake. And some copies were sitting in Google Drive where Windows Search couldn’t see them. The file existed. Finding the right one was the problem.

I did the hard work. Then the system still couldn’t find it. While I was in the Army, before there was #AI that could help, I had to write a new Take-home Laptop security policy after someone left a Government laptop on a train. I expected to build on existing guidance. We're taught never to recreate the wheel. Instead I found outdated references, documents that wouldn’t open, and policies I knew existed that simply never appeared in search. I rebuilt the entire policy from scratch and spent long hours getting it right. Once approved, I numbered it correctly and saved it in the proper #SharePoint location. Anyone who already knew exactly where to look could open it. But the search itself remained inconsistent and unreliable. That was the real frustration. What’s the point of creating the correct document if people still can’t find it when they need it? I see the same pattern inside #architecture and #engineering firms: → A detail that already exists gets redrawn because no one can locate it → A hard-won decision loses its “why” the moment the project closes → A new team member spends the first days just trying to find the current version of key files This quiet cost rarely shows up as a line item. It simply burns hours on project after project.AI tools only amplify the foundation you give them. If the foundation is hard to search, the answers stay incomplete. What’s the last piece of project knowledge your team already created but still struggles to find? #AEC #KnowledgeManagement #AIReadiness #Microsoft365 #Findability

The detail already existed. It was even drawn correctly last time. This team just couldn’t find it, so they redrew it under deadline. Again. That’s not a design failure. It’s a findability failure. What’s the last thing your team had to recreate because nobody could locate the original?

Serious question: Is #Robinhood disrupting traditional brokerages like Charles Schwab and Merrill Lynch, or do the established firms still have the edge? What’s your experience?

Everyone's rushing to add #AI at work. Almost nobody's ready for what it does first: AI won't fix your knowledge problem. It will reveal it. Point an AI agent at your organization and it surfaces everything you've been living with: • Conflicting policies • Tribal processes • Buried SharePoint pages • SOPs nobody updated • Files with no owner AI doesn't know your organization is messy. It just retrieves what exists. Confidently. The teams that win with AI won't be the ones with the best prompts. They'll be the ones with knowledge worth retrieving. What's one place AI would embarrass your organization today?

Maybe I'm just old-fashioned here... If you don't document your decisions, will you remember: -why you chose A over B? - how critical the decision was? - the consequences that could've resulted? The written word is what we can pass on to help those who come after us. If you take the time, and also do Lessons Learned, it can help teach along the way.