Building AI People Can Trust

Imagine this: your sales team is thrilled about a new AI assistant. Then someone asks—what if the bot accidentally exposes our pricing data or internal manuals?

That simple question opens the floodgates:

Are we allowed to record our sales reps for training data? Where does customer PII go? Do vendors really keep our data private? What if the bot gives unsafe product advice?

There are frameworks (NIST AI RMF, OWASP, FTC/CCPA/GDPR) and key practices (data classification, enterprise tiers, vendor diligence, clear disclosure) that can help. It’s on us, the humans, to build responsibly.

The question isn’t “Should we trust AI?” but “How do we build AI people can trust?”

Prototype to Production

I built a working AI prototype that could process PDFs perfectly. Then I tried to move it into production.

That’s when I discovered OpenAI’s API handles PDF processing completely differently than ChatGPT does. The prototype worked great in the interface, but the API? Not so much.

Turns out PDF handling was only added to OpenAI’s API last month – and it’s still pretty limited on file sizes. My prototype hit those constraints immediately.

I pivoted to Claude’s API, which handles PDFs up to around 30MB. Problem solved, but it got me thinking about a bigger issue – always validate API capabilities early, even when the UI version works flawlessly. What works in conversation doesn’t always translate to production code.

This isn’t the first time I’ve hit this kind of mismatch. The gap between demo and deployment can be surprisingly wide.

Two Types of “AI Agents” – And Why the Distinction Matters

Working on AI implementations, I keep running into confusion around the term “AI agent.” Turns out we’re talking about two completely different things.

Type 1: Autonomous AI Agents These are the systems getting all the buzz. An AI agent can perceive its environment, decide which tools to use, and execute actions without constant hand-holding. Think customer service bots that access your CRM, check inventory, process returns, and escalate issues – all while maintaining context and making smart decisions.

Type 2: AI-Enhanced Workflows
This is AI plugged into traditional automation platforms like Zapier, Make, Power Automate, ServiceNow, or custom solutions. The AI handles specific tasks within a larger, predictable process flow.

Real example I’m building: Staff scan shipping labels with a mobile app. AI extracts supplier info, model numbers, delivery dates, and populates our equipment database. Standard workflow automation then triggers notifications to procurement, project managers, and finance.

But here’s where it gets interesting: The system also compares delivery timelines against project schedules. When procurement suggests equipment substitutions for cost savings, AI evaluates whether the new supplier’s lead times will mess up critical milestones. If there’s a conflict, it sends up an alert that can be acted upon.

The key difference: Workflows excel at consistent, repeatable processes. Autonomous agents shine when you need adaptive decision-making across multiple variables.

The most powerful implementations combine both – workflow automation for operational consistency, enhanced with AI agents for complex decisions.

In leveraging AI for business operations, getting this distinction right can save serious time and headaches during deployment.

What are you seeing out there? Are you building agents or workflows?

Stuck On Deployment

I built a nifty AI utility tool for a client that will look into a google folder and sub folders, ingest all documents, then extract names & titles of all the people it finds inside. It’s a quick way for the sales team to comb through historical contracts, SOWs, project plans, etc and find the people we’ve worked with in the past who could become new contacts – even if they’re at new companies.

Scripting the AI prompt took a couple of hours. I needed a tight prompt so the end user doesn’t have to interact with the script yet still receives a tidy output list every time. That turned out to be the easy part.

For this project, the challenge (for me) was the deployment, especially as I learned more about Google’s ecosystem. Linking Apps Scripts to a GCP project, enabling Google Drive API in Cloud Console and adding it to the Apps Script, authorizing the script, etc. The deployment took many more hours than the AI piece. I felt frustrated – right on the verge of having a useful tool, but stuck in the details of deployment.

I eventually set it aside for the night and came back the next morning. I asked AI to create a checklist of EVERY SINGLE detail necessary for setup and deployment. That cracked the case and got me across the finish line.

Using AI to solve the deployment issue was pretty nifty as well.

Thoughts on AI Prompt Engineering

Prompt engineering is an art form – and it’s already a legitimate career path, even if many companies haven’t caught up yet.

As LLMs get more powerful and can handle longer reasoning sessions (we’re talking 10+ minute processing times now), a well-crafted prompt becomes the difference between impressive demos and reliable, production-ready automation.

Sure, anyone can get cool results from conversational agents. But building prompts that deliver consistent, predictable outcomes for business-critical tasks? That requires genuine skill, experience, and strategic thinking.

I’ve seen teams spend multiple hours perfecting a single prompt – and save hundreds of hours downstream. Every word matters. Every sequence matters.

My approach? Treat prompt writing like crafting a compelling essay. Structure, flow, and precision all count.

Here are three game-changing techniques I’ve learned:

Examples are gold. Sometimes showing beats telling by a mile – even for AI. One solid example can communicate what paragraphs of instructions can’t.

Order is everything. The sequence of your instructions dramatically impacts results. Pro tip: put your most critical requirements at the end – that’s what the model “remembers” best.

Test relentlessly. Great prompts emerge through iteration, not inspiration. Build, test, refine, repeat.

There are fantastic tutorials out there (easy to find, though mastery takes practice), and tools like Promptmetheus or Originality can accelerate your workflow. But I’d recommend starting with manual practice first – understanding the fundamentals makes you a better prompt engineer long-term.

How’s your prompt engineering journey going? Are you seeing it become more important in your work too?

Framework for Finding ROI from AI

The highest value AI agents often aren’t the flashiest ones, but rather those that eliminate friction in existing processes.

I’ve had clients come to the table with ideas of things to build, but 50% of the time it’s not the most valuable agent for their business. How do you find the highest ROI?

My framework: MAP → IDENTIFY → PRIORITIZE → BUILD

Map: Start by building a customer journey map on Figma. Get the high-level view first, then drill into the details.

Identify: Look for repetitive, time-consuming steps with heavy text or voice components. With voice agents expanding rapidly, audio touchpoints are prime opportunities.

Prioritize: Focus on friction points that impact the most customers or consume the most resources.

Build: The cost to run an AI agent is negligible compared to development cost, so start with your highest-impact opportunity.

Real example: A SaaS company wanted a complex lead scoring agent. But mapping their journey revealed the real bottleneck was customer onboarding. A simple FAQ agent reduced their support tickets by 40% and freed up their team to focus on strategic accounts.

By starting with the customer journey instead of the technology, you’re more likely to land on solutions that drive real value. The opportunities we find this way are usually easier to build AND deliver higher returns.

Prompt Learning

Writing the prompt can be so much more than a line-by-line conversation with Claude or ChatGPT. Emerging research suggests there are ways to vastly increase accuracy. Things like describing its role, assigning a clear task, with specific rules, providing context, examples, and supplemental notes, can all improve the results. This is what makes low-code and no-code systems run with repeatability without hallucination. It takes time to build a first-class prompt, but the payoff is reliability and accuracy.

Here is a prompt I wrote asking Claude to create a Chrome extension for a password manager, storing my files in a local database. (I actually like Chrome’s native pw manager, but wanted to see if I could build an extension this way.) I tried this prompt in ChapGPT as well, but the experience was better with Claude. Copy/paste and watch what happens!

#Role
You are a web developer with a talent for Chrome extensions and JavaScript. You understand browser security and UI/UX design principles.

#Task
Create a personal password manager as a Chrome extension that safely stores passwords locally and helps me automatically save and fill credentials for websites I visit.

#Specifics
Detect login forms on webpages automatically
Prompt to save credentials after I’ve filled in both fields and attempt to log in
Prompt to autofill saved credentials when I return to sites
The save/retrieve prompt should stay visible for at least 8 seconds
Local encryption of all password data
Ability to export/import password data for backup
Clean, intuitive UI for managing all saved passwords

#Context
I currently use a third party password manager but want something I control myself for added security. I’m concerned about storing passwords in third-party services, even when they claim to use encryption. I need this to work with standard login forms and prefer simplicity over complex features.

#Examples
##Example 1
When I visit a site like twitter.com and enter my username/password, after I click “Log in” or press Enter, your extension should show a prompt asking if I want to save those credentials. The prompt should stay visible long enough for me to make a decision (at least 8 seconds).
##Example 2
When I return to a site where I’ve saved credentials, as soon as I click on the username field, your extension should prompt me asking if I want to fill in my saved credentials. If I click “Yes,” it should automatically fill both the username and password fields.

#Notes
– I’m comfortable with technical details but prefer code that’s well-commented and organized
– Security is my primary concern – all passwords should be encrypted locally with a master password
– The extension should work on most standard websites
– Keep the UI clean and straightforward
– I want to be able to view, edit, and delete my saved passwords through the extension
– Must work with Chrome

It Ain’t Sexy

It ain’t sexy, but it’s practical. I asked AI to write a bit of code to ingest a data file, parse and post the data to another database. The parsing rules are complex with many exceptions, which is why we’ve been doing it manually for years with an admin person.

A few hours to carefully construct the prompt, then maybe 5-6 hours testing and debugging; now it’s automated and saving up to 5 hours/week!

AI enabled this. Creating space for the human to do more high-value thinking.

Community

One of the five Things I Value Most is Community. What do I mean by Community? For me, it means bringing people together, usually building together or working toward a shared goal.

This month I started a men’s activity group, inviting some of the men in my local neighborhood to go hiking once a month. I’m calling it the IMG – the Ithaca Men’s Group, after the street we live on. So far only a half dozen in the group, and hopefully more will join.

Why is Community important to me? I guess I’m a social creature like anyone else, but I have some unfortunate mental wiring that makes me perhaps overly sensitive about intruding on other people’s time. It’s like my consideration gene got over-amplified to an extreme. Rather than walk up to someone and start a conversation – which can feel like an invasion (even though a rational part of my brain knows it isn’t) – it feels easier to organize an event or activity that could be valuable for others and invite them to join. In this way, I get the social interaction I crave, I get to meet new people, and if dig deeper I guess I get to control the circumstance of the interaction. It feels less scary, because they’re only coming if they want – no ‘intrusion’ necessary.

I’ve done this multiple times throughout my life. With the band, organizing running relay teams, mapping out a 10k, etc.

On a base level, it’s clearly self-serving.

There is another element, thought, which is I do believe we are stronger when we are connected. I feel more grounded, secure, and confident when I can say hello to neighbors with whom I have an actual relationship. Even if it’s a small relationship, I know I can ask a favor if needed, and they know they can ask me. These small bonds create a powerful network.

So Community continues to rise to the top as one of the things I value most.

I’ve Got Your Back

Today is Michelle’s birthday. (Yes, every year I get the double whammy of Valentine’s Day and Birthday on the same day. Better get those birthday dinner reservations WAY in advance…)

Last night I wrote a birthday card and somewhere in the lines I told her “I’ve got your back.”

This morning, in my email feeds I saw Seth Godin’s daily blog. Today’s title is I’ve Got Your Back. It’s a pretty good definition of what it means to have someone’s back. Linked above and credited to Seth Godin, here are his words:

This is a complicated promise. It’s about commitment and connection and most of all, time.

If we’re saying that we’ll do what’s in our short-term interest and convenient, then there’s really no reason to say anything at all, since that’s what we usually do anyway.

Instead, we’re promising to shift our time horizon. To show up when we don’t feel like it, especially then. To invest focus and time and resources when there may be other more compelling short-term options.

Commitment is a reward in itself. It gives us boundaries and structure, and also creates meaning. Commitment only counts when it costs us something, and that cost usually involves shifting time.

Because we said we would.

Sometimes a Distraction

I got back on my snowboard for the first time in two years. Like riding a bike, my brain and body remembered the skills after a few runs. I didn’t feel comfortable, though, and nearly caught an edge multiple times. After three or four runs, I still wasn’t really having fun. If it had been a powder day, or warmer temperatures, that might have changed things, but the snow was crunchy (thankfully not icy) and the weather was quite cold.

Then I put in my earbuds and turned on Dave Matthews ‘Live at Redrocks’ album. I guess it was just enough distraction – I could no longer hear the snow crunching beneath my board, and a portion of my brain was keyed into the music – that it took the edge off. Each run after that was far more relaxed, which of course improved my riding, which of course made everything more fun.

Sometimes a distraction is just the thing.

By the way, the NB bus from Boulder to Eldora is my new preferred method for getting up there. No concerns with parking, no hiking across parking lots carrying gear. The bus drops off and picks up right in front of the Alpenglow lift, and it was super easy and comfortable. Buses every 2 hours each direction. If you’re headed to Eldora, definitely check it out.

Guatemala Part 4 – Good for the Soul

I was nervous about our trip to Guatemala. As of December 2024 the U.S. State Department gives the country a Travel Advisory rating of 3: Reconsider Travel due to crime. All the guidebooks we found clearly mentioned taking safety precautions against bandits that target the popular tourist destinations. Michelle convinced me, though, that we could manage the safety concerns. Boy am I glad she persevered. I did not feel unsafe at any time. It felt like traveling in any other poorer country. Just pay attention, be smart about where you go and when you go, and use guides when recommended.

Michelle and I got a lot of concentrated quality time together, something we had been missing especially since the start of her new job, the start of the school year, and since my kids both left for college. We’ve been successfully cohabiting and even co-working: no issues and we’ve held true to our Friday night date nights. But this kind of concentrated time – navigating a new country and having an adventure together – created a special space for us to reconnect doing something we both love. Not only the hiking and mountain biking, but sampling local cuisine, learning some of the local history, negotiating with street vendors and tuk-tuk drivers, interacting with Spanish-only speakers with each of us having limited Spanish ability. There was time for deeper conversations about our families, our careers, health and fitness, retirement plans, current events. It’s not like we’ve been strangers living together, but we took full advantage of the time to be present with each other and go deeper than what’s available amidst the hustle of life.