INBOX INSIGHTS: Local AI Migration Part 4, Practical Responsible AI Part 3 (2026-09-23)

INBOX INSIGHTS: Local AI Migration Part 4, Practical Responsible AI Part 3 (2026-09-23) :: View in browser

👉 Take our new course, Introduction to Deep Research with AI!

Watch this video on YouTube

Local AI Migration: Part 4

Why is this taking so long?

Last week I told you the requirements document would be finished this week, and that part 4 would be what’s in it. On Thursday, September 18th, I approved it. That’s twenty-four days after I wrote the first sentence of the user story.

Here’s what has changed about how I work since then: nothing. I haven’t installed anything, bought anything, or moved a single task off the tools I use today. However, I have written an impressive amount about all three.

I suspect some of you will be surprised at how slow this is. I’m surprised too. But not really suprised, because it’s me, hi. I’m the problem, it’s me.

What’s in the document I promised

The PRD is solid at about 11,000 words. It is a good visual representation of how I am overthinking this whole initiative. The short version fits in four decisions.

1) Buy nothing. The 2021 MacBook Pro on my desk already handles everything I’d want to run locally, and its manufacturing carbon is already spent.

2) There are four places work can run, not two. A script with no AI at all, my own laptop, a cheaper hosted model Chris pointed me toward, and the frontier model I use today. I started this project thinking it was a choice between the last two.

3) Nothing moves until it passes a test, and the test has no time limit. I’ll give up speed. I won’t give up quality. Every candidate has to produce work as good as what I get now, judged against a checklist I write before I see any results, so I can’t quietly move the bar.

4) My most valuable work stays exactly where it is. Strategy, synthesis, anything feeding a decision. That’s on purpose, and it turned out to matter more than I expected, for reasons I’ll get to.

There are two kinds of people

With any new initiative there are two kinds of workers, camps, that you’ll run into. First is the cautious documenter (that’s me). Second is the doer without caution (that’s Chris).

If you ask me, documentation comes first. You write it down, think it through, and find out what you’re wrong about before it costs anything. Documentation is my comfort zone. For example, there’s a folder for this project with more than two dozen files in it and nothing has happened yet.

If you ask someone like Chris, you do it now and figure out the documentation later, if ever. He’d have had a local model running the afternoon I wrote the user story.

Both camps have a failure mode, and this project has now shown me both.

My camp saved me real money. Three days into writing requirements, I found out my premise was wrong. Running a query on one machine is far less efficient than running it in a data center, where many queries share the same hardware at once. If we’d done this Chris’s way, I’d have bought a new laptop to solve a problem it would have made worse. Chris’s camp would have caught the other failure. Nearly four weeks in, the only thing that’s moved is paper.

There’s nothing left to plan

Both failure modes are real, but only one of them is still happening. The premise mistake got fixed for the price of three days. The paper-only-moving problem is still true right now, while I’m writing this.

I’ve documented enough. My own requirements document already has the tests in it: what I’ll run, what counts as passing, what I’m measuring it against. Running them is the thing that’s actually due next, not another version of anything.

That’s going to take a few weeks, and it’s going to feel like procrastination with better paperwork behind it. It isn’t. The log I started this week is how I’ll know the difference, it’s what turns “I’m testing” into something I can check instead of something I can just claim.

If this were somebody else’s project, I wouldn’t let them sit on a finished plan for four weeks. I’d tell them the documentation phase is closed, so start running what you wrote. I owe myself the same lack of patience.

Doing something means committing to something

This is the part I didn’t want to write.

I’m a creature of routine. Routine is safe and predictable, and I’ve built a lot of my working life around making things predictable for other people. That’s not the same as being unable to change. I’ve changed constantly, and I’ve steered this business through more economic turmoil than I’d like to count. But I only recently got comfortable with frontier models. I learned how to get what I need out of them, I built my processes around them, and I got good at it.

Now I’m disrupting that on purpose, and holding myself publicly accountable for it. I’ll be honest: I don’t love it.

Writing another document doesn’t commit me to anything, but changing how I work does, which is the real reason the paperwork keeps growing. The uncomfortable detail is that I saw this coming. My own requirements document has a risk in it called planning as avoidance: planning produces publishable content, and publishable content is an incentive to keep planning. (You’re reading the evidence.) I even wrote the test for it. The question isn’t whether I’ve planned enough. It’s whether the next document is going to change what I do, or only how confidently I describe it.

The first change doesn’t touch the part I’m afraid of losing

In the 5P Framework by Trust Insights™ (Purpose, People, Process, Platform, Performance), my fear lives in Process. I’m not attached to a tool. I’m attached to a way of working that I finally have under control.

When I reread my own requirements with that in mind, the answer was already in there. The work I care most about, the thinking and the judgment, stays on the frontier model. It isn’t eligible for optimization at all. What moves first is the clerical work: converting files from one format to another, regenerating reports after the analysis is already done, and fetching data before I think about it.

That isn’t disrupting the process I mastered so much as taking the chores off it.

What I actually did this week

I started a log that is one row for each piece of AI-assisted work: what it was, how long the AI part took, and how long I spent editing the output afterward. The goal is to spend under a minute a row using estimates. I don’t need exact figures for this.

This exercise changes nothing about how I work. In a few weeks I’ll need to know whether the new way is better than the old way, and I can’t answer that without a baseline.

If you’re leading an AI initiative, you have two camps on your team, and they’re probably annoying each other right now.

The usual mistake is to let them fight it out, or to pick a winner. The more useful version is to decide in advance which camp owns which phase. Documentation owns the decision about what to build and what “better” means. Do-it-now owns the first attempt, once that decision is made. Each camp is good at the phase the other one dreads.

Planning is how I avoided a bad decision. It can’t also be how I avoid a good one.

Which camp are you in, documenting or doing?

Reply to this email or join the conversation in our Free Slack community, Analytics for Marketers!

– Katie Robbert, CEO

Weekly One Question Poll 🐶

Click or tap just one answer below — there’s no form, this is the entire thing. One tap and you’re done, with our thanks.

How are you managing the expectations of using AI?

Do you have a colleague or friend who needs this newsletter? Send them this link to help them get their own copy:

https://www.trustinsights.ai/newsletter

In this episode of In-Ear Insights, the Trust Insights podcast, Katie and Chris discuss how everyday users mishandle AI research tools and waste valuable time on unhelpful outputs. You will discover why vague prompts trigger false answers and drain your resources. You will learn how to set strict boundaries so machines gather the data you need. You will master a simple framework that turns scattered findings into reliable plans. You will build a clear workflow that catches errors before you publish anything.

Watch/listen to this episode of In-Ear Insights here »

Last time on So What? The Marketing Analytics and Insights Livestream, we looked at the agency of the future and how billable hours must go away. Catch the episode replay here!

This week on So What? we’re Examining how to use deep research for sales outreach. Are you following our YouTube channel? If not, click/tap here to follow us!

Here’s some of our content from recent days that you might have missed. If you read something and enjoy it, please share it with a friend or colleague!

Take your skills to the next level with our premium courses.

👉 Take our new course, Introduction to Deep Research with AI!

Get skilled up with an assortment of our free, on-demand classes.

You could be reaching 37,000+ marketers, analysts, data scientists, and executives directly with your ad. Want to learn more? Reach out and contact us.

This week, we dig into part 3 of 5 about what responsible AI is in practice. Responsible AI is one of those amorphous terms that can mean whatever the speaker wants it to mean, and it has perilously few practical tactical things we can do.

In this five-part series, we aim to correct that by providing practical tactical things you can do.

We have five core practices you can use to reduce your AI impact:

Use AI for what it’s good at.
Use cloud AI as little as possible.
Use AI for the tasks you hate.
Use local AI as much as you can.
Prohibit AI for what it’s bad at.

Part 3: Use AI for the tasks you hate

In this third part, let’s address an angle of responsible AI a lot of companies and organizations miss. Many AI deployments tend to be hasty affairs, with leaders exhorting staff to use AI for everything, to find efficiencies, etc.

However, this tends to be a fairly undifferentiated ask, and one that lots of folks resent. Why? Because we’re often told to use AI for the things we’re good at and enjoy. This is especially true for creative tasks, like writing, art, music, etc., tasks that people have invested a lot of time, effort, and expense in themselves to be proficient at.

One of our guiding rules, built right into the TRIPS Framework by Trust Insights is that the more painful a task is, the better a candidate it is for AI. What are the tasks that no one will miss, if machines start doing them? What are the tasks you dread, the tasks you just would wish you could delegate to someone else?

When it comes to AI, those are the tasks that should be on the chopping block. A writer doesn’t want to give up their favorite parts of writing, but they’d happily give up filling in their time sheets. A musician doesn’t want to hand composition of music to a machine, but they’d gladly never fill out an expense report again.

At my old agency, the system for filling in time sheets was so obtuse and difficult to work with that it routinely took staff an entire hour to navigate the ERP system that controlled our timesheets. No one enjoyed that task, and it was so onerous that some teams simply lied just to get through the process faster. No business generates revenue from employees filling in timesheets, not directly (yes, they track and bill hours, but in the agency world almost no one does that accurately either). So if we can hand off tasks like that to AI, we can not only recover valuable hours to be spent on more meaningful tasks, we will also make ourselves and our people deliriously happy.

If you’re using and deploying AI to make people unhappy, that’s irresponsible use of AI, and no matter what productivity or efficiency gains you might make in the short term, in the long term it damages the business.

If you’re using and deploying AI to make people happy, you might see fewer “quick wins” but you’ll have happier teams and retain more valuable people for longer.

Here’s a practical exercise to do, one I recommend everyone do. Keep a running list in the format of your choice of each task you do every week, and then score it 1-5, where 1 is your loathing of that task and 5 is your giddy eagerness to indulge in it. Any task that scores under a 3 is a task you should be asking your AI tool of choice how to automate.

If you’re not sure what tasks to even consider, take a screenshot of your calendar software, take a photo of your paper diary, or capture your weekly timesheet and put it into the AI of your choice with this prompt:

Today we’re doing task management analysis using the TRIPS Framework by Trust Insights (https://www.trustinsights.ai/trips) which you can retrieve with WebFetch. Our goal is to identify which tasks are highly repetitive, likely have sufficient examples for AI to learn from, consume the most time, are not core to driving business value in my role as a {your job title} and are the most unpleasant for me. Based on the calendar shown here of my day to day activities, infer what tasks have tangible outputs, then ask me up to 20 clarifying questions until you have enough information to succeed at the task. Then assign a score of 1-5 for each of the TRIPS dimensions and show me a table in descending order by score.

If you do this exercise with a month’s worth of calendar entries, you might be surprised at what things AI could take off your plate that you’ve forgotten about or are so repetitive you block them out.

Give AI the work you don’t want to do.

Almost every AI course is the same, conceptually. They show you how to prompt, how to set things up – the cooking equivalents of how to use a blender or how to cook a dish. These are foundation skills, and while they’re good and important, you know what’s missing from all of them? How to run a restaurant successfully. That’s the big miss. We’re so focused on the how that we completely lose sight of the why and the what.

This is why our capstone course, the AI-Ready Strategist, is different. It’s not a collection of prompting techniques or a set of recipes; it’s about why we do things with AI. AI strategy has nothing to do with prompting or the shiny object of the day — it has everything to do with extracting value from AI and avoiding preventable disasters. This course is for everyone in a decision-making capacity because it answers the questions almost every AI hype artist ignores: Why are you even considering AI in the first place? What will you do with it? If your AI strategy is the equivalent of obsessing over blenders while your steakhouse goes out of business, this is the course to get you back on course.

👉 Take this course now to become an AI leader

Here’s a roundup of who’s hiring, based on positions shared in the Analytics for Marketers Slack group and other communities.

Are you a member of our free Slack group, Analytics for Marketers? Join 4000+ like-minded marketers who care about data and measuring their success. Membership is free – join today. Members also receive sneak peeks of upcoming data, credible third-party studies we find and like, and much more. Join today!

Where can you find Trust Insights face-to-face?

LPA, September 2026
MEMA, October 2026
Women in Engineering, October 2026
MAICON, October 2026
SMPS AEC.AI, November 2026
MarketingProfs B2B Forum, November 2026

Going to a conference we should know about? Reach out!

Want some private training at your company? Ask us!

First and most obvious – if you want to talk to us about something specific, especially something we can help with, hit up our contact form.

Where do you spend your time online? Chances are, we’re there too, and would enjoy sharing with you. Here’s where we are – see you there?

Read our disclosures statement for more details, but we’re also compensated by our partners if you buy something through us.

Imagine a world where your marketing strategies are supercharged by the most cutting-edge technology available – Generative AI. Generative AI has the potential to save you incredible amounts of time and money, and you have the opportunity to be at the forefront. Get up to speed on using generative AI in your business in a thoughtful way with our workshop offering, Generative AI for Marketers.

Workshops: Offer the Generative AI for Marketers half and full day workshops at your company. These hands-on sessions are packed with exercises, resources and practical tips that you can implement immediately.

👉 Click/tap here to book a workshop

Some events and partners have purchased sponsorships in this newsletter and as a result, Trust Insights receives financial compensation for promoting them. Read our full disclosures statement on our website .

Thanks for subscribing and supporting us. Let us know if you want to see something different or have any feedback for us!

Need help with your marketing AI and analytics?

Leave a Reply Cancel reply

Pin It on Pinterest

Watch This Newsletter