This Week's Focus ⤵️
AI definitely took off like wildfire in 2025.
It is faster, better, and more capable than ever.
That being said, we are still far from ubiquitous enterprise AI, and definitely far from the kind of AGI some futurists talk about.
Most enterprises are still barely scratching the surface of what’s possible, not because they lack ambition, but because their systems, incentives, data models, and governance were not built for continuous, high-velocity iteration.
The good news is that the best capabilities are no longer locked inside AI labs.
Models, connectors, design tools, and coding copilots are now mainstream. And while I still recommend enterprises use top talent to implement these tools responsibly, you don’t need huge teams anymore. Small, crack AI teams of 3 to 4 people can run circles around bigger, slower groups if they know what they’re doing.
That’s the new game. And we are just getting started.
One of the tools I personally used a lot this year was Magic Patterns. My internet friends Alex and Teddy founded it in 2023, and in November they raised a $6M Series A after bootstrapping the company up to that point.
Their year-end stats email said I landed in the top 5% of users. I crafted close to 2,000 code-specific prompts in 2025, and those prompts translated into 37 separate enterprise POC starter designs. I’m kind of proud of being one of their top users. 😀
I’m sharing this because it captures what I believe to be an essential component in enterprise value creation using AI: prompt velocity is becoming a real competitive advantage, but only when it is paired with deep business understanding and real engineering discipline.
A lot of what I do at Sprinklenet is lead small AI teams to build stable, integrated, high-impact dashboards and intelligence layers that help executives and operators connect data dots faster. But before we write production code, I almost always do the business process mapping and exploration myself. That means talking to stakeholders, executives, operators, internal technology owners, admins, and sometimes external vendors. We build a clear picture of what’s happening, what’s breaking, where the friction lives, and what decisions are being made with incomplete information.
Then comes the sprint.
Once I understand the current business and data model, I jump into tools like Magic Patterns and do these intense, focused sessions where I try to imagine what the next generation of the system should look like. I mean intense. I’m usually exhausted after two or three days of super focused prompting. Then I step back, get stakeholder feedback, validate assumptions, and run another cycle.
The output is not just screens; it’s clarity. What the real product should be, what the data contract needs to look like, what workflows matter most, and what we should build first so we can ship value quickly without creating risk.
That’s why I like being a power user. Teddy and Alex have helped me clean up code, narrow scope when an MVP starts growing teeth, and keep the work grounded in what a real team can actually deliver.
So, if you missed my last few editions, this one is a simple 2025 recap: the year I doubled down on iteration as a system, not a vibe, and why I believe 2026 will reward the companies that can iterate fast without getting sloppy.
Build systems with AI and use AI to deliver better, faster, easier to access results.
Diving Deeper 🤿
What changed in 2025? A lot.
AI got good enough to feel inevitable.
The models improved, sure, but the bigger shift was in tooling. The design-to-code loop got tighter. Integration patterns got clearer. Evaluation tooling matured. The ecosystem became usable for real teams, not just researchers.The bottleneck moved from “model quality” to “organizational throughput.”
Most enterprise initiatives don’t fail because a model is weak. They fail because the org can’t move: unclear ownership, brittle data definitions, slow approvals, unclear security boundaries, no feedback loop with actual users.Small teams gained massive leverage.
A focused team that understands the business, has a clean data access plan, and can iterate daily can now outpace a much larger team that ships quarterly.
Core Principles in Practice 💡
The Sprinklenet Loop
This is the workflow I have refined all year. It looks simple, but doing it well takes discipline.
Discovery
Talk to the people who live in the workflow. Define outcomes, constraints, KPIs, and risk boundaries.
Data Reality Check
Identify systems of record, permissions, identity model, and what “truth” actually means in this environment.
Prototype Surfaces Fast
Use rapid UI and code prototyping to force clarity early. Make assumptions visible.
Validate with Users
Get real feedback, not internal opinions. Watch where confusion shows up.
Lock the Data Contracts
Define stable IDs, event definitions, and minimal required fields.
Engineer the Production Path
Build the integration, guardrails, auditability, and role-based access like they are core product features.
Ship and Measure
Deploy something small that reliably improves a decision or reduces effort. Measure adoption and impact.
Repeat
Do it again, get faster and better.
If you can’t describe what you are doing as a process, you don’t know what you’re doing.
Legacy Spotlight 🔧
In 2025 I got more convinced of something I’ve suspected for a while: most legacy systems are not “bad.” They’re simply not designed for the kinds of questions leaders now want to ask in real time.
Older systems were built for recording.
Modern businesses need systems built for understanding.
That gap is the opportunity. You don’t always need to rip and replace. You can often build an intelligence overlay that respects the system of record and makes the organization smarter without destabilizing the foundation.
Balanced & Insightful ⚖️
If you’re leading enterprise AI initiatives, here’s the honest tension: you need speed and you need control.
Speed without control creates risk and distrust.
Control without speed creates irrelevance.
The win is a system that can move fast inside boundaries.
A simple decision filter I like:
Start with an overlay when:
You need value fast
You cannot pause the business for modernization
You have usable systems of record
You can define permissions and audit trails cleanly
Consider deeper modernization when:
Identity is a mess and no one trusts the data
Definitions vary by team and no one can agree on “truth”
The system itself blocks any meaningful integration work
AI trends I’m watching going into 2026
Agents Will Be Real, But Only in Governed Environments
The dream is “tools that do work.” The reality is “tools that do work with permissions, logs, approvals, and limits.” Enterprises that build the governance layer now will move faster later.Retrieval Quality is Becoming a Board-Level Concern
Not because boards care about chunking, but because reliability and auditability are the difference between a trusted assistant and a liability.UI Matters Again
We are about to see a wave of enterprise “intelligence layers” that sit above legacy systems. The winners will not be the ones with the most features. They will be the ones that reduce cognitive load and make decisions easier under pressure.
Jamie Thompson
If you’re an enterprise leader and you want 2026 to be the year you move from AI interest to AI outcomes, my recommendation is simple:
Pick one high-value workflow.
Define what “better” means in measurable terms.
Build a small, crack team.
Run the loop weekly.
Ship something useful in weeks, not quarters.
And if you want help doing that, I’m looking for two or three enterprise design partners for early 2026. The best fit is a team that has meaningful legacy systems, serious data, and a real appetite to move fast with discipline.
Need Expert Guidance?
Book a focused 1-hour strategy session with Jamie.
✅ Evaluate current architecture and readiness
✅ Identify quick wins and hidden risks
✅ Get tailored, actionable next steps
👇🏼 Book a Strategy Call
Jamie’s Weekly Spotify Mix
My favorite genre of music is classic American rock, but a close second is Spanish crooners like Julio Iglesias and Luis Miguel, plus classic Latin pop from Shakira and Maluma. I listen to all kinds of music.
This week’s mix is a handful of my go-to tracks when I want Spanish beats in the background.
Keeping it real. Enjoy the rest of your day.
🎺🎧 Note: Web Edition Only
