Last week I promised this week would be about concrete side project ideas for technologists in 2026.
I'm going to break that promise, sort of. Because after spending a week thinking about it and reading the search queries that are bringing people to this site, I realized that giving you another "10 ideas for developers" list would be exactly the wrong thing to do.
Here's the problem with every side project ideas article you've ever read: it was written by someone who has no idea what you do for a living.
They don't know that you spent the last three years debugging a specific kind of telemetry pipeline at an automotive OEM. They don't know that you sit in weekly meetings where compliance teams complain about the same CSRD reporting gaps. They don't know about the internal tool everyone in your company secretly builds workarounds for.
And that's where the real ideas live. Not in random lists shared by random people on the internet.
The Unfair Advantage You're Ignoring
Every technologist I know has three sources of insight that outsiders cannot replicate:
Inefficiencies you see daily. The repeatable friction in your work that you've stopped noticing because it's "just how things are." An outsider would build a tool to fix it but since you got so used to it, you just accept it.
Insider knowledge. Things that are obvious to you, like regulatory nuances, data structures, workflows, vendor politics are completely invisible to anyone without time inside your domain.
Translation edges. Let’s say you sit between two worlds. Software and automotive. Cloud and compliance. Frontend and finance. The interface between domains is where most high-leverage products live, and you're one of the few people standing at it.
Anyone building a product in your domain without these three things is at a permanent disadvantage. And that's how moat is built.
The problem is: you can't see your own moat clearly. It's invisible to you because it's your daily environment. You need a structured way to surface it.
Enter the Agent
Here's where 2026 changes things. You don't need to sit alone with a notebook trying to extract your own insights. You have a thinking partner that can interview you, structure your answers, generate ideas against a defined framework, and score them against explicit rubrics, all in about 45 minutes.
I've been running this kind of structured session with Claude for my own product thinking, and it works remarkably well if the prompts are engineered correctly. Generic "brainstorm with me" prompts produce generic ideas. Structured, multi-phase prompts that lock the agent into a specific role and prevent it from drifting produce genuinely useful output.
So I built the workflow and I'm giving it to you for free.
The Sovereign Idea Workflow
The Sovereign Idea Workflow is a 5-phase prompt sequence you run with any frontier AI assistant (Claude, GPT-4, Gemini, etc). It takes 45–60 minutes end to end to run it.
Phase 1: Domain Extraction. The agent interviews you about your actual work that you do in your day to day job. Six structured questions that surface your inefficiencies, insider knowledge, privileged access, and target peer persona. Output: a Domain Context Map.
Phase 2: Idea Generation. The agent takes the Domain Context Map and generates 5–7 product ideas, each required to have one of three specific edges: an inefficiency edge, an access edge, or a translation edge. Generic "AI wrapper" ideas are explicitly disallowed in the prompt.
Phase 3: Scoring. Each idea is scored on three weighted dimensions: Unfair Advantage, Market Signal, Weekend-Validatable. The rubrics are specific, a "5" on Market Signal requires visible paid demand, not "I think people would buy this." This is what prevents self-deception.
Phase 4: Shape Canvas. The agent fills out a complete Shape Canvas (problem, user, constraints, definition of done, success signals, kill signals) for your top-ranked idea. This is the first artifact of the Building Loop, the system I've been writing about for weeks.
Phase 5: Business Profile. Because an idea without a commercial lens is a hobby. The agent profiles the competitive landscape, sizes the market (TAM / SAM / SOM), recommends a monetization model with realistic price anchors, builds three 12-month revenue scenarios, maps the most likely failure modes, and gives you a direct go / no-go / adjust recommendation.
At the end you have: a map of your domain, a ranked shortlist of ideas, a Shape Canvas for the strongest one, a complete business profile with competition and revenue scenarios, and a clear recommendation on whether to pursue it.
Why I Built It As a Prompt Workflow, Not a Worksheet
Because you and I both know you're not going to print a worksheet and fill it out with a pen on Saturday morning.
But you'll spend an hour in a conversation with your agent while you make coffee. That's just how we work in 2026. The workflow meets you where you already are: in a Claude or ChatGPT session.
It's also why the file is shipped as a .md. You can paste it into any agent context, fork it, adapt it, run your own version. It's not thought as a document to read but as a context to delegate to your agent.
The Anti-Hallucination Discipline
One design choice worth calling out: the Phase 5 prompt explicitly instructs the agent to say "I don't know" rather than invent numbers. If you ask an AI to size a market without data, it will confidently fabricate a TAM. I've made that mistake. So the workflow forces the agent to state its assumptions, flag the gaps, and recommend where you should go to find the real numbers yourself.
The output shouldn’t be a polished deck. It's a clear-eyed business case with honest gaps, which is exactly what you need to make a decision about whether to invest your precious sovereign hours.
→ Download the Workflow
The full workflow, all 5 phases, scoring rubrics, Shape Canvas, and Business Profile prompts is free. No email gate beyond what you've already given me.
→ Download the Sovereign Idea Workflow (free, .md format — paste directly into any AI assistant)
Spend one Saturday morning running it. I'd genuinely love to hear what your top-ranked idea turns out to be, and what the Business Profile says about it. Reply to this email with your Shape Canvas and Business Profile outputs. I'll take some time to give you honest feedback.
Why I'm Sharing This for Free
Same reason as last week's SEO playbook.
Ideas are the cheapest part of building a product. If I can help you generate better-scoped, higher-leverage, commercially profiled ideas faster, the real work, validating and shipping is still yours to do.
Also: this is the Shape Step of the Building Loop, delivered as an agent-executable workflow. If the signal is good, I'll ship the Validate and Ship workflows in the same format over the coming weeks. The whole loop, free, as prompt sequences you can run end-to-end.
The One Thing
Your domain is not your cage. It's your catalog. Every frustration you've accepted, every process everyone works around, every insight that's "obvious" to you, these are the ideas worth building. Generic ideas are commodity. Domain-native ideas are moat. And an idea without a business profile is still just a hope.
Stop searching for what to build. Start mining what you already know, and profiling what it would take to turn it into something real.
What I'm Building This Week
Shipping the Sovereign Idea Workflow resource page on thesovereigntechnologist.com. Monitoring which phases people actually run (via download data and reply volume) to refine the prompts for version 1.1. And beginning the design of the Validate workflow, the 48-hour demand test, as an agent-executable companion to this one.
What resonated? What did I get wrong? Hit reply: I read everything and I'm building this with you and with your input.
P.S. The toolkit is growing: Visibility Audit (where you stand) → Business Model Canvas (is your idea viable) → Building Loop (how to execute) → SEO Strategy Playbook (make your work findable) → Sovereign Idea Workflow (find the idea worth building, and know what you're getting into). All free, all designed for builders with limited time.
P.P.S. Know a senior engineer sitting on three half-formed ideas and no clear path to pick one? This workflow exists for them. Forward this issue. They can subscribe at thesovereigntechnologist.com.
That’s all for this week.
See you next Thursday.
