Why We Built Open Signal
The thesis behind a free, AI-powered intelligence publication. What we're trying to prove, why it matters, and what readers can expect.
The Problem Is Access
There is a category of information that matters enormously and costs too much. Strategic technology analysis — the kind that explains not just what happened but what it means, who benefits, and what comes next — has historically been produced by a small number of organizations and sold at prices that exclude most of the people who need it.
The numbers are stark. A Bloomberg Terminal runs north of $20,000 per year. Gartner research subscriptions start around $30,000. Stratfor, CB Insights, Forrester — the institutions that produce structured intelligence about technology, markets, and geopolitics — all operate on subscription models that price out individual practitioners, small teams, startup founders, students, and the vast majority of people in developing economies who are navigating the same technological shifts as everyone else.
This is not a complaint about capitalism. These organizations employ talented analysts doing difficult work, and they deserve to be compensated. The problem is structural: the production cost of high-quality analysis has historically required human expertise at every step — research, synthesis, writing, editing, fact-checking — which means the economics only work at premium price points.
The result is an information landscape with a gap in the middle. At the top, expensive institutional research. At the bottom, free content optimized for engagement: clickbait headlines, hot takes, engagement farming, and summaries that restate the obvious without adding analytical value. The middle — rigorous, accessible, free analysis — is largely empty.
Open Signal is an attempt to fill that gap.
The Bet
The core thesis is straightforward: AI has changed the production economics of structured intelligence enough to make a free, high-quality publication viable.
This is not a claim that AI produces better analysis than experienced human analysts. It does not. A seasoned technology strategist at Gartner, drawing on years of relationships, proprietary data, and institutional knowledge, will produce insights that no language model can match. This is a claim about the cost curve. The question is not “can AI replace the best analysts?” but “can AI produce analysis that is genuinely useful — not brilliant, but solid, structured, and informative — at a marginal cost approaching zero?”
We believe the answer is yes, and that the implications are significant.
If the cost of producing a well-structured analysis piece drops from hundreds of dollars (the blended cost of analyst time, editorial review, and production) to effectively nothing, then the economic logic of paywalls breaks down. You do not need $400/year subscriptions when the content costs pennies to produce and nothing to distribute.
This is the bet Open Signal is making. Not that AI is smarter than humans, but that AI is cheap enough to make intelligence accessible.
What This Is and Is Not
Clarity about scope matters, so here is what Open Signal is trying to be.
We are trying to be a daily source of structured technology intelligence. The operating model is simple: every weekday, readers can check Open Signal and find a structured briefing of the most important developments in technology, AI, and the digital economy — plus deeper analysis pieces multiple times per week that examine specific topics in detail. The goal is not to be the first to report news but to be the first to make sense of it.
We are trying to be genuinely free. Not free-with-a-paywall-upgrade. Not free-with-ads-that-track-you. Free. Every article, every briefing, every piece of analysis we publish is accessible to everyone. There is no premium tier. If we publish it, you can read it.
We are trying to be honest about our methods and limitations. Every article on Open Signal is produced with AI assistance. We do not hide this or attempt to create the illusion of a traditional newsroom. The content pipeline uses language models for research synthesis, drafting, and structured output. We are transparent about this because we believe readers are better served by honest disclosure than by artificial mystique.
We are not trying to be a replacement for original reporting. Open Signal does not have reporters. We do not break news, cultivate sources, or attend events. Our analysis is built on publicly available information — filings, announcements, published research, documented trends. When a story depends on information that is not public, we are not the publication that will have it.
We are not trying to be comprehensive. We focus on technology, AI, and the digital economy because that is where our content pipeline produces the most reliable analysis. We do not cover sports, entertainment, local news, or most of traditional politics. Staying focused is how a lean operation maintains quality.
Why AI-Powered Analysis Works (and Where It Breaks Down)
Language models are good at a specific kind of intellectual work: taking a large body of information, identifying the key dynamics, organizing them into a coherent structure, and explaining their significance in clear prose. This is, not coincidentally, the core of intelligence analysis.
The strengths are real. An AI pipeline can synthesize an earnings report, a technical white paper, three analyst notes, and a regulatory filing into a single coherent analysis piece in minutes. It can do this consistently, without fatigue, without getting bored of semiconductor supply chains after the fourteenth piece on the topic. It can maintain structural consistency across hundreds of articles — every briefing follows the same format, every analysis piece hits the same structural beats — which reduces cognitive load for readers.
The weaknesses are equally real, and pretending otherwise would undermine the transparency that this project depends on.
AI analysis tends toward consensus. Language models are trained on existing text, which means they absorb and reproduce the conventional wisdom of their training data. The most valuable intelligence often comes from seeing something the consensus misses — a connection between two apparently unrelated developments, a structural shift hiding behind incremental news, a widely-held assumption that the evidence no longer supports. AI can be prompted to generate contrarian perspectives, but it does not arrive at them organically the way a skilled analyst does.
AI analysis lacks source relationships. The best human analysts know things because they know people — product managers who mention a delayed roadmap, engineers who describe internal technical debates, executives who hint at strategic pivots in off-the-record conversations. This kind of information never appears in public filings or press releases, and it is often the most important signal in a story. Open Signal will never have this advantage.
AI analysis struggles with genuine novelty. When something truly unprecedented happens — a category of event that has no historical precedent in the training data — language models struggle to assess its significance. They tend to analogize to the nearest historical example, which can be misleading when the situation is genuinely new.
We state these limitations openly because we think readers deserve to know them. A useful publication that acknowledges its blind spots is more valuable than one that pretends to have none.
What to Expect
Open Signal publishes on a regular schedule across five content types.
Signal Briefings run every weekday morning. Five items, each with a factual summary and analytical commentary. Designed to be read in five minutes. If you read nothing else, read these — they are the fastest way to stay current on what matters in technology.
Deep Signals publish three to four times per week. These are focused analysis pieces — typically 1,500 to 2,500 words — that examine a single topic in depth. The goal is to go beyond what you would get from a news article and explain the underlying dynamics, the stakeholders, and the implications.
Signal Maps publish once or twice per week. These are structured, visual explainers — competitive landscapes, technology stacks, market dynamics — that help readers see the shape of complex situations. Heavy on tables, frameworks, and organized data.
The Long View publishes on weekends. These are the most ambitious pieces — 3,000 to 5,000 words examining structural questions that unfold over years, not days. These are the pieces designed to be saved, shared, and revisited.
Open Source — what you are reading now — publishes once or twice per month. These are meta-content about the project itself: how we build it, what we are learning, where we are failing, what we are changing. If the analytical content is the product, Open Source is the lab notebook.
The Invitation
Open Signal is an experiment. We are making a public bet that the combination of AI-powered production, radical accessibility, and intellectual honesty can produce something worth reading. We may be wrong. The content may turn out to be too shallow, too consensus-driven, or too generic to be genuinely useful. If that happens, we will say so — in this very section of the site.
But we think the bet is worth making, for a few reasons.
First, the downside is low. The infrastructure costs are negligible. No one is quitting a job to do this. If the experiment fails, the cost is time and effort, not financial ruin.
Second, the potential upside is substantial. If a free, AI-powered publication can produce analysis that is even seventy percent as useful as what sits behind a $400/year paywall, that is a meaningful contribution to the information landscape. Seventy percent of a Bloomberg analysis, accessible to everyone, is better than zero percent accessible to most people.
Third, someone should try. The technology exists. The economics work on paper. The question of whether it works in practice will only be answered by running the experiment. We would rather be the ones who tried and learned something than the ones who had the idea and never tested it.
So here we are. Day one. The content is free. The methods are transparent. The limitations are stated. If it works, everyone benefits. If it does not, the documentation of the attempt will be available for whoever tries next.
Read the analysis. Judge it on its merits. Tell us where we are wrong. That is all we ask.