1. Hearing “MCP” and asking “So… can it read the posts directly?”
A friend says D-Lab has launched MCP, making it easier to connect with assistants such as ChatGPT. Instead of asking for the definition, the first question is: “What can it do? Can it read the posts directly?” Before learning the acronym, the person is already guessing the workflow: can AI fetch the information from the service instead of being manually fed context? That is close to D-Lab’s actual description of MCP: direct D-Lab knowledge search from assistants such as ChatGPT and Claude.
2. The real skill is not knowing everything; it is guessing the right function quickly
The mental chain is: external service connects to AI → maybe AI can access the content → maybe copy-paste disappears → maybe it can read the posts directly. This is operational reasoning. Some people meet new software and read documentation from page one. Others ask, “Which manual step can this remove?” That second habit is powerful in automation and process improvement.
3. GitHub stops being “where websites live” and becomes an AI factory warehouse
When someone says “GitHub is where you can make websites, right?”, a lecture on distributed version control would be accurate and socially fatal. A better explanation is “a very convenient data warehouse.” Then the workflow becomes: store site data → AI reads it → automated checks → fixes when needed → updates → scheduled repetition. ChatGPT becomes a component inside an operating system for work. Technical note: OpenAI’s official ChatGPT GitHub app is mainly read-only; writing and pushing use write-capable workflows such as Codex.
4. “Make it fetch the evidence” turns AI into a research team
For health, psychology, learning, and cognition, AI can be instructed to move beyond secondary summaries and inspect PubMed, PMC, publisher pages, original studies, and reviews. It becomes search assistant + paper triage + summarizer + comparator + skeptic. New studies can sometimes be surfaced before someone summarizes them in another language. PMC, however, is not an original-research-only vault; it also includes author manuscripts and preprints, so study design, peer review, sample size, and limitations still matter.
5. What is D-Lab MCP actually selling?
As of August 30, 2026, the DaiGo plan is ¥2,199/month and D-Lab Pro is ¥989/month, with DaiGo required for Pro. MCP is a Pro feature, making the simple total ¥3,188/month. But this is not a paper-access fee. The product is: DaiGo selects information → explains it → turns it into action → accumulates it in more than 4,000 videos and archives → makes that curated knowledge searchable through AI.
6. Then comes the forbidden thought: “Isn’t ChatGPT Plus enough?”
ChatGPT Plus costs $20/month and includes broader model/tool access, file analysis, and Deep Research. For someone who can search primary sources, the D-Lab route looks like research → DaiGo → D-Lab → MCP → ChatGPT → me, while the direct route is web/paper databases → ChatGPT → me. Several layers disappear. The products are not identical, but if the desired result is an evidence-backed answer rather than DaiGo’s explanation, the direct route can substitute for a lot of the utility.
7. D-Lab makes more sense as “DaiGo fan-club spending”
Two people can read the same paper and create completely different experiences depending on what they select, their examples, pacing, humor, and how they turn evidence into action. Information value and narrator value are different. D-Lab can therefore be seen as an explanation fee, curation fee, editing fee, and—said affectionately—a DaiGo fan-club membership. Nobody tells a concertgoer, “Why pay? The song is on streaming.” The experience is the product.
8. The middleman-margin sensor is permanently set to maximum
Some people automatically ask: “What value does this layer add? Can I go straight to the source? What labor am I paying to remove? If I can do it in five minutes, why subscribe?” That is not merely stinginess; it is transaction-cost analysis. The better someone is at search, AI, and workflows, the less valuable some intermediaries become. But “I can do it myself” is not the same as “I should.” If a small fee deletes one hour of annoying work, it buys time. The alarm should scream only when almost no time is saved.
9. DaiGo is still impressive — especially “2× speed without pressing 2×”
Even after dismantling the subscription logic, mentalism, card reveals, attention control, timing, and narrative construction remain impressive. Those skills do not come from downloading the original paper. And there is a remarkable UX feature: you do not need 2× playback speed because he already talks fast. If the speaker’s biological hardware feels like 1.7× by default, the speed button becomes decorative. Intentional optimization? Probably not. Probably just fast speech. But for people who like dense information at high velocity, it is part of the appeal.
10. The annoying part: a paywall right before the core answer — and what remains valuable in the AI era
A free video that raises the problem and curiosity, then moves to paid content exactly at “So what should I do?”, can frustrate. Paid depth is fine; a satisfying split is free = answer the title-level question, paid = deeper evidence, additional studies, edge cases, and implementation. When the core answer itself sits behind the wall, research-oriented viewers may simply find the original paper. As MCP shortens the distance between AI and information sources, merely possessing information becomes less valuable. What remains valuable is selection, interpretation, explanation, and action. For evidence retrieval, some users can replace a lot with ChatGPT. For DaiGo’s explanation style, no. And if 1× already feeling like 2× is part of the charm, that is full fandom. In the AI era, people may ultimately pay not for data but for “I want to hear it from this person.”
