Libretto — Latest Insights

Future&Tech

코히어 파스 5, AI 문서 처리에서 벤치마크 79점이 84점을 앞서는 조건

AI 도구를 도입하는 논의에서 벤치마크 점수는 거의 첫 번째 비교 항목이 됩니다. 어느 모델이 어느 테스트에서 몇 점을 받았는지를 먼저 살피고, 그 순위를 근거로 구매 결정을 정당화하는 방식이 일반적입니다. 점수가 높은 쪽이 더 잘 만들어진 모델이고, 더 잘 만들어진 모델을 선택하는 게 합리적이라는 논리는 그 자체로는 틀리지 않습니다.

Mgmt&Insight

Order First, Invest Later: A Lesson From Microsoft's Backlog

When inquiries keep coming in but last month's orders still aren't out the door, most solo business owners have wondered whether to stop taking on new work. Last week, Microsoft offered a clue: the company that used to ask whether AI demand was coming has started asking how fast it can deliver on the demand it already has.

Dev&Data

Getting the Most Out of Claude Code: How to Save Your Mistakes to Memory

Why the moves that matter aren't the commands you ran, but the judgment calls behind them

Every time you open a new Claude Code session, you may find yourself re-explaining your project structure and re-fixing mistakes you already corrected yesterday. The difference between someone who ends up with a sharp, well-trained assistant and someone stuck retraining a new hire every time comes down to one thing: what you choose to save.

Mgmt&Insight

What Café Branding Actually Means: How to Build a Shop Customers Come Back To

Most new owners mistake a logo and an Instagram account for branding. Here's what actually makes a café memorable — and a three-question test to see if yours is.

Even in a saturated café market where new shops keep opening anyway, most first-time owners treat branding as a logo, a sign, and an Instagram account. This piece explains what café branding actually means and offers a checklist for building an identity customers remember and come back for.

Issue&Insight

In the AI Training Data Wars, the Government Played Its Hand First

Naver's crawling lockdown, the DOJ's intervention in the OpenAI lawsuit, and the Astra safety controversy — a record of the day three stories collided

In May 2025, the US Department of Justice filed a brief arguing that AI training on copyrighted works is fair use and vital to American competitiveness — a move that, alongside Naver's crawling restrictions and safety concerns over OpenAI's Astra model, signals a new phase in the battle over who controls AI training data.

Money&Invest

How to Read Your R&D Ratio: Size, Direction, and Capacity

A solo entrepreneur earns steady income from online courses, but pours a large share of it into outsourced app development each month — with no revenue yet to show for it. Without a clear standard for telling a growth investment from money down the drain, they hesitate every month in front of the transfer button.

Mgmt&Insight

오류가 보이지 않는 작업에서 사람이 확인합니다

어떤 단계를 자동화해도 되는지는 출력 정확도가 아니라 책임의 귀속 방식으로 정해집니다. 이 기준을 모르면 사람을 잘못된 자리에 배치하게 됩니다.

번역 에이전트를 납품하고 계약 갱신을 논의하러 간 자리에서 종료 통보를 받은 사례가 있습니다. 오역이 많았다면 이해가 됩니다. 실제 이유는 달랐습니다. 검수 이력에 사람이 읽었다는 흔적이 없었습니다. 발주처 담당자는 계약서를 꺼내기 전에 이것 하나를 확인했습니다. 이 텍스트를 누군가 직접 읽었습니까.

Dev&Data

FPV Drone Parts Compatibility: Why Picking the Best of Everything Still Fails

Motors, ESCs, and batteries aren't chosen one by one — they're chosen as a chain

There's a common way first-time builders pick parts: search for a well-reviewed motor, do the same for the ESC, then the battery. Each part is good on its own — but that's not enough.

Money&Invest

When Claude Gets Cheaper, Agents Start Grading Their Own Work

As AI tool costs fall fast, who's building the filter to catch the errors quietly piling up inside?

In May 2025, Anthropic cut Claude API pricing again, and developers celebrated near-free usage at scale — the same week reports surfaced that Anthropic was facing multiple copyright lawsuits in the US.

Career&Upskill

What to Hand Off to Claude Cowork, and What to Keep

AI used to live in a chat window, answering questions you typed in — now desktop agents like Claude Cowork can open folders and edit documents and spreadsheets directly. This piece offers a single principle for deciding what to delegate and what to keep: AI takes over tasks, not entire jobs, so break your job into individual tasks before deciding what goes to Claude Cowork.

Mgmt&Insight

A Seasonal Menu Strategy to Survive Your Café's Three-Month Slump

A café that's packed with strawberry drinks and desserts from winter through spring goes quiet the moment the season ends, with foot traffic and sales dropping sharply and owners scrambling for the next seasonal menu every year. This piece lays out a café seasonal menu strategy for bridging that three-month gap.

Mgmt&Insight

AI Meeting Notes Are Full of Errors — Does 'A Human Checks It Last' Really Hold?

You skim an AI-written meeting summary for smooth prose, and if it reads clean, you paste it into Notion without a second look. A new audit of AI medical scribes finds that this kind of skim-review catches almost none of the errors that matter most — the fabricated ones, which happen to read the smoothest.

Issue&Insight

The Day Sony and Warner Sent Anthropic the Bill, the Whole AI Industry Flinched

A copyright lawsuit over training data has, for the first time, put a major music publisher in the plaintiff's seat

In May 2025, Anthropic was the subject of two contradictory headlines within a single day. One day it walked away clean from a retaliation suit tied to the U.S. Department of Defense; the next, Sony Music and Warner Chappell Music filed suit alleging copyright infringement.

Future&Tech

The Numbers Say OpenAI's Chip Beat Nvidia. Should We Believe Them?

How one line in a SemiAnalysis report shook the AI infrastructure landscape — and what it actually takes to read a benchmark

In May 2025, semiconductor analysis outlet SemiAnalysis quietly published a report claiming that OpenAI's custom chip, "Jalapeño," had outperformed Nvidia's latest Blackwell architecture on inference throughput and power efficiency.

Mgmt&Insight

What Apple's App Store Crackdown Reveals About Platform Fee Dependency

If most of your revenue comes from someone else's platform — a marketplace, an app store, YouTube — your profits move whenever that platform's fee rules do. Apple just showed, at the largest scale imaginable, what happens when those rules turn against you.

Future&Tech

The Day Z.ai Shipped Two Models and Bill Gates Wrote 6,000 Words

What a single day of speed colliding with caution asks of solo entrepreneurs

On one Monday morning in May 2025, the Chinese AI lab Z.ai released two models within hours of Bill Gates publishing a 6,000-word essay voicing concern about AI. The two stories, landing in the same feed almost at once, pointed in opposite directions — and together they offer a useful gauge of where solo entrepreneurs stand right now.

Dev&Data

Does Making AI Agents Collaborate Actually Help?

If you've ever had multiple AI models review and critique each other's work to get a better result, a new paper suggests that approach may backfire — a single round of cross-talk can erase the very diversity that made using multiple models worthwhile in the first place.

Mgmt&Insight

Great Café Service Isn't About Friendliness — It's About Emotional Design

Sleek interiors and friendly staff are now standard at coffee shops, yet only some places earn repeat visits while others see customers just once. The difference comes down to deliberately designing the emotional journey customers move through from the moment they walk in to the moment they leave.

Money&Invest

FPV Drones and Pilot Licenses: What Class Does Your Build Fall Into?

Weight gets decided first, licensing follows — and since you're building it yourself, you're the one who has to do the math.

Search for drone licensing rules and you'll find a table running from Class 1 to Class 4, based on maximum takeoff weight: under 250 grams needs no license, above that requires online training, and above 2 kilograms requires a written exam and logged flight experience.

Mgmt&Insight

대기업 파트너십에서 스타트업이 공유한 정보는 인수 협상의 근거가 됩니다

협력 논의와 M&A 실사가 같은 서류로 시작되는 구조

티오더와 KT의 관계는 2022년 말 협력 논의로 시작됐습니다. 그로부터 M&A 실사 단계까지 갔다가 합의에 이르지 못하고, 지금은 분쟁으로 이어졌습니다. 이 흐름에서 주목할 대목은 협력 제안과 인수 실사가 스타트업에 요구하는 정보가 거의 동일하다는 점입니다.

Mgmt&Insight

Why 622 B2B Negotiators Stopped Fact-Checking When a Story Walked In

An experiment with 622 procurement professionals reveals how factual judgment shifts once information is wrapped in narrative

A study of 622 B2B sales and procurement professionals found that when identical information was presented as a story rather than as a flat statement, participants became significantly worse at telling fact from fiction — regardless of how much negotiation experience they had.

Issue&Insight

The Government Blacklisted an AI Company for Speaking Up. A Court Reversed It.

What the Anthropic ruling quietly means for how solo operators choose their tools and structure their contracts

In May 2025, a U.S. federal court ruled that barring Anthropic from government procurement over its public stance on AI safety violated the First Amendment — a decision handed down the same day OpenAI claimed to have reached AGI and a self-improving AI system was shown outside the lab for the first time. Together, the three events point to a shift in how much any business can safely depend on a single AI vendor.

Mgmt&Insight

도구를 먼저 산 팀과 업무를 먼저 쪼갠 팀

같은 AI 플랫폼을 쓰는 두 팀이 석 달 뒤 다른 결과를 내는 이유

AI 도입 프로젝트를 진행한 팀들 사이에는 공통된 출발점이 있습니다. 경영진의 지시가 내려오고, 벤더 데모를 참관하고, 구독 계약을 맺습니다. 그런데 6개월이 지난 뒤의 모습은 갈립니다. 어떤 팀은 AI가 실무 안에 녹아들어 있고, 어떤 팀은 여전히 "어떻게 잘 쓸 수 있는지 방법을 찾고 있는 중입니다"라고 말합니다.

Career&Upskill

Remote Workers Reported the Highest Well-Being in a 7,700-Person Study — So Why Are Companies Still Calling Everyone Back?

What a University of Colorado study found, and the distance between that data and actual workplace policy

In August 2026, a University of Colorado research team compared well-being across 7,700 U.S. employees working remotely, hybrid, and in-office. The result pointed one way: fully remote workers scored highest of the three groups — yet that same year, major employers kept tightening their return-to-office mandates.