Hasaki
AI Makes Avoiding Decisions Cheaper
You ask AI to compare two options. It lists the advantages and disadvantages of each, but that does not feel like enough…
Writing Prompts in a Non-Native Language
I have noticed that when I write prompts for an Agent in Chinese, I instinctively soften what I say. “You could start by…
Repo as an Agent
A well-designed repo can itself become an Agent that continuously improves. Here, a repo is more than a place to store c…
Do What Is Clearly Right and Easy to Begin
No one can map out the entire path to becoming a professional footballer in advance. Which club you will join, which coa…
Tacit Understanding Cannot Be Expected Up Front
「この件、考えておいて。」 “Give this some thought.” The sentence is easy to understand. The hard part is knowing what to do next. Am…
Learning English Doesn't Have to Start with a Vocabulary App
When people decide to learn English, their first move is often to download a vocabulary app. They pick a word list, set …
Can a Company Without Its Own Product Still Have FDEs?
Lately, I have been thinking about a question: can a company without a product of its own have genuine FDEs? My answer i…
Saturday Is Not the Tail End of the Week
I used to think of Monday as the start of the week. If Monday went well, the rest of the week probably would too. If I f…
Control the Blast Radius
When teams discuss AI coding, their first questions are often still the same: Who reviews the code after an Agent writes…
Some Prompts Should Deliberately Leave Room
Sometimes I give an Agent an article, a conversation, or even a pile of unorganized material and ask just one question: …
Nōmiso: What Remains After Translation Gets It Right
I recently came across a Japanese word while listening to a podcast: 脳味噌（のうみそ）. 「脳」 means brain; 「味噌」 means miso. Put to…
Why Definitions Matter
One of the best ways to learn something is to return to its most basic definition. Do not stop asking questions just bec…
Don't Mistake Confusion for Thinking
When I sit down at my computer in the morning, my mind is still clear. A few emails are waiting, Slack has several red b…
Give Agents Primitives, Not Prepackaged Workflows
When designing a Harness for an Agent, start with a small set of clear primitives and let concrete tools and workflows e…
A Long-running Agent Is Not Always Running
We usually understand “long-running” as a process that remains alive. A database, queue, or web service occupies memory …
What End-to-End Delivery Means
Clients rarely hand you a neatly defined task. They do not say, “Analyze lead quality and the sales funnel, identify why…
Why I Still Chose Haiku at Ten Times the Price
Recently, I needed to choose a model for a context-aware definition feature. When a user selects an English word, the sy…
I Am the Agent's Guide Dog
The Agent is not my guide dog. It is more like a blind person with extraordinary abilities, moving quickly but unable to…
The Difference Between Information and Knowledge
The Same Label Can Hide Different Questions Search X for AI memory, and you will find discussions that seem to concern t…
Don't Rush to Turn the Graph into a Tree While It Is Still Growing
Strings in programming languages hardly seem worth designing at first. Are they not just sequences of characters? Store …
AI Is Already Good Enough—the Hard Part Is Choosing the Right Harness
Not long ago, I was breaking down a HubSpot query request. The goal fit into one sentence: employees should be able to u…
AI-Native Starts with Redesigning the Work
Being AI-native is not about knowing AI jargon, using more tokens, or adding agents to every workflow. It means recogniz…
AI Programming Is Still Waiting for Its Spring Boot
AI programming has moved beyond “Can models write code?” It has not yet moved beyond “How can teams deliver reliably?” P…
The Interface Is Shaping How We Use AI
We open an AI, ready to hand it a difficult task. In the input box, we type, “Help me redesign the homepage,” and hit se…
In the Agent Era, Branches Are No Longer the Default
Branches became the default under particular conditions. They name a line of commit history and gradually took on isolat…
Learn to Adjust the Friction in Learning
When people talk about learning efficiently, they often try to make everything faster. AI has strengthened that impulse.…
Why Workflows and Autonomous Agents Will Coexist
Many people believe that once Autonomous Agents become intelligent enough, Workflows will no longer be necessary. Handin…