DEV Community: Yulin Cheng

Why I Spent Six Months Designing an AI Morning Briefing

Yulin Cheng — Sat, 27 Jun 2026 17:31:51 +0000

Why I Designed an AI Briefing Instead of a Dashboard

The default output of an AI-powered product is a dashboard. AI reads your data, AI summarizes your data, AI displays the summary in a clean UI. You now have a "smart dashboard."

I didn't want to build a smart dashboard.

Here's why.

The Dashboard Problem

A dashboard tells you what's there. It organizes information into categories, applies some intelligence to the presentation, and puts it in front of you. The burden of interpretation is still yours.

This is fine for data you're actively analyzing. It's terrible for information you need to act on.

The problem with dashboards in productivity tools is that they add a layer between you and the decision you need to make. You open the dashboard, you scan the widgets, you figure out what matters, you close the dashboard, then you do the thing. The AI didn't save you a step — it added one.

What a Briefing Does Differently

A briefing skips the scan. It starts from the question: what does this person need to know and do, right now, today?

That's a fundamentally different output. Not "here's all your data organized" — but "here's what your day is about, and here's what to focus on."

The difference sounds subtle. It isn't. One is an information tool. One is a decision tool.

The Design Challenge

The hard part of building a briefing — rather than a dashboard — is that you have to make editorial decisions. A dashboard shows everything and lets you choose. A briefing makes a choice for you.

Making choices for the user is scary. Get it wrong and you've made their day worse. Get it right and you've given them back time they didn't know they were losing.

The tradeoff is worth it. Here's why: the alternative is a dashboard that the user has to interpret. And interpreting a dashboard is work. Work that a good AI should be doing for them.

What "Briefing" Actually Means in First Light

Every morning, First Light reads:

Your open tasks and their urgency signals
Your calendar for the day ahead
Recent email that might indicate pending actions
Patterns from recent days (what keeps getting deferred, what's been on the list too long)

Then it produces three things:

Three prioritized actions — not your whole list, just the three that matter most today, and why.

One insight — something about your current situation that you might not have surfaced yet. A pattern, a risk, an opportunity.

One thing to drop — because the enemy of focus is the task that doesn't need to be done, the meeting that could have been an email, the promise you made that you don't need to keep.

That's a briefing. Not a list of everything. A clear signal about what today requires.

The Philosophical Point

We have a tendency to think more information leads to better decisions. It doesn't. More information leads to analysis paralysis, not better decisions.

What leads to better decisions is the right information, structured for action, delivered at the right time.

That's what a briefing is. And that's why First Light is a briefing, not a dashboard.

Try First Light at firstlight.to

I Built a Bilingual Daily Planner — Here's What I Learned About AI and Language

Yulin Cheng — Sat, 27 Jun 2026 17:23:09 +0000

Building a Bilingual Daily Planner: What I Learned

Building a productivity app in English and Traditional Chinese at the same time sounds like a localization challenge. It is. But it's also something else: a forcing function for clarity.

Let me explain.

The Language Problem Is a Thinking Problem

When you have to explain the same concept in two languages with very different grammatical structures, you discover something uncomfortable: you didn't understand the concept as well as you thought you did.

English lets you be vague. "Get things done" works as a tagline because English tolerates ambiguity. Traditional Chinese doesn't. The language demands specificity. You can't say "get things done" — you have to say what things, for whom, by when.

This sounds like a constraint. It is. But constraints are useful. The vagueness that English lets you get away with is often the same vagueness that makes productivity tools less effective than they could be.

What Bilingual Taught Me About UX

Most productivity apps are designed in English and then "localized" — which usually means translated. The menus get translated. The onboarding gets translated. The error messages get translated.

That's not localization. That's translation.

Real localization starts from a different question: what does a user in Taipei need from this product, on a Tuesday morning, when they're already behind? That answer is not the same as the answer for a user in San Francisco. The context is different. The mental models are different. The relationship to "productivity" as a concept is different.

Building First Light bilingual meant asking those questions early, rather than bolting on a Chinese translation at the end. The result is a product that works naturally in both languages — not a product that works in English with Chinese subtitles.

Three Things I'd Do Differently

1. Start bilingual from day one

Localization at the end is expensive and painful. Localization from the beginning is just design. If you're building for more than one market, make the bilingual decision on day one, not day ninety.

2. Hire for cultural clarity, not just translation

The best Chinese copy for a productivity app is not written by a translator. It's written by someone who understands the productivity culture in that market — what people expect from a morning briefing, what "clarity" means in that context, what a "task" actually is in daily life.

3. Build the English version lean

English has a tendency to bloat. We add features because we can describe them, not because they're useful. Building the Chinese version forces you to ship only what you can explain clearly in both languages — which is usually exactly the right feature set.

Why This Matters for First Light

First Light's Daily Edition ships in two languages simultaneously. Not because it's a nice-to-have — because the clarity that Chinese demands makes the English version better.

If you're building for more than one market, don't think of it as extra work. Think of it as a quality filter.

First Light is live at firstlight.to

The Morning Routine Problem No Productivity App Wants to Solve

Yulin Cheng — Sat, 27 Jun 2026 17:23:05 +0000

The Morning Routine Problem Nobody Talks About

You've read the articles. You've seen the Twitter threads. Wake up at 5am. Cold shower. Journal. Meditation. Exercise. Blue light exposure. No screens for the first hour.

The people selling you morning routines have figured something out: the morning is when your willpower is highest, your context is cleanest, and your capacity for intentional action is at its peak.

What's weird is that the productivity industry talks constantly about having a morning routine, and almost never about what you're actually supposed to do in it.

The Routine vs. The Briefing

A morning routine is a set of habits. A morning briefing is a decision-making process. They're not the same thing, and confusing them is where most people lose the plot.

Here's the test: when you finish your morning routine — when you've done your journaling, your meditation, your cold plunge, your green smoothie — do you know more clearly what today is about? Or do you just feel like you've done something healthy?

If it's the second, your routine is a ritual, not a tool. Rituals have value. But they're not the same as clarity.

What You're Actually Looking For

The thing people want from a morning routine — the thing they think they're getting from the cold showers and the 5am wake-ups — is a felt sense that today is manageable.

Not a list of everything on your plate. Not a count of your unfinished tasks. A sense that you have a handle on what matters today, and a reasonable plan for handling it.

That's not a routine. That's a briefing.

Why This Is Hard to Engineer

Here's the problem: your brain is not a reliable source of truth about what matters today. It's a great source of anxiety about what might matter, what could go wrong, and what you shouldn't have left unfinished.

When you sit down in the morning and try to figure out what to do, you're starting from a biased dataset. Your brain is going to surface the thing that caused you the most stress yesterday. It's not going to surface the thing that will matter most tomorrow.

Getting to real clarity — the kind that makes the rest of the day feel manageable — requires an external view. Something that sees all your tasks, your calendar, your commitments, and tells you, from a position of actual context: here's what today is, and here's what to focus on.

The Gap First Light Fills

First Light generates a daily briefing. It reads your tasks, your calendar, your recent email, and produces:

Three prioritized actions for today
One key insight about your current situation
One thing to drop or defer

It's not a task list. It's a decision-making tool for the first fifteen minutes of your day — the fifteen minutes that determine how the other nine hours go.

The morning routine industry wants you to wake up earlier. First Light wants you to wake up with more clarity. Different goals. Different tools.

Try the Daily Edition at firstlight.to

What If Your AI Actually Did Things — Not Just Reminded You?

Yulin Cheng — Sat, 27 Jun 2026 17:17:33 +0000

What If Your AI Didn't Just Read Your Tasks — It Acted on Them?

Most AI integrations with task apps look like this: you ask a question, the AI answers. You copy the answer somewhere. You act on it. The AI never touches your task list.

This is fine if you want a sophisticated search engine. It's not fine if you actually want AI to do work on your behalf.

There's a difference between an AI that answers questions about your tasks and an AI that acts on your tasks. First Light is built around the second thing.

The MCP Difference

First Light is built on a model context protocol (MCP) architecture. That sounds technical, but the practical meaning is simple: the AI has direct access to your task list, your calendar, and your inbox — not as read-only data, but as a live system it can write to.

What does that actually mean in practice?

Instead of: "AI, what should I work on today?" → AI gives you advice → You manually update your task list → You do the work

With First Light: AI reads your tasks and calendar → AI surfaces what matters today → AI marks tasks complete when they're done → AI creates follow-up tasks automatically

The loop closes. The AI isn't just in the conversation — it's in the system.

Three Things MCP-Native Changes

1. Your AI knows what's actually on your plate

Most AI assistants are stateless. Every conversation starts from scratch. MCP-native means the AI knows your open tasks, your deadlines, your calendar constraints — and factors all of that into every response. It's not guessing what you have going on. It knows.

2. The AI can complete tasks on your behalf

When you finish something — send an email, close a deal, ship a feature — First Light marks it complete. No manual updating. No "I'll do it later" task hygiene that never happens. The system stays current because the AI keeps it that way.

3. Follow-ups happen automatically

You don't have to remember to follow up on something. First Light creates the task, sets the due date based on the context of the conversation, and tags it appropriately. The act of deciding to follow up is the last manual step.

Why This Matters More Than a Better UI

Every new productivity app promises a better interface. Cleaner design. Smarter filters. Better keyboard shortcuts. But underneath the UI, the model is the same: human enters data, human decides what to do, human does the work, human updates the list.

MCP changes the model. The AI becomes part of the workflow — not a consultant you consult, but a collaborator that acts.

That's a fundamentally different product. And it's why this moment in AI-native tools feels different from the last wave of "smart" productivity apps.

First Light is built MCP-first. Try it at firstlight.to

I Stopped Using Task Apps. Here's What I Use Instead.

Yulin Cheng — Sat, 27 Jun 2026 17:17:30 +0000

I Stopped Using Task Apps. Here's What I Use Instead.

Fourteen years of to-do apps, and I kept ending up in the same place: a longer list than when I started.

I want to tell you about the moment something shifted — not with a new app, but with a different way of thinking about what a task app should actually do.

The Problem With Task Apps

Here's what almost every task app does: it gives you a place to store things you need to do. That's it. You dump your brain into a list, and then you're responsible for deciding what to do with that list. Every day. Forever.

And the result?

A study from the University of California found that the average knowledge worker has between 150 and 300 tasks floating around their head at any given time. We built apps to handle that. But the apps just moved the pile from your head to a screen. The cognitive load didn't go away — it just changed address.

The real problem isn't storage. The real problem is decision fatigue. Your brain is not a good place to store tasks. It's a terrible place. It's an even worse place to prioritize tasks, because it keeps resurfacing the urgent ones while burying the ones that actually matter.

What I Was Actually Doing Wrong

I was using my task app as a storage device when I needed it to be a thinking device.

Every morning I'd open my app, see 47 things, and default to the one that felt most urgent — which was almost never the most important. By noon I'd done six things that felt productive and zero things that moved anything meaningful forward.

The trap: task apps reward completion, not progress.

The Moment I Switched

About six months ago, I started using an AI-native daily planner called First Light. Here's what changed:

Instead of looking at a list of 47 tasks, I got a morning briefing. Three things. One insight. One thing to drop. Written from the context of my actual day — not a generic productivity framework, but something that understood what mattered today.

It sounds small. But here's what happened: I stopped starting my day with a panic scan of everything I'd promised to do. I started it with a clear signal about what this particular day required.

The task list still exists. But it's not what I open in the morning anymore.

Why This Matters More Than You Think

The productivity industry wants you to believe the answer is a better app. Better sorting, better tags, better filters, better views. But the underlying model hasn't changed in twenty years: you enter tasks, you see tasks, you do tasks.

The better model isn't a smarter list. It's an AI that thinks on your behalf — one that reads your calendar, your tasks, your context, and tells you what to do today and why.

That's what First Light is doing. It's not another task app. It's a daily briefing for people who have too much to do and not enough clarity about what actually matters.

What I'd Tell Myself Six Months Ago

Stop optimizing your task list. Start optimizing your morning clarity.

The task app will always be there. What you need is something that helps you decide — before 9am, before the Slack messages start, before the day runs away from you — what today is actually about.

First Light: firstlight.to