How we started Yuki lean

Yuki started with a simple rule: spend money on what people feel, not on what they never see. Here is the full stack, what each tool costs or offers, and where the money actually goes.

Pricing and programs below are as of November 2026. They change often, so check each linked page before you plan around a number.

When people ask what it costs to build an app like Yuki, they usually expect the answer to be servers. It isn’t. The infrastructure behind a modern product can cost very little in the early months. The bill that grows is the one for the tools that help you build faster.

This is how we set Yuki up: what we use, what is free, what we pay for, and what we would tell anyone starting today.

The principle: spend on what users feel

Yuki is an AI personal assistant for life admin. It turns the emails you already have into events, trips and bills, and it helps you plan your week. Everything a person notices about Yuki happens in the product: how fast it understands an email, how clear a screen is, whether a plan makes sense.

Nobody notices which crash reporter we use. So the rule was easy to state. Keep the invisible parts on free tiers and usage-based pricing for as long as they hold up, and put the money into design and building.

A second rule followed from the first. Pick tools whose pricing grows with you. Paying when the product earns is far easier than carrying a fixed monthly bill before it does.

The stack, by category

Monitoring and analytics

Sentry, for crash reporting. Sentry for Startups offers up to $5,000 in credits that expire after 12 months, for young companies that are new to paying for Sentry.
PostHog, for product analytics. PostHog’s free tier includes 1 million analytics events and 5,000 session replays every month.

Cloud

Google for Startups Cloud. The Start tier offers up to $2,000 in Google Cloud credits, valid for one year. The larger tiers require venture funding, so most early founders should plan around the Start tier. Accepted members can also get 12 months of Google Workspace Business Plus at no extra cost, for a new signup. That is a separate application, and the company domain must not have had a paid Workspace plan in the 31 days before applying.

Hosting

Cloudflare. Our website, docs and this blog run on Cloudflare’s free plan, which is more than enough for early traffic. If you outgrow it, Cloudflare for Startups offers $10,000 in credits for bootstrapped companies that have raised under $1M, and up to $350,000 for funded ones. We haven’t needed them yet.

Code, builds and releases

GitHub. The free plan includes private repositories and 2,000 GitHub Actions minutes a month.
Expo. The free plan includes 15 iOS and 15 Android builds, which is plenty while a mobile app is young.

Payments and subscriptions

Stripe. No setup or monthly fees, just a cut of each web payment.
RevenueCat. Free until $2,500 in monthly tracked revenue, then 1% of what you track. This is the clearest example of pricing that grows with you.

Database and storage

Supabase. This one we pay for. It is cheap at our stage. If you incorporate through Stripe Atlas, its perks include up to $500 in Supabase credits, according to Stripe.

Launch and marketing

Clueso. The free plan got our launch video made.
Buffer. The free plan covers up to 3 channels and 10 scheduled posts per channel, enough to keep a small posting schedule running.

Where the money actually goes: AI

Our biggest bill, by far, is Claude. We use Claude Design for UX work and Claude Code for building.

That is a deliberate choice, and it is the part of the story where working on my own matters. I build Yuki by myself. There is no design team and no second engineer. The tools that let one person move from an idea to a tested screen in an afternoon are the tools that decide how fast Yuki gets better. That is worth paying for in a way that a bigger database plan is not.

It also changes how we think about cost. A few years ago, the advice was to watch your hosting bill. Today, for a small team, infrastructure is rarely the expensive part. The expensive part is capacity: how much you can design, build and ship in a week. AI is where we buy that capacity.

What we would do the same

Start on free tiers and usage-based pricing. Almost every tool above costs nothing, or close to it, until the product has real usage.
Pay for speed, not for comfort. We spend on the tools that make the product better, not on upgrades that only feel safer.
Keep the stack small. Fewer tools means fewer bills, fewer logins and fewer things that can break on a Friday night.

What we would do differently

Time the startup credits. Credits expire. Sentry’s last 12 months, so apply when you are ready to use them, not the day you get the idea. The same thinking applies to any program with a clock on it.
Read the eligibility rules first. Many credit programs have conditions, such as company age, funding stage or being new to the product. Knowing them early saves applying for perks you cannot use yet.
Track the AI bill from day one. It is the line that grows fastest, so it deserves the most attention.

Closing

Starting lean is less about being cheap and more about being clear on what matters. For Yuki, that means the invisible parts run on generous free tiers, and the money goes into the product people actually use.

If you have found a free tier or startup credit we missed, tell us. We would like to keep this list useful for the next founder.

Try Yuki. Yuki is an AI personal assistant for life admin. Its free “What if” feature plans around your real calendar, then adds the blocks after you approve. Start free at yukihq.com, with a 14-day Pro trial and no card required.