Nazar O (@ohdanskyi) header
Nazar O (@ohdanskyi) profile picture

18, Da Nang. I build bots that chase client documents, bookings and reminders for businesses. 365 days: 0 → $3k/mo.

Da Nang, Vietnam
Joined December 2024
Photo by Nazar O (@ohdanskyi)Photo by Nazar O (@ohdanskyi)Photo by Nazar O (@ohdanskyi)Photo by Nazar O (@ohdanskyi)Photo by Nazar O (@ohdanskyi)Photo by Nazar O (@ohdanskyi)Photo by Nazar O (@ohdanskyi)Photo by Nazar O (@ohdanskyi)Photo by Nazar O (@ohdanskyi)Photo by Nazar O (@ohdanskyi)
New chapter on this account. I'm Nazar, 18, living in Da Nang. I build bots that chase client documents, bookings and reminders for businesses, so the owner stops doing it by hand. Day 27 of 365: from 0 direct clients to a steady $3k a month. In public.
3
6
259
day 34/365. 18, da nang, building bots for businesses. goal: $3k/mo. yesterday i said: don’t pitch, just ask who decides. a clinic took me for a patient. a bare question with no reason confuses people. new order: why i’m writing, one question, then what the bot does. 32 first messages 21 follow-ups 10 linkedin invites 6 no’s, 2 passed to management 0 calls evening: first offers to managers’ personal contacts. results in a week.
1
1
14
the fix is small: one line saying why i’m writing goes before the question. the bot description comes last. i’ll post what happens in a week.
6
I spent 10 minutes troubleshooting my laptop’s Wi-Fi… before realizing the entire building had no electricity. I woke up, opened my laptop, and started my usual claude based workday. No internet. First thought: “It’s my laptop.” So I updated drivers, toggled Wi-Fi, restarted everything. Then I checked my phone. No signal either. I messaged my landlady. Two minutes later, I understood: The whole building had no electricity. I had mobile data, but only about an hour of laptop battery. So I went hunting for a café. First café: no electricity. Second café: finally electricity Also, guess how much this breakfast + coffee cost me. Back in Europe, I’d expect to pay around €15. The lesson? Things won’t always go according to plan. But the real skill is adapting fast and keeping moving. Have a productive day anyway.
1
1
53
i agree with half of this any new dependency needs a reason first, and my own tooling is plain stdlib python on purpose but compression, crypto and auth i’d never rewrite with a prompt. a known library has thousands of people finding its bugs. my copy would have one
What’s going to happen to open source? A prediction: Dependencies are over. The same Brotli vulnerability impacted .NET, Node, Bun, Envoy and more. Dependencies are too big of a target. When rewriting a dependency is one prompt away, why take the risk?
1
2
46
i started building at 15, before AI could write a working function 3 years later AI does more of the build every few months. the part that stays mine: i define the problem, the logic and the testing, and i’m the one the client calls when it breaks
Working with AI requires perpetually operating at the edge of sanity. You have to recalibrate your mental model of AI capabilities and your own job description every few months, without getting delusional about seemingly infinite productivity gains, or feeling threatened by losing what you thought was the uniquely human part of your job, or obsessing over the perfect setup, or losing your mind over the clumsiness and jaggedness of agents, or ... Maybe AI literacy needs to be less about prompting and more about improving emotional regulation.
1
3
78
code readability has a reader nobody mentions: the client on a document bot i built for an accounting firm, everything they might change, the client list and the deadlines, lives in a google sheet they already use the agent reads the code. the owner reads the sheet. nobody has to read both
We need to reconsider our priors on code readability. Humans, compared to agents, are pretty slow at understanding large quantities of text. So in the old days, readability was important so the human could efficiently grok what was going on. So readability now comes down to: 1. Can a human, if needed, jump in and understand the code 2. Does the code use the same domain language as the human describing the code? 1 helps with observability and the occasional spot checks you need to do on important PR's. 2 helps the agent translate the words you use into code changes. In other words, readability is now a nicety, not a necessity. The way agents 'read' code is different - good agents are all about gathering data about how the code works. They'll prod around with it, execute parts of it. So 'readability' to agents really means 'verifiability'. How easy it is to mess about with that code to see it working. Let's take this scenario: - You have a huge module in your codebase that uses all sorts of complex performance tricks - It has 100% test coverage A talented human would need an hour or two to get comfortable working in that module. An agent needs zero time. They verify it with the tests, poke about a bit, and they're good to go. The takeaways: - When judging a module (or a codebase), don't think about how easy the code is to read - Instead, think about how easy it is to verify - Design your codebase so modules can be tested and verified independently - Feel free to sacrifice a bit of readability for performance - Cyclomatic complexity and function size don't matter to agents
1
41
day 33/365. 18, da nang, building bots for businesses. goal: $3k/mo. my first message doesn’t pitch anymore. one line about the bot, one question: who decides? a dental clinic answered in 2 minutes: “we passed it to him”. 24 cold messages 35 follow-ups 15 linkedin invites 0 interested, 8 no’s total: 484 sent, 36 interested. friday i ask if the owner watched the video.
1
1
25
honest numbers on the “who decides” message: it gets the most replies of anything i send (29%), but fewer turn into interest than my long offer (5% vs 10%). it gets me to the owner faster. the owner still has to want it.
9
the bots i build for businesses keep one rule the AI does the volume, a person keeps the decisions. the booking bot hands anything unusual to staff. the document bot shows the team who’s late, and they decide what happens next that’s “people in control” at the size of a barbershop
People should always be in control of the future. OpenAI is against the notion that it would be acceptable for people to be relegated to a role of doing things that “AI systems think makes sense to reward in humans”. People should not cede control of the future to superintelligence, and that requirement should ultimately constrain AI progress.
1
3
20
312 cold messages to businesses. two versions of the first message. the short one got more replies. the long one got more people actually interested. short (101 sent): 25% replied, 6% interested long (211 sent): 21% replied, 10% interested the only difference: 2-3 lines on how it works. what their client sees, what their team sees. my read: a short message is easy to answer, and that includes "no". the long one answers the first questions before they ask, so whoever replies is already halfway in. small numbers still. testing more, results here every day.
3
3
74
the record nobody’s counting: same quality, 40% cheaper i build bots for businesses. the model is the one cost that grows with every client message, everything else is a flat server fee so this isn’t a leaderboard story for me. the same bot just got cheaper to run, and the price i quote can drop with it
if this goes on for 20 more days we might have the first time a model has reigned supreme for TWO MONTHS in a row, which never happened before?
1
19
day 32/365. 18, da nang, building bots for businesses. goal: $3k/mo. 42 first messages today. the only yes came from a follow-up. an accounting firm ignored my first message last week. today's follow-up got "yes, let's see it". 42 cold messages 30 follow-ups 1 interested, 7 no's total: 428 sent, 0 calls. tomorrow: more follow-ups.
1
2
44
the follow-up didn't repeat the pitch. it brought one new idea for their work: sole traders and companies have different deadlines, so the bot reminds each client on their own schedule. that's the one they answered.
1
9
Day 31/365. 18, Da Nang, building bots for businesses. Goal: $3k/mo. Yesterday I wrote a parser. Today it read 913 Telegram channels. 170 new leads 13 cold messages 3 LinkedIn accepts, 1 no Week: 157 messages, 12 interested, 0 calls. That's the problem now.
3
3
122
What the parser looks for: channels that sell paid access and still hand it out manually. 913 read, most were news and airdrops. 170 made it to the list.
1
20
Day 30/365. 18, Da Nang, building bots for businesses. Goal: $3k/mo. Spent Saturday on a script that finds Telegram channels to pitch. Today: 20 cold messages, 1 follow-up, 5 LinkedIn invites, 1 interested, 3 no's (2 already have a bot). Week: 98 sent, 3 offers, 0 calls.
1
1
157
All 20 went to a new niche on Telegram. So far: 33 messages, 4 replies, 1 interested. Colder than accounting firms, too early to call it. If you sell to Telegram communities: where do you find them?
1
17
Day 29/365. 18, Da Nang, building bots for businesses. Goal: $3k/mo. Zero motivation yesterday. Did the plan anyway: 10 cold messages 2 follow-ups 8 LinkedIn invites 2 offers sent, 1 no This week: 72 messages, 4 offers, 0 calls. Then I walked for 2 hours. Mood fixed.
1
3
66
Side quest yesterday: how to deploy a client's bot so it's as easy as possible for them. Weighing a private repo vs a private Docker image, hosted in the EU. If you ship bots for clients: how do you hand them over?
2
25
He was ready to start. 20 minutes later there was no budget. Day 28/365 of going from $0 to $3k/mo with automation. Yesterday: 23 cold messages 13 follow-ups 4 interested, 9 no's 1 full offer sent Not a fit for now. My lesson: ask who pays and how before I talk about the setup
1
2
36
the bot he wanted: reminders that chase clients for documents before the VAT deadline. run a tax firm? DM me "bot" for the demo
1
28