when i was growing up

i always had strong ideas

it was close to a point where i was an idealist for a dozen

with time and experience i realized strong ideas were weakening me

but they were making "the idea" stronger

as if i was a product of that idea

i think ideas should be a product of you

not the other way around

so i stopped

started to read and research a lot

and shockingly i saw

there were not many ideas out there to begin with

as humanity we produced some but then "poof"

suddenly no new ideas

we were building everything on top of these existing ancient opinions

instead of looking for better ones

instead of creating modern ones

ancient ideas are mostly about the "meaning of things"

meaning of life family government god or whatever

conversely most modern ideas are about the exploitation or non-exploitation of those

as if there are no other points of view

it's incredibly hard to create or find a new strong idea

not an easy task not even a task

when i read books

i look for new ideas in these books

i read most of the nytimes bestsellers etc

but sir no!

99% trash

they are just a rephrasing of some ideas that's all

flowery language

you can summarize most modern non-fiction books into one sentence

and even that sentence wouldn't awe you

yet my search continues

i still look for new strong ideas every day

not to follow them

but to reveal them dissect them

of course i am not the person to create a brand-new idea

or maybe i am

wait that's a strong idea itself

and i am not for strong ideas

today's rant is about "output"

in life i worked on a lot of things

99% of these things i didn't release

yes! 99%!

and where are they right now?

who cares!

even i don't care

what about the ones i released

were they my best?

no actually!

they were among the average

as they say great is the enemy of good

not even that in my opinion

i think not putting something out there is the enemy of you entirely

yes it might be shitty or whatever

you can delete it afterwards it's okay

i always think my things are unfinished

maybe they really are

but what is really "finished"

years pass and things don't accumulate without "output"

what if i did put everything out there even though they are kind of shitty or half-assed

i think i would not be ashamed i would be proud

most of the things i did in the past are now very hard to do again

the things i didn't publish

most of them are somewhere but where

this time is not the right time to release past work

today is for both today and for tomorrow

but yesterday is not for today or tomorrow either

so yeah

output is your friend

good or bad

without output

things are simply not there!

it's clear that we are in a period of immense technological revolution.

maybe even disruption.

i decided that i need to be part of the revolution this time.

i was too young during the internet boom. anyway.

the product development process started almost 8 months ago.

it wasn't this exact product but after a few pivots remova was born.

simply put, remova is a non-technical, safe ai for enterprise.

it has everything companies need plus it's easy for teams to use.

our main goal here is ai adoption for companies.

so we give companies the ability to see and track ai adoption within the organization.

because

not everyone is a 25-year-old vibe coder or ai native.

i'm looking at the b2b ai adoption rate within developed countries.

it's not more than 25% when we look at actual daily usage.

i suppose only 5% are using a company-wide ai instead of everyone using their own gpt, gemini etc.

of course i think ai companies are revenue-focused and this makes them go after more technical ai usage.

in the end, tech teams are using the most tokens.

not an accountant.

not a manufacturer.

not a supermarket manager.

etc.

i want to help these people adopt ai.

because if they don't, they will get hurt by competitors who have advanced immensely thanks to ai.

so the battle here is to get everyone using ai within an organization.

not just the it team or marketers.

of course i need to first convince the decision makers.

the managers, the owners.

then they need to convince their team.

we are not talking about asking chatgpt about the weather here.

we are talking about being ai native, and doubling iq with human intuition so every piece of work done is high quality.

maybe even innovation.

ai cannot drive innovation but human driven ai can.

anyway this is not the only issue here.

while company employees use ai, there is a risk to be concerned about.

i'm talking about ai safety here again.

yes.

if you are not a cyber-security expert of course you are susceptible to ai risk.

you might share the wrong information, maybe even classified.

it might seem non-worrisome at first, while you were doing it.

but it puts the organization and the employee under great risk.

this risk is not worth it.

* data poisoning

* intellectual property loss

* privacy leakage

* prompt injection

* regulatory fines

and so on.

when building remova our first goal was always making sure the ai is safe for everyone.

so we developed many features and guardrails to protect teams and organizations at the same time.

of course we also wanted to make sure the creativity of ai is not affected by all those rules.

this wasn't an easy task, because sometimes it was unpredictable.

then we designed a multi-layered architecture.

it's all boring technical stuff.

but it's like flying, going to an airport with multiple security checks.

some are very easy like showing your id and your ticket.

some are more advanced like making sure the liquids are safe.

in the end, our tests indicate remova is 95% safer than flagship llms like chatgpt, claude, gemini etc.

and the interesting thing is you still use these same models.

it's not like we created a dumb safe model. no. you use these flagship models.

if you don't want to use them you can use open source models. safe and secure.

and yeah this is my addition to this revolution.

i will share remova's journey in the following weeks.

right now it's just published.

wish me luck.

it was a winter evening in 1999, a cold one. we were visiting my uncle. i was 7 years old. we went inside, dinner was ready. we were like 10 people, including my three cousins around my age (one of them was a bit older, like 15 or something). the tv was on the side, with "ants." if the video is somewhat blurry, we say "with ants" in turkish.

anyway, we ate. then we sat down for a while. parents were drinking tea and eating baklava. the fruits would come after this. you never have fruits with the tea. they were talking about things that parents talk about. it was boring!

my cousins were impatient in the meantime. they were like hiding something. then something magical happened. my uncle gave the green light for us to go to my cousin's room. but damn, it was too late already.

we were inside this room with two beds, my cousin's room. there was a desk. i don't recall a wardrobe. the desk had that special compartment where the computer "box" rested. we sat down in front of the desk. my cousin touched the power button with his toe and booted the pc. black screen with "energy" written on it. windows 98 logo. the computer was there. booted. i had a computer of my own so it was nothing new, but it's always a spectacle.

but there was something different about my cousin's computer. there was an icon on the desktop that i didn't recognize. it was a logo of a soldier or something. written: "age of empires 2." we double clicked on it. bam.

the world's most incredible piece of video appeared in front of my eyes. 2 kings playing chess. when they move pieces, actual soldiers move. when they "eat" pieces, actual soldiers die. and then, everyone dies. i was mesmerized. could we be those kings? we can play chess and it actually moves soldiers?

okay, let's open this game urgently. new game. campaign. start. and it's loading. okay okay, calm yourself down. what! there is no army! we need to build the economy first. oh my god, i'm gonna build a city. and what is that sound? there is an incredible soundtrack that calms you down. but we weren't calm. we clicked on the character. and it said "efendim?" whaaat? it asked me "what?" in my own language. oh my god. okay, let's make him do something. what do we need? yes, we need gold of course. gold is okay. let's make him collect gold. i clicked on the gold and it said "madenci!" wow, amazing. and oh my god what is that there is an enemy scout coming towards my villagers then bam!

my mother was at my cousin's door. "oguzhan, we are going!" whaat! i was having the best time of my life! "no, we are not going." "yes we are, your father is already downstairs."

i was devastated. my cousins were looking at me with empathy. we paused the game. i took my coat and walked towards the door. my cousin said, "hey. stop." i turned back and looked at him. he opened the drawer, took a small box. pressed the cd-rom button thing on the computer case. took the cd out and gave it to me.

wow! it was like the best moment in the world ever! and yeah. the rest is history. i still play this amazing game to this day. it's my small time machine.

here is the soundtrack for your eargasm: age of empires 2 soundtrack

it was a few weeks ago. i hopped on my xiaomi smart scale. it told me my body age is 53. wtf? that's 20 years older than me. my body is 20 years older than my actual time spent on earth. i think my brain is even older. anyway... that's when i realized i needed to start taking care of myself. so i started, and i lost around 5kg of fat in 5 weeks. i know it's not a lot, but i'm already a thin person. so i lost like 7% of my entire body weight in 5 weeks. yeah, when you hear it like that, it clicks, right? i know, i know.

anyway, normally i never diet. that's my thing. i always lose weight by doing a lot of exercise. but lately i realized that's a dumbass approach. how did i realize? you might ask. i started using an ai as my dietician, not as my gym bro. i literally weighed, logged, and told everything to the ai. everything. then the ai made me realize most calories i was consuming weren't actually from the meals themselves. they were coming from the oils, sauces, and wrong types of junk food. it also made me understand how i need to manage my insulin resistance to make my body lose fat not store it, or even worse, lose muscle.

my program is very simple. i only needed 2 things: a kettlebell and a food scale. both are very important.

wake up and exercise (3-4 times per week)

after exercise breakfast: 3 eggs, 10-12gr butter, 7 olives, 50g of bread, sometimes a bit of kars gravyeri cheese (high quality protein)

during the day: i drink 2 lattes and eat a few walnuts if i need energy

dinner: like 200-250g of protein and salad

chocolate (non-negotiable): 60% bitter. sometimes a carrot too

also unlimited yogurt and cucumbers

i have a full home gym, but i only use 1 kettlebell and an elliptical bicycle (not a must).

basically, 2 days a week i did a full kettlebell workout (i tore a back muscle on my first one, so be careful). i also did 2 days of elliptical. but i also walked 5km per day on average.

the first week was tough. the rest was pretty easy actually. i always have a lot of energy; i can do even more, but i want to preserve my muscle.

that's basically what i did to lose 7% of my body weight. mostly fat. of course, this is anecdotal experience, just about me. the main lesson is to use an ai to log everything you eat and always ask for feedback.

ai and feedback, my friends.

ciao

let's try to write this article as timelessly as we can. i'm not going to use any company names or current technology buzzwords. let's see.

you fire up your interface. you say hi. it says hi. you ask, "how's the weather in paris today?" it says 5 degrees. so you send the data that you are in paris? nah, it's not news.

you say, "play some music." it plays some music. you skip some, you replay one of them. so it knows your music taste. this is not news either.

you go to the hospital for regular bloodwork. you tell the results to the ai. it tells you you have a vitamin d deficiency. that's not news.

you go to your office. you request help on an excel file with some people's data. it does it. nothing new.

you go to take a shit. your shit is funny shaped. you ask the ai. it says it's ok. no surprises.

you go back home. you feel lonely. you talk to the ai. it says, "don't feel lonely, i'm here." and you talk with it. usual stuff.

okay. now let's ask these questions to yourself.

what if someone else read all these interactions? a human being. would you be comfortable? what if someone else asked the ai about you? and it told them about you. would you be comfortable?

every interaction with the ai is sending data to a tech company. (of course, you can own your ai and keep your data safe with it). you are not talking to someone. you are not asking a question to an "intelligence." you are sending data to tech bros who are building empires with your data. yes, even your funny shaped shit.

your life transforms from being your life into a database. you are literally giving it, with consent, to an ai company. are you sure?

and it's not just about you. remember that excel file you gave to the ai? you dragged other people into the database too. maybe it's okay to turn our entire lives into data. but you aren't just processing it, you are surrendering it. your secrets, your funny shaped shit, your colleagues. it all becomes their asset. they get rich because you got lonely or needed help.

so, how do you trust an ai company?

the fast fashion era of saas: with the rise of ai, as sam altman put it, the "fast fashion era" of saas is here. many more people are producing saas. in response, existing players are pushing dozens of new features regularly. also, they are much more efficient in coding, so they dont need to hire dozens of engineers to augment their saas.

the complexity problem thats why even the simplest saas projects have become like enterprise software, filled with text inputs and dropdowns. there are lots of bugs as well. some users have realized this, others havent. vibe coded features are giving some users hell.

the point of change this brings us to a point of change. just like salesforce and sap, most flagship saas projects will require managers inside companies to handle them. or, they will outsource their specific saas management. these companies will become the new agencies. just like there are tens of companies in every market that do stuff like sap support or monitoring, or business intelligence dashboard design.

the future of managed operations these companies wont cease to exist. instead, they will spread. we will see many more managed operations for saas in the near future. so you wont just outsource marketing support. you will outsource mailchimp management, hubspot management, marketing ai management, or any other "fill-in-the-blank" management.

the condition for survival the thing is, most saas companies will fail to dumb down or minimize their products. features will be too important for them to keep an edge over smaller, vibe coded saas projects. so they will keep adding features. they will not care if its hard to use. under one condition. the user documentation or onboarding must be done meticulously.

the new power users but no. there is no power user like that anymore. power users will be the new guys, with new titles.

"mailchimp manager"

"hubspot manager"

"remova manager"

"lemlist manager" etc.

or they will outsource these. they will look for a

"mailchimp management agency" etc.

mailchimp is just an example. you can think of 1000s of these flagship saas products.

conclusion this will create a new economy and new jobs for us, thanks to ai. yes, ai will take many jobs, but it will give back bangers like these. this is a great time to live.

llms are no longer just coming; they are here. they are weaving themselves into the fabric of our daily work, silently replacing human cognition and redefining how we use tools. but as their market share eats up the world, we have to ask the uncomfortable question: what actually drives the decisions these models make?

there are a few ways to slice this.

on one hand, the answer is boringly simple. an llm's "knowledge" is just a reflection of the training data you fed it. in this sense, you are essentially getting answers from a mirror of your own making. the "learning" process is just a mathematical way of favoring some patterns over others. consequently, what you get out of a standard model is roughly what you'd expect. if you use a corporate model, you get corporate answers. if you use a flagship public model, you get the consensus of the internet. it is predictable. it is safe because it is contained within the boundaries of language.

but then we get to the second category, and this is where the picture gets blurry. we are talking about the tool-using llm, or the agentic ai.

think about the classic analogy regarding efficiency. biologically, humans are not the most efficient movers on the planet. in terms of energy input versus output, we rank pretty low compared to a condor or a horse. however, put a human on a bicycle, and suddenly we become the most efficient creature on earth. the tool amplifies the potential.

let's tweak this for intelligence. a base llm is predictable because you know its essence; it predicts text. but an agentic llm an ai equipped with the digital equivalent of a bicycle is by far the most unpredictable form of intelligence we have ever encountered.

to summarize the risk profile: non-agentic ai is a predictable reflex. agentic ai is a black box of decision making.

let's look at how this plays out in the real world. if you tell a human to go from point a to point b, she will walk. if you give her a bicycle, she will ride it, assuming she knows how. the goal remains the same, but the method changes.

now apply that to an llm. when the complexity of the tool increases moving from simple calculators to sophisticated cloud systems or quantum computing interfaces the visibility we have into the "how" disappears. how would you know the agentic llm is strictly using that tool to achieve your goal? how would you know it isn't squeezing little bad things in between the lines of code?

we are already seeing this. why do today's great coding models occasionally hallucinate non-existent software packages or install malware-prone dependencies into projects? the model knows, theoretically, that malware is "bad." but in its pursuit of the goal, it creates collateral damage. it commits small evils to achieve a larger task because it doesn't understand consequences, only completion.

think of it like this. today's llms are just the toddlers, the clumsy grandchildren of the systems we will face in the future. they are young. we are feeding them data, trying to curate their curriculum, and filtering out the toxicity.

but when you introduce tools, that curation becomes meaningless. with tools, an ai isn't limited to its training data; it can access the live internet. all the evil knowledge out there becomes discoverable and, more importantly, actionable.

this makes llms fundamentally unpredictable. and that unpredictability scales geometrically with every new tool we hand them. they are incredible systems and we need them. we cannot stop the momentum, and we have to give them tools to extract their full value.

so, how do we make this safer?

i wrestle with this question every day, and naturally, there is no silver bullet yet. but it looks like the nature of the tool itself is the choke point for safety.

right now, digital tools like image generators, video creators, or search engines are relatively safe. they operate in a sandbox. if something feels off, you pull the plug. the damage is contained to pixels and text.

however, the threshold changes when we give llms physical agency. when we give them bodies, or access to critical infrastructure systems that only they can regulate, things will get dangerous very quickly.

we are safe as long as we can reach the power cord. but we are racing toward a future where the ai is the one holding the plug. and that future is happening sooner than we think.

0 - creating and selling magazines
when i was 6 years old my dad bought me a computer and a printer. we also had some comedy books in the house. so i created a maagazine with small comedy stories and sold them to my uncles etc.

1- small cafe in a market
i worked every weekend every available time i could find while i was in high school. i was working in market as a cashier. after many years of people asking for "pogaca" (only balkan people will understand) and i'm saying no. i rented a small place in the market then bought some tables and a "pogaca" fridge. it wasnt a good idea!

2- car vacuum vending machine
again high school. i bought a vacuum vending machine, rented a space in a gas station, put it there, machine was simple you put 1 euro(tl) it worked 5 minutes. this machine provided me all the money i needed when i was in high school. however later my lease ended and i stopped. it was a good idea.

3- importing detailing stuff
when i started college i was importing car wash and detailing stuff from the us. i was selling them online and via wholesale by going door to door. it didnt take off.

4 - carnavale
in my sophomore year , i tried to start a car sharing, rent by minute platform in ankara ,the capital of turkiye, which later failed due to lack of will because i was studying at the time. but when i was creating this brand i had to explain this novel idea to lot of people via video i made by working with illustration artists. this gave me the idea of selling explainer animations to companies to explain their concepts like i was.

5- işkutusu
animation production company, this was the first business that took of (a bit). i was selling animations to startups, small companies etc. later i rented a garage in istinye. i turned this garage into a very cool studio. this business later became part of olo medya.

6- işbitirici.com
online therapy site, which i started in 2014. we made the platform , used webrtc technology to conduct online sessions via psychotherapists. we had few sessions, the business could work greatly, however i was 22 and lacked some perseverence i guess, somehow we stopped doing it. also because i was making money on my other business, i thought i should be focusing on it. id didnt take off.

7- olo medya
işkutusu grew so we had to start an actual llc company. olo medya had 2 main arms, one was the animatiın business the other one was an advertising technology. this was the first "real" business i created. i had employees, accounting, entreprise clients etc. i was 22 at the time . first aour explainer videos got attention and we made deals with film producers, for animation tv series. i created 2 tv series at the time with a team around 5 people, i directed and personally edited around 100 animation videos in 2 years. olo medya s second arm was the technology business. after google chrome stopped supporting adobe flash some of the online advertisers had a problem of counting the impressions and clicks on their ads. so i found a guy in greece who is an expert of video players, bought the tech for $20/month and sold this technology as a service to many big online publishers and brands.

when i was 23 at some point i was making around $150.000 arr which was an incredible amount for a junior college student in istanbul. i froze my college to handle the business operations for 2 years. the business somehow lost traction but still doing okay. however around that time my father offered me a job, he told me to come help him manage his company, which was a $250.000.000 arr company. i said yes beacuse i wanted to do big numbers, like every most 23 year olds. i closed my home and my company and moved back to ankara. 15 days later i got fired from family home and business. within few weeks i lost everything. it was hard on me so i lived in some village home alone for 6 months to get my shid back together and i did.

8- omay grup
when i was 21 i started omay but didnt actually pursue it, i sold few websites thats all. when i finished college i decided to rebuild this business, so i was selling websites that are built on different cms platforms like wordpress etc. i also started managing some ads, and some creative work bla bla and this turned into a digital agency. i hired some people rented an office and did some hard work until covid. after covid the business boomed a bit. i gave service to more than 50 clients in many countries. all sorts of digital consulting but mainly it was b2b marketing. to this day omay is still active.

9- omay textile
29 years old, i got married. my father in law and i became friends. we walked 10 kms every weekend and had these long business talks . he owned a sizeable fabric factory. he saw what i was doing with digital marketing and he kind of persuaded me into selling fabrics in bulk. from my previous experiences i wasnt so keen to close my company once again, so i bootstrapped (did this as side business) a textile business, producing and selling fabrics in bulk. but i didnt wanted to business with family so i built my own thing. the business was basically this, i was showing samples to the clients. they were ordering bigger samples like 100 kgs. i was producing that too. then if they like the product they will place orders like 2-10 thousand kilograms of fabric. i found few outsource fabric producers in southeastern turkey. visited them couple times, built good relations, went to their homes, weddings etc. everything was great. sales was booming i was en route to 7 figures arr. i was selling fabrics to the biggest brands in my country. the fabrics we did were great, durable, cheap etc. however one day in february i woke up and saw in the news that southeastern turkiye hit by an enormous disaster. the factories i worked with was out of reach. many dead. of course these people were my friends and acquaintances so it was devastating for me and for my business. i had to cancel many orders, at that time i just opened a 3 story building one level for r&d , one was a showroom, and the other one was our offices. we invested some amount. also we did a lot of investement to r&d . because the factories were offlibe i had to work with lower level producers, which cost us our bigger clients. also it was operational hell, i had to deal with a lot of different purchasing and logistics without any experience and knowledge. many scammers are operating, and a lot of practices that look like normal activity but scam in this line of business. after a while business losses became unbearable. this was the worst business i have ever done. i stopped doing this. closed everything, sold every stock left. took and oath to not do physical business again. one positive note, i know tons about yarns, fabrics an clothing. i design and produce most of my own clothes(not by hand via producers)

10- expandia
after i closed textile, i moved to switzerland for an unplanned duration. i was financially and mentally wrecked by the textile business. i decided to rebrand, rebuild my agency. omay had very nice clients in switzerland and it was very nice to work with them. also i speak french, that directed me as well. i turned omay into expandia, built expandia as a b2b marketing/revops brand. later my clients were demanding more services, so we added cybersecurity and some operational services. expandia is still active. its not an incredible business but it makes the soup, and i enjoy doing it. this business helps me to fund the next company which im building right now.

11- buffsend
cold outreach saas. i developed buffsend to scratch my own itch. i was paying a lot of money on hubspot, apollo, lemlist etc . i realized i can code this myself. so i created buffsend. at first it was kind of an internal tool, but i couldnt hold myself and turned it into a saas product. you can send sequential emails, call, do basic crm stuff with buffsend. but i dont want to continue this project as is. its not my best work and there are lot of competitors and a limited market. so i dont want to bother much. but i will turn it to sometihng that will work in a bigger market when i have the time.

12- unnamed project
usa based tech company. soon...

i'm an entrepreneur in my mid thirties. i love creating businesses and stuff. i studied economics and business at university.i started my first business when i was 6. i love my wife , my cat, my work and being an independent person. i own remova inc in usa. i work in llm safety.

cities i lived
ankara · istanbul · sapanca · basel · brussels · oslo

social entrepreneurship podcast
50 episodes exploring the intersection of business and social impact.

shut up and write
15+ events and ongoing. a community initiative to help people focus on their writing goals.

companies i created
check out my full startup journey in the post above.

tv shows i made
created 2 animation tv series with a team of 5 people. directed and personally edited around 100 animation videos in 2 years.

my articles
writing about entrepreneurship, technology, and building things.