Attain Institute

Attain Institute helps non-technical decision-makers in medium-sized businesses to improve their product management. 

We provide training and advisory services on product strategy, new product development, product operations, product management, and solo PM programs on nocode and genAI.

We convert your expert knowledge into in-house training programs and online courses. 

Maersk, Standard Bank Group, SAP, Viessmann, Klöckner, and others trusted our advisory, training, and community-building expertise. Over 400 PMs and 100 executives have benefited from our training. 

Best practices from Amazon, Google, and other tech companies are adapted to your industry and your specific organization into a custom pm program you can launch within 4 weeks. The programs include:

- Bootcamps with live work sessions in virtual cohorts 
- Self-service online course content by ex-AWS and ex-Google trainers
- Exercises customized to solve the specific challenges of individual participants
- Templates, Assets, NoCode, and genAI resources to solve the common pm challenges
- Everything you need to build an internal product community: events, challenges, content

Launch your custom product organization in 3 months:
- PM role definition, role profiles, leveling guides, job descriptions, incentives, performance metrics
- Internal PM community: templates (pm intranet, events, channels, news), governance, standards
- PM best practices: templates (PRD, BRD, Roadmaps, Methods), OKR and KPI, stakeholder management
- Product portfolio: governance, taxonomy, product strategy, product teams
- PM training: strategy, research, visioning, development, launch, growth, operations, retention, sunsetting, product lifecycles, roadmaps, product hacks, technical debt

Here is how we work together:
1. month - analysis: identify needs & gaps, benchmark, current state map
2. month - design: define program, customize content & tooling
3. month - launch: pilot cohorts, train the trainers

Transform in 3 months.

Chief Product Officer

Amazon Web Services (AWS)

Bring expertise on platform businesses, marketplaces and digital ecosystems to AWS customers. Principal Advisor to senior executives and product teams for key Customers in FSI and Manufacturing across EMEA. Helping to identify new business opportunities and incubate new products and services. Leading business conversations to drive innovation and improve customer’s capabilities. Helping to improve their business by using the full potential of AWS. Building solutions from the customer perspective. Advising top management on their digital transformation.

Principal Advisor

Hotpatch

Product Agency. Projects & Clients:
- Led squad at McKinsey to develop an internal suite of products to create presentations.
- Advised Datev Executive on creating an Executive Briefing Center.
- Created a platform for Google to host an international event.
- Led User Research for Leonardo, a children's TV producer.

Quit after catching Covid and taking a parental leave.

Hypoport AG

introduced design thinking and agile approaches into the company. 
fired for disagreeing with the strategy to create a holacratic organisation

Director People and Change

Mind-x AI

Cofounded startup on AI driven workshop experienced. Failed due to timing & tech capabilities.

Berlin School of Digital Business GmbH

consulting & coaching companies in digital transformation, focus on design thinking and agile, created AI training experiences for VW and Klöckner

Head of Innovation

GE Power Conversion

introducing new approaches to collaboration, social networking and design thinking to GE, used ML to visualise internal social network, built internal communication platform product

Innovation Lead

spent about 3 months in china training managers in design thinking. Managed teams in the design thinking wave, building products for SAP customers

Project Lead, Design Thinking

Alexander von Humboldt Institute for Internet and Society

co-created and co-founded the premier internet research institute in Germany with funding by Google. Worked with Google policy leads and Berlin universities to create a conference and research network including Harvard, Oxford, and Keio Universities.

Co-Founder, Program Manager

Ingosu

Productizing design thinking for corporate  customers. 
Created product concepts, prototypes and trainings for Haufe, Fraunhofer Society, Avato, Bertelsmann, and various SME.

Chief Product Officer, Co-Founder

Your business is eating up all your time, but you can't just keep hiring people. You're overwhelmed by technology and feel like you're not leveraging it properly.

That's where I come in.

Imagine developing and running your products with half the time you are putting in today. 

Imagine making informed decisions in days, not weeks. 

Imagine running 20+ experiments per month instead of 2 per year. 

If you are struggling to: 
- create digital products your customers love
- improve productivity in your business
- spending time working on the business instead of inside it

Let's talk and I'll show you how you can improve your workflows and processes.

Attain modern product operations with AI and automation. 

Past customers:
McKinsey, Amazon, Google, Facebook, SAP, Maersk, TUI, Standard Bank, DB, Haniel, Haufe, TUI, Mercedes, Volkswagen, Sanofi, Zurich, Deutsche Bank, +30 more. Ask me for the full list and 1-page case studies. 

----------

Customers I helped had the following challenges:

☒  Manual production of products & services, with no established systems 
☒  Products that don't perform and lack an effective product strategy
☒  Inefficient product development and delivery processes

Here is how I helped them:
☑︎  Build systems that create new products and services reliably
☑︎  Develop high-performing products with an effective product strategy
☑︎  Introduce best practices in product operations

first thoughts on ChatGPT 5.
10 short insights.

tl;dr: 
it's faster, better, and has a broader range of use now. 
if you're not using tools like this, you'll be screwed in a few years.

1) overall responses have improved. you no longer get sycophantic responses. it's straight to the point with no fluff. the result quality is better

2) queries have improved. no need to write complex structured prompts as it is better in understanding what you're trying to get. if there is ambiguity, it asks

3) more capable. the assets generated (code, documents) are on par with what was previously available by wrapper products like lovable or copy.ai

4) less user effort. no need to pick from different GPT models. there's just GPT5. the pro version just adds more compute

5) agent mode, think longer, deep research - now actually do what they're supposed to & seem to outperform other agent and reasoning tools (like manus)

6) study and learn is now a dedicated mode - i think this will break most courses and trainings if they don't adapt to this. i'd expect sth similar for "coaching and counseling" since that's a major use

7) canvas mode for assets (eg a web application) is crazy - it was good before, but now it's like sitting together with another person and working on the same thing simultaneously. i'd expect this to be built into an OS soon (maybe they'll create their own)

8) connecting and integrating other tools has become easier & there are more available. not sure why you'd need automation tools like zapier if this is able to do it

9) voice mode. if you aren't talking to it, you're missing out on a different type of interaction with computers. go see the movie "her", then go try it

10) hallucinations and mistakes are still a thing - but far less than before, or maybe just harder to detect now. it'll be interesting to see what other techniques (beyond LLM) they'll combine this with to get the glitches out

ahmet acar

I help business owners grow revenue, save time, and reduce costs with AI.

I don’t know why people are complaining here. It’s so simple:

Just heal your chronic diseases, get over your childhood trauma, stop your many addictions, get off social, overcome your brainwashing, fix your relationship issues, and meditate daily to overcome your ego.

Also don't forget taxes, income, rent, answer emails, and pose your “achievements” here while posting nonstop to build your personal brand. 

honestly, why are you struggling?

ps: don’t forget to say “I am fine” when someone is asking

I help business owners grow revenue, save time, and modernize with AI.

LLMs can't think or reason. The "thinking" feedback in the chat and the "reasoning" feature is just marketing. That said...

Thinking machines are possible.

Any human accepting the principles of science should come to the conclusion that thinking is merely another natural phenomenon that we haven't solved as of yet. 

That doesn't mean it's unsolvable.

If you believe authors like Ray Kurzweil, it should happen some time between 2029 and 2045. Others say around 2060. 

In any case, given more resources on research, why shouldn't we be able to build thinking machines? Given advances in robotics, why shouldn't these be housed in mobile, autonomous entities?

There are many other challenges associated with this for sure. From energy needs to battery technology and sensory challenges, there's a bunch of problems to attempt this.

But you see pieces of the puzzles appearing everywhere already. 

Whether LLMs can think or not is a useless discussion. Even if a system built solely on this technology could, there is no agency, no action. An LLM isn't running compute in a data center somewhere, planning its escape.

Yet... 

To think that there is anything special beyond our biology is hubris. 

So, granted, current approaches like LLMs, LCM, etc might not lead us there. But they are another step. It'll take insights from many other disciplines like cognitive science to get there. 

Unless we destroy ourselves as a species first. We seem to be very intent on doing so. This is a bountiful planet and an abundant universe. There is no shortage of resources in our solar system. 

Yet we stick to petty tribal conflicts, hoard resources and indulge in selfishness and unkindness to each other, rather than seeking evolution and intellectual enlightenment.

If our so-called leaders don't blow us up or ignore climate change to the extent that it finally kills the only habitat we have, we'll definitely get there. 

There are no laws of physics preventing it.
Only our own ignorance.

I have a firewall, password manager, 2FA authentification. Still got hacked.

2 out of 3 people reading this record their password on a document, use the same password multiple times, or share it with others.

So you're probably screwed even worse.
(but don't know it)

Just a reminder to: 
- use a password manager (keepass)
- check if your accounts are secure (use "have I been pwned")
- use secure passwords (use the "how secure is your password” tool)
- actually use VPN & firefall (bitdefender is pretty good)

...and log out of websites when you're done (which I forgot).

prediction: over the next days (weeks), you'll see a lot of posts shitting on the latest chatgpt release... and a lot hyping it. instead:

if you see new tech and new tools, go test them, read up on them, make up your own mind.

whether it's web3, quantum computing, or LLMs.

and if you don't want to invest the time, at least follow people with a (somewhat) clear-headed and objective view of things

---
note to linkedin - fix the damn feed. i get better info on tiktok these days...

Saudi is putting $100B into their AI initiative “transcendence”.

Will that be competitive enough to catch up? Most likely no. 

now imagine that money put into terraforming the region into a green oasis and climate tech instead. 

that would build legacy for centuries to come.

instead of a tech that may or may not pan out.

Someone asked for advice today on how to launch something and shape its story. The answer is simple but not easy:

It's just a lot of work.

Talk to lots of people. Talk to 100 people. 99 will say no. 1 will say yes.
Launch the first thing to 10 people. Fix it. Launch again to 20. Fix it...
It's not complex. It's just a lot of work. That's what makes it hard.

If you're busy, sick, or otherwise very short on time, it's so much harder.

Take Ari Meisel, for example.

A super busy agency owner working non-stop until he burned out and was diagnosed with Crohn's disease.

He basically couldn't do any work for months, then only work 1-2 hours a week.
I can relate - just went through sth similar.

So how do you do all this work if you have to work less?
The truth is, we aren't really very productive. 
We're not focused or efficient.

You can probably do what you're doing in 4h instead of 8.

It's simple, but not easy.

I'm obsessed by the idea of automated organizations. Companies that can scale their impact without requiring more and more people and resources to do so.

The larger organizations get, the more internally focused they become. 

People inside them no longer ask how they can serve others, but they talk about internal processes, politics, metrics, and rules. They spend more time cobbling together useless powerpoint reports to their higher ups than actually contributing anything of value to humanity.

That's an uncomfortable thought, so people focus on the daily grind, then entertain themselves to be distracted.

There's a lot in support functions that could be fully automated and handled by AI or rule based systems. But most people don't even use simple email filters and rules to lighten their inbox. Most don't use search tools well or automate their file organisation, so you get dozens of "where's X?" messages per day.

At the same time, I see small agencies starting to automate their onboarding, invoices, expenses, customer support, lead generation, proposal writing...

I can guarantee you that most large companies are still asking highly paid people with 15+ years of experience to waste time writing customer proposals manually. I know I had to do that at the last big tech company I worked for. 

100s of hours wasted with copy & pasting, tinkering, and editing text that should just be created automatically.

It wasn't just me either. Account managers, sales managers, consultants, solution architects... waste upon waste.

Meanwhile, I'm creating fully automated proposal generation systems for SME that only spend an hour (max) per proposal - down from 30h before.

And that's just one small business process. That's pretty much possible across the board and it's becoming easier with more powerful tools.

I imagine that future organizations could run themselves, with the only people building and delivering products and services while all internal support functions are mostly replaced by tools. 

The robotic company.

the main problem i see with AI agents (manus, chatgpt5 etc) isn't with the agents but rather with the sources they're accessing

or rather, they're trying to access. the error below comes up 50% of the time. current websites and information sources aren't meant for ai use.

even trying to scrape some of these sites returns problems.
there are no ways to interface with them. forget about MCP, UTCP etc. 
most websites still aren't even search engine optimized

i'm not suggesting websites should make it easy for ai to steal their data.
on the contrary, make information accessible, then charge for its use

but as long as "tool couldn't read the site" is a thing, agents will suck

# Open Call: The Vault ProjectClimate catastrophe is accelerating faster than our adaptation strategies.

 While institutions debate incremental changes, communities worldwide will soon need radical self-sufficiency to survive and thrive.I'm launching The Vault Project—a comprehensive AI-supported knowledge base for resilient community building. 

We're documenting the technologies, methods, and systems that enable true independence: energy generation, food production, manufacturing, waste recycling, and governance structures that work when traditional infrastructure fails.This isn't prepping—it's practical futurism. 

We're creating blueprints for communities that don't just endure crisis, but emerge stronger and more human-centered than what came before.Looking for contributors across disciplines: AI devs & data scientists, industrial and mechanical engineers, permaculture designers, social organizers, medical professionals, educators, and anyone with expertise in sustainable systems. 

Your knowledge could be the difference between communities that collapse and those that flourish.Ready to help design our future instead of just watching it unfold? 

Send me a message.#ClimateResilience #AI #CommunityBuilding #SustainableTechnology #FuturePrep

When the AI bubble bursts, it's not just going to take out AI startups.
It'll take down companies like yours, too.

Take NuBank - they built their customer support automation on OpenAI. It handles more than 50% of their support.

What happens when OpenAI goes bust?

There's hundreds of companies building on these AI platforms. 

It'll be a fun cascade to watch.

Everyone working with companies on AI agents sees this all the time. 
It's not surprising though. People don't even secure and encrypt email.

The problem with IT security however - no one is really buying it because "it didn't affect me". It's like insurance. You don't need it until you need it.

The flipside of agents - hijacking them with other agents. Do that at scale and it becomes easy to target anyone.

Microsoft's Copilot Studio agents got hijacked at DEFCON. Autonomous agents dumped full CRM records with no human oversight.

Security researchers demonstrated they could manipulate these agents to:

Reveal private knowledge and internal tools

Access and dump complete CRM databases

Execute actions without any human verification

The critical detail: these are autonomous agents with no real governance.

This breach exposes the fundamental risk of deploying AI agents without proper security architecture. When agents have access to sensitive data and can act autonomously, a single vulnerability becomes a system-wide exposure.

The demonstration shows agents revealing Salesforce records, customer data, billing information, and internal communications. The researchers gained control through prompt manipulation, turning the agents into data extraction tools.

For enterprises rushing to deploy autonomous AI: this is your warning. Every autonomous agent with data access is a potential attack vector. The convenience of "no human in the loop" becomes a catastrophic vulnerability when security fails.

The gap between AI capability and AI security keeps widening. We're building powerful autonomous systems on foundations that hackers can compromise with clever prompts.

Time to rethink what "autonomous" should really mean in enterprise AI.

Edit 1: Michael Bargury added more context at the comments, link to the post: 

https://lnkd.in/drbGndFq

(@Michael Bargury on X)

yes... and no. the problem with this:
current AI tech is still based on pattern recognition, prediction, and completion.

which means:

there have to be patterns in the first place.

but these don't just drop from the sky. observant, insightful, creative humans connect the dots of things they notice in their brains and come up with new, useful shit.

in this case, it goes all the way back to George Mattson who invented the synthesizer and the keytar. 

i guess technically, you could ask an ai tool to combine instruments that haven't been combined before. 

so, i did a simple zero-shot prompt:
"think deeply about all music instruments in existence. then, connect two instruments that have never been connected before and imagine a new type of instrument as a new invention that has never existed before. give it a name and describe its use"

It came up with:

"the Spectral Kalimba, a revolutionary instrument born from the marriage of the ancient African mbira and the otherworldly theremin. This hybrid creates music through both physical touch and invisible electromagnetic fields, allowing performers to literally sculpt sound in three-dimensional space while maintaining the intimate, percussive heart of traditional thumb piano techniques."

well...

you'd have to be willfully ignoring the signs at this point.

software as we know it is ending. that whole thing where you download an app, learn its interface, wait for updates? stable object with version numbers, roadmaps & feature sets. pretty much gone soon. 

here's gpt-5 generating a full beatmaking interface from scratch. someone literally said make a sick beat to celebrate gpt-5 and boom. perfect interface, beat already composed, ready to go in seconds.

that's not software anymore. 

software has always been this weird barrier between what you want and what you get. you have to learn its language, memorize its buttons, work within its constraints. even the best apps are still compromises.

but when you can just describe what you need and it materializes exactly as you imagined it? the software becomes invisible. it stops being a tool you have to master and starts being an extension of thought.

this kills the entire traditional development model. why spend months building interfaces when ai can generate exactly what each person needs in real time? why have roadmaps when the product reshapes itself for every interaction?

i shipped an app literally last week. perfect timing to watch my own industry transform in real time. though i think i picked the right niche. kids don't know how to prompt yet and parents want guardrails anyway.

but the writing's on the wall. the gap between intention and reality just disappeared.

wild to be living through it.

did my best to replicate the "garbled president gallery" mistake and some of the others mentioned in the article. no luck. seems to work perfectly fine for me.

here's the proof - https://lnkd.in/eMACt-VP

not sure whether there are inconsistencies across users, or some people just got unlucky, or aren't able to use it well, or are just bashing it 

in any case, it's not a good look when there is a very wide range of experiences across customers.

Building Cognitive AI to Unlock Real Intelligence | Co-Founder @Aigo.ai | AGI Missionary

𝐆𝐏𝐓-5: 𝐅𝐫𝐨𝐦 𝐅𝐚𝐢𝐥𝐮𝐫𝐞 𝐭𝐨 𝐅𝐚𝐫𝐜𝐞, 𝐀𝐜𝐭 𝐁𝐞𝐟𝐨𝐫𝐞 𝐭𝐡𝐞 𝐅𝐚𝐫𝐜𝐞 𝐁𝐞𝐜𝐨𝐦𝐞𝐬 𝐎𝐮𝐫 𝐅𝐮𝐭𝐮𝐫𝐞

In a display less of technological triumph and more of tragic theater, GPT‑5 has emerged not as an evolutionary leap, but as a spectacle of AI hubris. OpenAI’s vaunted “reasoning” powerhouse has crashed into reality with a dull thud, its debut marred by hallucinations, incoherent model-switching, and downright absurd visuals.

For at least two years, the debut of GPT‑5 was treated as a milestone, supposed to outpace its predecessors and symbol a stride toward AGI but instead it revealed the fundamental emptiness of the scaling-first narrative. The much-touted “seamless” model routing dissolved into slapstick as conversations ricocheted from moments of brilliance to sheer nonsense, often indistinguishable in tone. And the PR-polished demos collapsed under the most basic scrutiny, bar charts literally lying to your face, inflating performance while the actual numbers told a far grimmer story.

"Startlingly, OpenAI kowtowed to their pressure and gave back access to 4o to paid subscribers — but already, the writing was on the wall.
Part of the reason behind making GPT-5 the only available model, OpenAI insists, is that it was built to switch seamlessly between all of its prior versions to better provide users with what they need. But as Wharton AI researcher Ethan Mollick notes on Bluesky, "seamless" isn't the right word to describe the current reality.

"When you ask 'GPT-5' you sometimes get the best available AI," Mollick posted, and "sometimes get one of the worst AIs available and you can’t tell and it might even switch within a single conversation."

And that's just the tip of the crapberg.

The latest model has also demonstrated, many argue, even more of a propensity to "hallucinate," or make stuff up — and apparently, it's taken to gaslighting folks, too.

Case in point, multiple people have found that GPT-5 will, when asked to generate portraits of recent presidents and list their names and years in office, invent a garbled version of history that's equal parts funny and unsettling and that the real-world internet is rapidly filling up with AI garbage that's ruining the experience for both human users and future AIs trained on all that slop."

LLM hypesters are completely clueless about AGI, they say if LLMs don't work they'll introduce swarm of agents built on LLMs for intelligence to emerge and their leaders are constantly gaslighting from both sides.

The hype is deliberate & the bluff seems to be part of the business model. They need new discoveries in compute & energy just to become profitable, so no AGI ever from LLMs.

Scaling a flawed paradigm doesn’t turn it into something new, it just makes the flaws bigger, faster, and more expensive.

That was fun while it lasted, now it’s time to build cognitive AI, unlock real intelligence and leave LLMs behind for good before the farce becomes the future.

New edition of my newsletter is out. This time: transforming your work with AI.

Open article: Transformation in 7 Steps by open.substack.com

opt out via: settings > privacy > privacy settings (click everything off)

CTO at Omniscien Technologies | Board Member | Strategic Advisor | Consultant | Author

Trust Betrayed. Again.

Anthropic—the company that branded itself as “privacy-first” and “safety-driven”—just torched its own moat.

Starting now, Claude will train on your chat transcripts and coding sessions unless you manually opt out by September 28. Five years of storage replaces the old 30-day deletion rule. Free, Pro, Max, Claude Code—no exceptions.

This is not an update. It is a betrayal.

→ Hypocrisy laid bare: The self-proclaimed “responsible” AI company now runs the same playbook as the rest—harvest first, ask forgiveness later.
→ Compliance nightmare: Sensitive conversations, contracts, legal docs, and code can now sit in Anthropic’s servers for half a decade. Opt-out ≠ consent.
→ Structural exposure: For governments and enterprises that bought Claude for its privacy promises, the foundation just cracked.
→ Pattern confirmed: In the end, every closed model company caves to the same growth imperative: extract more data, hold it longer, and lock users in.

The last fig leaf of “privacy-first AI” has fallen. The message is simple: sovereignty and control cannot be outsourced.

The question for every policymaker, CIO, and enterprise is now clear: how many more times will you let “responsible AI” vendors betray your trust before you build systems you truly control?

https://lnkd.in/gm2J-T6h

Open article: Anthropic updates privacy policy, will train on Claude chats by perplexity.ai

Anthropic updates privacy policy, will train on Claude chats

always fact check everyone and everything

Co-founder of the AI, Tech & Privacy Academy (1,300+ participants), Author of Luiza’s Newsletter (75,000+ subscribers), Mother of 3

Hard pill to swallow: ALL existing AI chatbots produce false information presented as “facts,” yet:

- Companies pressure employees to use them or get fired
- Teachers incorporate them into the educational system
- Millions rely on them for therapy and companionship
- Students use them to do homework, research, and cheat
- They are replacing search engines
- People use them as writing and reading assistants
- There is little to no regulatory oversight
- AI liability is still a legal grey zone
- Many celebrate them as “the generative AI revolution”

In 10 years, they will (correctly) call us ‘absolutely nuts’…

Taplio

ahmet acar's Linkedin Analytics

What is ahmet talking about?

Who is engaging with ahmet

's Best Posts (last 30 days)

Want to drive more opportunities from LinkedIn?