The manifesto behind the launch of Bleep AI.
On 1st of April 2025 we launched a new hardware product. Only it wasn't real.
Why did we do this? Simple really. There's so much noise going on in AI that we've all somewhat gotten away from the real purpose of building AI products.
AI should be delightful — not daunting
We've seen one too many product launches with the claim that their AI is "so good you can stop hiring humans." We think this is fundamentally wrong.
AI is a tool — just as Excel or email are tools to get work done better, faster, and so on.
As a tool, it should keep its primary user in mind at all times: us. The consumers of software. Not the VCs, or the AI hype train, or the models themselves.
We believe in building human-friendly AI that starts with the user first, then builds all of the fancy bells and whistles after.
Because the reality is adoption is far behind technical advancements. And in the next 10 years, the real innovation will be how to get people to use AI in the best possible way — not in making it better at imitating people.
Users are leaving companies in the dust
Put our product in the hands of any user and they will intuitively pick it up right away.
"Oh so it's like live scan of an image — that's such a better experience!"
We hear that all the time from end users — the ones parking their bikes, delivering packages, returning items. They are the ones taking the photos.
But companies can be so risk averse in this modern age. Rightfully so. But this gets them stuck in a mindset of limiting liability, rather than seeing the immense potential.
"What if we create a claim? What if we add user friction?"
Instead of thinking: "What if we build such a delightful experience our customers won't be able to believe how smooth it is and come back to us every time?"
AI reinventing the wheel
It's a trap we all fall into. Short term memory makes us think we're inventing something brand new. When in reality, we're often just rebooting.
1. Slack / Discord — IRC and AOL-style chatrooms (1980s–1990s)
• Channels, commands, bots — it's all IRC, just with GIFs and dark mode.
2. Threads / Mastodon / Bluesky — RSS feeds, forums, and Twitter-style posting
• Every few years, social media platforms "rediscover" decentralized or minimalist posting.
3. ChatGPT and LLMs (2022+) — chatbots since the 1960s
• ELIZA (1966) was a rule-based chatbot. LLMs just gave them superpowers.
4. Ring Doorbell — intercoms with cameras from the 1990s
• Ring added cloud storage and app-based alerts.
5. AirPods (2016) — wireless earbuds in the 2000s
• Brands like Plantronics and Jabra had Bluetooth headsets ages ago. Apple made them slick, seamless, and desirable.
That's why the much more interesting conversation is not about the latest AI features and whose foundation model is better — but about the base human behaviours that remain consistent over time. And why we are trying to solve for those.
Dedicated to Clipy — the OG AI assistant
Spark conversation
Most of all, we want to spark a conversation about where AI has come, where it's going, and what this means for your users.
