It’s the beginning of 2021 and we are slowly coming to the realization that our startup Sentinel, which provides AI-based deepfake detection, is about to fail. As a founder, this is one of the hardest realities — facing the collapse of something that you have built for years is like eating glass and looking into an endless void.

Yet, the journey didn’t end there. Instead, I will tell you the story what happened next and how it turned out to be one of the best things that happened to us because it led us to the belief that the Internet should be owned by citizens and why we are building NFTPort, the Stripe for NFTs to fulfil that mission (if you wish to skip the pivot story, then head straight here).

Origins — Protecting Democracies Against Deepfakes

‍

With our team previously working in organisations like NATO, Nvidia and having a background in cybersecurity and AI, we realised in mid-2019 that the threats from AI-generated media — deepfakes — are growing and that they will increasingly be used maliciously to wreak havoc in societies by enabling the spread of disinformation. This is why we started Sentinel — to build an AI-based “deepfake antivirus” to protect citizens and democracies against disinformation of which negative effects all of us have felt during the pandemic.

During the last year and a half, we built a leading AI-based deepfake monitoring and detection platform, supported by an in-house big data infrastructure to gather the largest dataset of real world deepfakes. Our go-to-market strategy focused on the public sector as the first target and we worked together with the Estonian Government and the European Union to support the fight against information warfare, and built an extensive global network of leading organisations in democratic countries — from YouTube all the way to DARPA.

Although there was a lot of interest and support towards what we were doing, by the beginning of 2021 we came to the realisation that building a sustainable business providing AI-based deepfake detection was simply not viable. The decision came down to a wrong market timing and our realisation that there is a better way to solve the problem.

‍

It Doesn’t Matter how Good the Dog Food is if the Dog doesn’t Eat it — The Importance of Timing & the Right Solution

Although the data in 2020 showed an exponential growth of deepfakes, these were mainly generated for entertainment and porn purposes. Our hypothesis of 2020 being the tipping point for the manipulative use of deepfakes was wrong — they simply weren’t causing as much problem as expected. Even as deepfake technology was and is developing at a rapid pace with examples of malicious uses constantly surfacing, synthetic media has not yet caused a burning problem that would force organisations to use and pay for an operational detection solution. This is not to say that there is and will be no threat. To use an analogy, it seems deepfakes are today where malware was at the end of the 90’s — malware existed but criminal groups started spreading it rampantly in the beginning of the 2000’s when they found a working business model by stealing personal information such as credit card numbers, etc. through malware. So, the timing was off. We still firmly believe that synthetic media will have immense — both negative and positive — effects on future societies.

Second, during our customer engagement and sales we came to the conclusion that AI-based deepfake detection is likely not the most suitable solution to address the looming synthetic media problem. Although we managed to build a leading product that provided near real-time deepfake detection with 95% accuracy, the reality was that maintaining this level of quality would become almost impossible in the long run. This is largely due to a major imbalance between deepfake creation and detection efforts, with the former receiving multiple times more attention and resources.

‍

In the long-term, synthetic media will become pixel-perfect. This means that there will be no discernible difference, for example, between AI-generated media and a smartphone video recording. Thus, detecting deepfakes with AI is only a short-term solution. We further understood that high quality AI-generated synthetic media will inevitably be part of life with mostly enabling positive uses, be it for entertainment, marketing, education or healthcare. This is again similar to malware: most software is created with a positive purpose and a small part of it is weaponised to cause harm.

Consequently, we came to the conclusion that truly effective solutions do not actually need to address the “technical fakeness” of media itself, but the focus rather needs to be on ensuring the authenticity of the source. In other words, one needs to be able to verify and have trust over who is the legitimate author of the media, be it synthetically created or not. We see that the possible threats from deepfakes will be largely mitigated if there are solutions (for example, the Content Authenticity Initiative) to securely authenticate the source of the media, enabling people to come up with their own conclusion as to the validity of its content.

We believe that development and standardisation of authentication and data provenance infrastructure should be the core focus in the fight against deepfakes and disinformation on the technical level, in addition to various regulatory and educational efforts at the societal level.

‍

The Insight that Sparked NFTPort — Stripe for NFTs

Building on the realisation that trust-enabling infrastructures are likely the key to solve these types of issues, we decided to look beyond AI-based deepfake detection for alternative solutions. It soon became apparent that the answer could lie in blockchain technology. More specifically, the concept of non-fungible tokens (NFTs) got our attention. When we delved into the NFT ecosystem, iterated with different concepts and researched for options for better authentication or data provenance solutions, we experienced that it is extremely painful and time-consuming to build NFT-based applications due to the lack of developer tooling. Having used Web2 APIs such as Stripe, Twilio, etc. we knew how great APIs can empower developers to build new applications and bring existing ones to market with 10–100x less time and cost. On one hand, we were amazed that something like this doesn’t exist yet for NFTs but at the same time we knew scaling up data infrastructure is an extremely hard technical challenge because that’s what we did and got really good at Sentinel as any large scale AI-based product is 80% about the data infrastructure. This was the insight that sparked NFTPort.

Seeing the opportunity, we dug deep to validate our theses. We spoke with over 100 NFT developers to understand what their hair-burning problems are, goals and outcomes they are looking to achieve. And suddenly we were not alone in our experience: developers told us that it’s hard, time-consuming and expensive to develop NFT applications. One needs to have extensive prior blockchain experience, set up and maintain complex infrastructure or rely on existing APIs that are limited and unreliable. Some teams have even started to develop their own NFT infrastructure because of the lack of solutions. To build an NFT application which reads a lot of NFT data or mint NFTs programmatically feels like reinventing the wheel on every step of the way, leading to general frustration and painstakingly slow product development cycles causing months of wasted time in an extremely fast moving and competitive market.

NFTPort is the Stripe for NFTs — a one-stop, simple, and developer-friendly NFT infrastructure and APIs which help developers bring their applications to market in hours instead of months.

There are simply no dedicated NFT infrastructure purpose built for developers. NFTPort changes this through a one-stop, simple, developer-friendly NFT infrastructure and APIs that enable developers to build NFT-based applications — you can think of it as Stripe for NFTs. Before NFTPort, building an NFT application took weeks or months; now developers can do that in less than a day and ship products fast. Really fast. For example, before NFTPort it would’ve taken months to build a “Google for NFTs” — AI-based visual and reverse-image search for Ethereum and Polygon NFTs — but using our APIs it took only a few days.

‍

Simplicity is the Ultimate Sophistication

For example, to previously create an NFT programmatically, one had to learn Solidity, Web3.js, (+ a lot more complicated blockchain tech), maintain it and pay a lot of money for minting. Now, any developer/creator can turn anything into an NFT for free using a single API call.

‍

And as illustrated by our recent ProductHunt launch, people are absolutely loving it. In a few weeks, there are already 200+ builders in our Discord community creating new NFT experiences. Some of these include the NFT camera (already launched), Instagram for NFTs, Spotify for NFTs, game assets with NFTs, regular NFT collections, turning event tickets into NFTs, turning books into NFTs and many more. Most of the world sees NFTs by the lens which media portrays it which is mostly about speculation. We on the other hand, see so many creators bringing their dreams to life through NFTs and we are totally blown by the innovation that is happening in the space. The Great Migration is already happening as every day more and more Web2 builders realize that NFTs is a new world to be discovered which is not held back by today’s legacy monopolies — they join the movement to create this new world.

‍

The Reason Why NFTPort Exists — the New Data Layer for the Internet as an Open Protocol

You can’t connect the dots looking forward; you can only connect them looking backwards. So you have to trust that the dots will somehow connect in your future. — Steve Jobs

Coming to the realisation that Sentinel will fail was hard but looking backwards, it was one of the best things that happened to us because it set us up for the mission of empowering developers to break up today’s data monopolies and make the Internet ownable by citizens.

Although NFTs have found adoption only recently, the concept itself has a rich history going back over 40 years. In the 1980’s before WWW, Tim Berners-Lee tried to implement a standard similar to todays NFT’s which could later be part of the WWW. He did so again in the 2000s by advocating for the concept of linked data (i.e. the Semantic Web). These efforts were unfortunately not successful due to their sheer technical complexity and non-adoption by the industry. Instead, the biggest beneficiaries of Web2 became today’s data monopolies and the surveillance capitalism industry.

We thus believe that NFTs are short-term overvalued and long-term undervalued because the true opportunity lies in its ability to break up today’s data monopolies by reimagining the existing Web2 personal data model to a peer-to-peer one where we citizens own our personal data and ultimately our growingly digital lives. Tim Berners-Lee dreamt of an open data layer for the Internet and finally its promise is coming to life through NFTs. Our vision is to enable this by becoming the new data layer for the Internet as an open protocol.

‍

Roadmap to a Large Scale Public Infrastructure

In crypto, being transparent with communication, executing on the promises made and designing the right incentives are critical. That’s why we already share our roadmap with the community:

1. Adoption through a hosted service — Decentralization is one of our core values but one has to be smart about when is the right time to decentralize. Currently, our core focus is to move fast and bring real value to developers already today and not in the distant future. That’s why we decided to start with a hosted service on top of which projects can already build and create real value. We are still in beta but there are already a lot of projects building on our APIs. They range from a single person startups all the way to Ethereum’s leading projects such as POAP which has 200K+ users.
2. Community — We believe a new data layer for the Internet should be a public infrastructure which enables all stakeholders from developers, end-users, network participators all the way up to governments which form a large backbone of today’s Internet. Thus it should be inclusive and needs the community’s voice in its design, development, marketing, etc. to make a positive dent in the universe compared to serving only a few powerful groups which is often true with today’s Web2 monopolies.
3. Multi-chain support — The blockchain industry is moving multi-chain and developers need access to NFT infrastructure/APIs to move fast and onboard new users from other chains. We already support Ethereum, Polygon and will be adding support for other leading NFT blockchains.
4. Decentralized network — We believe the new data layer for the Internet can not and should not be controlled by one entity. Instead, the centralization pendulum has started to run its course and quoting Balaji Srinivasan we will most likely see a decentralized West and a more centralized East. That’s why, long-term NFTPort will become an open protocol where anyone can participate in the network by running a node to provide the necessary NFT infrastructure. This will be accompanied by an incentive aligned protocol design where consumers of the infrastructure will pay to the node providers for its usage.

‍

Why Join the Navy if You can be a Pirate

Wish to build and ship your NFT application in hours instead of days? Get a free API key from NFTPort, check documentation and start building.

Join our Discord community to be part of the mission, share your feedback, get support from our team and connect with fellow NFT developers.

In addition to growing our community, we are hiring. If you believe in decentralization and are good at infrastructure, big data, APIs, DevOps, smart contracts, full stack engineering & marketing then DM me in Discord or Twitter.

Finally, I’d like to thank all of our investors who have been believing and supporting us all along through thick and thin. Ragnar Sass, Martin Henk, Jaan Tallinn, Taavet Hinrikus, United Angels VC & Jaynti Kanani — we couldn’t be doing this without you.

For all the founders out there, when times are tough and you are in a battle then stop for a moment, pause and drown out the external noise because the answers will not come from there. Instead, look deep within you and listen to your heart — it already knows where you have to go.

— Johannes & the NFTPort Team

Follow NFTPort on Twitter

Join our Discord Community