Startin'blox | Data app framework for Ocean's ecosystem | Round 21

OceanDAO Round 21
#1

Project Name

Startin’blox

Project Category

Build & Integrate

Proposal Earmark

New Entrants

Proposal Description

Problem:

Data apps frameworks are not suited for web3, and therefore for the Ocean ecosystem. You can’t login using your wallet, standard web3 services are not supported natively and interoperability is not built in leading to centralization. On top of that, the most famous ones, namely Gradio and Streamlit don’t let data scientist support Python to do…

Solution:

We believe that there needs to be a user friendly data app builder on top of the Ocean ecosystem if we want apps to thrive. Startin’blox is proposing to build such a framework. We believe that a rich application layer would increase usage of both datasets and compute-to-data.

Street credibility:

We do not start from scratch, quite the contrary. We already have such an open-source low-code app builder for interoperable apps. It is used in production by 30 clients today, including the European Space Agency (ESA), providing us with close to 500k€ of annual revenue.

We already provide Self Sovereign Identity (SSI) to our customers as well as their own sovereign data storage. We want to go a step further and be able to offer all the web3 service in a couple of lines of HTML. Filling the gap in Ocean’s ecosystem by becoming the web3 native data apps framework, modular and interoperable, is totally in line with our current roadmap.

Financials:

Hours per day

8

Daily rate

$700

Hourly rate

$88

Days

5

Total hours

40

Hours per interview

2

# of interview

5

Total hours for interviews

10

Total cost per interview

$875

Hours per studied framework

5

# of frameworks studied

5

Total hours for interviews

25

Total cost per interview

$2,188

Total

$3,063

We are already in discussion with Algovera and DataUnion.

Our goal is to interview 7 projects of the ecosystem to understand their needs in terms of data app framework. Then we want to study what data frameworks on the market offer to understand what is missing and what additional features are required by the ecosystem.

Vision:

We strongly believe that the future of data is sovereign and decentralized. We are very excited by the “data lab” that is the Ocean ecosystem today, developping distributed machine learning, data unions, monetization, data NFTs and more.

We believe that a federated network of off-chain data storages with a low-code app builder on top is a nice complement to sustain that vision, and this is what we propose to bring to the table.

Grant Deliverables

The deliverable will be a detailed and priced roadmap of what needs to be built in order to meet Ocean ecosystem’s needs in terms of data app framework.

Project Description

Problem:

Data apps frameworks are not suited for web3, and therefore for the Ocean ecosystem. You can’t login using your wallet, standard web3 services are not supported natively and interoperability is not built in leading to centralization. On top of that, the most famous ones, namely Gradio and Streamlit don’t let data scientist support Python to do…

Solution:

We believe that there needs to be a user friendly data app builder on top of the Ocean ecosystem if we want apps to thrive. Startin’blox is proposing to build such a framework. We believe that a rich application layer would increase usage of both datasets and compute-to-data.

Street credibility:

We do not start from scratch, quite the contrary. We already have such an open-source low-code app builder for interoperable apps. It is used in production by 30 clients today, including the European Space Agency (ESA), providing us with close to 500k€ of annual revenue.

We already provide Self Sovereign Identity (SSI) to our customers as well as their own sovereign data storage. We want to go a step further and be able to offer all the web3 service in a couple of lines of HTML. Filling the gap in Ocean’s ecosystem by becoming the web3 native data apps framework, modular and interoperable, is totally in line with our current roadmap.

Financials:

Hours per day

8

Daily rate

$700

Hourly rate

$88

Days

5

Total hours

40

Hours per interview

2

# of interview

5

Total hours for interviews

10

Total cost per interview

$875

Hours per studied framework

5

# of frameworks studied

5

Total hours for interviews

25

Total cost per interview

$2,188

Total

$3,063

We are already in discussion with Algovera and DataUnion.

Our goal is to interview 7 projects of the ecosystem to understand their needs in terms of data app framework. Then we want to study what data frameworks on the market offer to understand what is missing and what additional features are required by the ecosystem.

Vision:

We strongly believe that the future of data is sovereign and decentralized. We are very excited by the “data lab” that is the Ocean ecosystem today, developping distributed machine learning, data unions, monetization, data NFTs and more.

We believe that a federated network of off-chain data storages with a low-code app builder on top is a nice complement to sustain that vision, and this is what we propose to bring to the table.

Final Product

The deliverable will be a detailed and priced roadmap of what needs to be built in order to meet Ocean ecosystem’s needs in terms of data app framework.

Value Add Criteria

We believe that there needs to be a user friendly data app builder on top of the Ocean ecosystem if we want apps to thrive. Startin’blox is proposing to build such a framework. We believe that a rich application layer would increase usage of both datasets and compute-to-data. 

Core Team

Benoît Alessandroni, CTO

https://www.linkedin.com/in/benoitalessandroni/

Benoît has been building decentralized and SSI powered apps for 8 years. He is a worldwide expert of the Solid standards, the most advanced specifications to build interoperable apps, with 30 apps deployed in production and several contributions to the W3C specifications.

Sylvain Le Bon, CEO

https://www.linkedin.com/in/sylvainlebon/

Sylvain is an entrepreneur for 12 years in the tech industry, focused on the decentralization of organizations. He co-founded the biggest freelancers collective in France called Happy Dev, which can be considered one of the first DAOs. He co-founded Startin’blox being convinced that interoperable tools were a crucial missing piece of a thriving decentralized ecosystem.

Louis Cousin, CSO

https://www.linkedin.com/in/louis-cousin/

Louis is a word class expert in cooperativism. He is one of the leading figures in the world cooperative movement. He is spearheading research on the impact of web3 in cooperativism. He convinced 10 international organization to give a try to decentralization, and is accompanying them in the process.

Alex Bourlier, CGO

https://www.linkedin.com/in/alex-bourlier-1725a83a/

Alex is an entrepreneur for 10 years in the tech industry, with the same focus as Sylvain. They work together for 8 years and therefore share the same curriculum.

Advisors

Philippe Honigman (philh), Mangrove

Alp Ergin, PrimeDAO

Funding Requested
3000

Minimum Funding Requested

Wallet Address
0x534958c32a3e282086a5D1737A320aE038D9deF6

#2

Project submitted deliverables:

Introduction

Several members of the Ocean ecosystem have shared their pain to access the data that they need within an intuitive interface. In our grant application for round 21, we had set out to understand what is missing in terms of data app framework so that projects of the ecosystem can easily build apps that grant them access to the data they need.

Interviews

We interviewed the following projects to understand their pains and their needs in terms of data sharing, processing and rendering. The interviews as presented here are my personal notes rather than actual transcripts of our conversations, as I thought it is much more clear and digestible.

Richard Blythman @ Algovera

Q. Do you need data from other members of the Ocean ecosystem?

A. Yes, a lot!!

Q. What are your needs in terms of data exchange within the Ocean ecosystem?

A. I need to be able to:

  • Access data from other projects
  • Run computations on those data
  • Display the results
  • Easily share the resulting module

Crowd sourcing data is a good summary of my need, and maybe integrate the result of my computations into something useful in your daily app.

In my wildest dreams, I’d have things like:

  • a DataUnions module
  • a Ocean Marketplace module
  • a Athena protocol module
  • a {name of your favorite Ocean ecosystem project here} module
  • a data processing module
  • a data rendering module

Those modules would be interoperable. I’d be able to easily chain them in whichever way makes sense to extract value from data + AI models.

There is no standard interface for datasets. On ML, text is becoming the interface.

We need standards for datasets otherwise we can’t train on several datasets.

Q. Why do you need that?

A. Save plenty of time

To improve the quality of the output as well, because less biased, because based on real world data

Q. Something else you want to add

A. I’d be happy to dig on that idea to integrate AI models into everyday apps. I think it would bring a lot of value to the end users and to Ocean.

Robin Lehmann @ Data Union

Q. Do you need data from other members of the Ocean ecosystem?

A. Hmmm… indirectly yes.

The job of a Data Union is to pool data to extract money out of it, sharing a part of the profit back to the data providers. We are a tooling solution to enable that business model so… if we can ease access to data, we bring more value to our customer.

Q. What are your needs in terms of data exchange within the Ocean ecosystem?

A. So far each Data Union we work with has its own “data supply”. Example: one of them works with healthcare professionals to collect health data, another one collects social networks data. At the moment we find mostly defy datasets on the Ocean Marketplace.

To sum up quickly what a Data Union does:

  1. Data collection
  2. Harmonize / label data
  3. Computation over data / Monetization
  4. Profit sharing

So yes in theory, in practice the real hurdle is consent. It is hard to convince individuals and companies to share their data. That’s what the Gaia-X ecosystem is trying to solve. Mats Broden is sharing that opinion.

We also have allergic reactions by some organizations which don’t want to be associated with the web3 space.

All that being said, interface components on top of the data we have once it is here are obviously welcome, but I couldn’t say exactly what yet, we need a bit more time to figure this out.

Q. Something else you want to add?

A. C2D on Ocean doesn´t take parameters at the moment. We need to make it evolve to become a bit more flexible.

Tom Husson @ Eden Protocol

Q. Do you need data from other members of the Ocean ecosystem?

A. We are very focused on talent and project data right now.

Q. What are your needs in terms of data exchange within the Ocean ecosystem?

A. Our goal is to help match the right project with the right person. We believe DAOs are the right playground for that, but it is probably applicable to any organization.

If we manage to have access to the members and projects data of the different members of the ecosystem, maybe via a job board of a members directory, we’d be interested. Right now we’ll keep our focus on DAOs.

Q. Something else you want to add

A. Not right now

Sal @ Algovera

Q. Do you need data from other members of the Ocean ecosystem?

A. Yes, and especially, we need to be able to have chainable interfaces components to be able to easily run computation on top of the data without having a PhD in Data Science.

We’ve combined ReactFlow with the Gradio data app framework to move in that direction. It is a proof of concept but it displays well where we’re heading.

Q. Why do you need that?

A. The solutions on the market are either not chainable, or not open source. We need both, plus integration with web3.

Q. Something else you want to add

Right now we focus on building the tooling. We are not yet at the step where we try to access and exploit any data, this will come.

Mats Broden @ Play Safe

Q. Do you need data from other members of the Ocean ecosystem?

A. Anything that can help my clients really. My job is to deploy sensors and to correlate the data they generate with other datasets to infer some insights about a given problem my clients are facing. It can vary a lot from one client to another so… I’m interested in pretty much everything. The more is accessiblem the more I can bring value to my clients.

Q. What are your needs in terms of data exchange within the Ocean ecosystem?

A. As Robin already told you, the big blocker is trust. Today “nobody” wants to share his/her data, especially not organizations. Everyone is wary about the consequences of sharing data they shouldn’t have shared.

Data NFTs and the whole infrastructure around them are heading in the right direction. We also need to educate people and to setup a legal and technical framework for data sharing that makes it obvious that sharing is the right thing to do, ideally economically.

Most of the time, people do not understand how their data could benefit others, nor what data sharing really means. For instance I have discussions with a Telecom company that have data relevant to many other actors, but they refuse to share it, and they can’t explain why they refuse, it is just a cultural thing. There is still a strong “protect your data” mentality.

Q. Something else you want to add

A. The road is long, we are heading in the right direction. There are other blockers before trying to work at the interface level I believe, even though interfaces are key to data sharing but it is not what is blocking me right now.

Nassim Dehouche @ DAOkit

Q. Do you need data from other members of the Ocean ecosystem?

A. Hmmm… a bit like Eden Protocol, if it is not project or talent data, then no for now.

Q. What are your needs in terms of data exchange within the Ocean ecosystem?

A. Our goal is to decentralize mechanical turk, we’ll focus on that.

Q. Something else you want to add

A. The interface layer on top of web3 is really not satisfactory and full of issues, so it is a nice angle to focus on that part of the stack.

Conclusion from the interviews

The interviews of the ecosystem didn’t reveal what we thought they would. We were expecting a strong need for data sharing among the different players of the Ocean ecosystem, and the necessity to have interface components to actually exchange data among the different projects.

50% of the interviewed projects, namely Algovera, Data Unions and Play Safe do express a strong need to access data, but what it means is very specific to each project, and we could not get them to converge on a set of interface components that would help them to do so.

We were still very convinced with Richard @ Algovera that something had to be done at the frontier with data and applications though. We decided to carry on with the market study of the different data app frameworks to try to identify what we thought was missing, and where we could bring value to the Ocean ecosystem.

Data app framework study

We studied Gradio and Streamlit, the two main data app frameworks, with the help of Sal and Richard from Algovera. Since we didn’t want to replicate the already available “feature comparisons” articles one can easily find online, we focused on why they are not satisfactory from a web3 perspective, and what needs to be added or modified to cover the Ocean ecosystem’s needs.

Large Language Models need low-code solutions

There is growing evidence from the AI community that we need to couple low-code and no-code solutions with the rest of the Data Scientist toolset [1][2][3].

Gradio is hard to embed in an interface

Gradio is only Python based. JS can’t easily be pushed into it.

This makes it hard to embed it in an interface or to customize it to do your own rendering.

It is not easily customizable. This kind of makes it unusable.

Streamlit apps are not web3 apps supercharged with AI

Streamlit apps miss the ability to interact with our apps. What we want is to be able to interact with real data, ideally live. We want Ai to become a commodity assisting our everyday interactions, not just something we demo in a sandbox context.

While Streamlit allows any dataset as an input, it fails at being integrated within our daily apps. Using Streamlit, you can’t summarize hours of discussions within your chat, you can’t generate privacy preserving avatars in your directory of members, nor can you transform a video call into a digest of 10 lines.

At that point of the study, we had several hours of discussion with Richard about what it means to have AI powered features easily assembled together in a low-code fashion. We came to the realization that this has never been possible before. It is the first time in human history that you can have apps that give you full control over your data, and run a trained AI model on top of your sovereign data to enhance the UX of your app.

For example:

  • Text to Avatar on a directory of members could allow users to retain their privacy while displaying an avatar that closely matches their persona
  • Text summarization can help busy chat members to stay up to date with the activity of the day in a few minutes
  • Audio to Text summarization can help anyone grasp the essence of several hours of meetings in a couple of minutes
  • Much more to come, AI is only just starting

We believe that these AI superpowered dapps can revolutionize the UX of our apps as well as benefit the Ocean ecosystem by creating new and exciting business models at the application level.

We decided to follow that path for the rest of the work we were set to do.

AI powered app builder

640×924 378 KB

We have all seen very impressive DALL.E 2 creations based on text inputs. What most of us failed to understand is that many AI models that are as impressive as DALL.E 2 are production-ready to assist us in our daily lives. In the same way that it is now possible to generate an image by providing a simple sentence, we can ask AI to summarize hours of meetings for us, to translate it in a foreign language and much more.

What matters is to have access to your data, and control over your app: two things web3 is striving to achieve.

Algovera and Startin’blox want to collaborate to provide the first live example of an AI powered feature anyone can add into her own app.

POC: a privacy preserving directory of members for OceanDAO

Startin’blox is a low-code app builder for interoperable apps. We have 15 ready made features that can easily be assembled in an app. The most deployed of these features among our 30 clients is the directory of members. 60% of our clients use it.

1600×1044 191 KB

With Algovera, we want to demonstrate that AI can now enhance our daily interactions. We believe that “text to image”, the DALL.E 2 feature, is the most impressive one. Therefore this is the one we want to integrate first, as we believe it is the one that will generate the most attention.

What we have in mind is to provide the OceanDAO members with a directory of members as a proof of concept of an AI powered feature. In that directory, people could upload their avatar, or they could choose to run an AI algorithm to generate one for them based on their profile.

Users would log on this members directory using their wallet. They could only create a user profile with a minimum of 500 $OCEAN on their wallet. On their profile edition page, they would have the option to generate an avatar using a generative image algorithm hosted on Ocean C2D infrastructure. Some $MATIC would be required for the algorithm to run, so transaction and community fees can be paid.

The roadmap below details what it would entail to develop such a feature. We designed it in a simple enough way so that it could fit Startin’blox’s round 22 ability.

Deploy a staging and a production environment to host a directory for the OceanDAO members

$700

Allow OceanDAO members to login with their wallets

$1,400

Block access to members having less than 500 $OCEAN on their wallet

$700

Deploy a generative image algorithm based on a trained AI model on OCEAN C2D

$1,400

Allow users having enough $MATIC to generate an avatar based the aforementioned algorithm

$700

Algovera’s mentorship

$1,500

Test, project management and deployment

$1,280

Total

$7,680

Such a roadmap could be completed by the beginning of November, and would only be the first steps towards a more elaborate project including the minting of NFT based on those avatars. We are just presenting the proof of concept as part of this deliverable.

#3

Admin:

Thank you for submitting your deliverables, they seem to match the outline of what was promised so I am accepting them.