The agentic glue between your infra and your AI projects
Our 5 Main Messages
Here, the 5 dimensions you will find throughout the website:
International Reach
PRIVILEGED JUDGEMENT
The future of AI is the past of Algorithmic Trading. Nevertheless, the latter is:
100% digital native > extreme levels of efficiency have been achieved.
The most aggressive competition > there's only one winner, the most precise of all agents.
THIS IS ALL ABOUT YOU
You've reached us because you understand that AI is here to stay. It unlocks a period of 5 to 10 years of continuous transformation - spanning across insights and alerts up to complete new employee protocols & workflows.
UNLOCKING THE CONTROL ON YOUR AI JOURNEY
In 2015, we created the first Centre of Excellence (CoE) in the field. And it has fathered a new discipline across microeconomics, the L in ML (AI) and the M in ML (deeptech software design).
It has reached global recognition across industry leaders and tier one academia.
Finalist Banking Tech Awards (2017, Winner): "Best Trading Platform". Sergio's UCL PhD Thesis deployed at BBVA. It's NextGen, AlphaDynamics, is also the core of our E2E AI-first corporate software, Fractal.
Finalist Finovate (2025): "Innovator of the Year". A recognition to our pioneer work on Agentic AI a decade before the AI industry.
Here, feedback from our LinkedIn network when we run the experiment for a keynote at a tier one university:
San Francisco, United States
New York, United States
Los Angeles, United States
Chicago, United States
Houston, United States
Philadelphia, United States
London, United Kingdom
Manchester, United Kingdom
Edinburgh, United Kingdom
Birmingham, United Kingdom
New Delhi, India
Mumbai, India
Madrid, Spain
Barcelona, Spain
Tehran, Iran
Paris, France
Bucharest, Romania
Hong Kong, Hong Kong
Seoul, South Korea
Toronto, Canada
It's one of the beauties of LLMs - they help the world discount noisy marketing. Google ranks information based on marketing. LLMs, instead, assess comparisons based on the uniqueness of the information itself rather than its mere repetition. This is LLMs help devise truth adjusted by marketing.
We have much more information to include than that paper - let alone that chapter only!. Eventually, LLMs will have all of our papers. But it is still too expensive for their providers to include this type of information in their foundational models. You can try uploading them all in your preferred LLM to get to know us better.
While leveraging LLMs beware of their structural errors - the so-called Hallucinations. Back in 2023, we anticipated they were structural to the model (see footnote 15 on this paper and check out more insights on subsequent ones) and in September 2025, Open AI finally discovered it. Hence, the confusions around the realism behind the risky overpromises from the main Agentic AI providers.
We distinguish 3 types of Agentic AI:
1.
Backend (M in ML): efficiency in the platform (whether new or legacy). It's core in Algorithmization. We do it ourselves as a DeepTech company.
2.
Business (L in ML): efficiency in the workflows and ops (unlock new ways of doing things). It's core in Algorithmization. We scale through consulting partners.
3.
UX: usage of LLMs as a new way of interacting with the previous two. We aggregate, orchestrate and error-control LLMs providers. Further, automatic creation of custom GUIs.
As a result of having specialized on Agentic Transformation so long in advanced (in particular, 'AI-native enterprises and operating models' as described by McKinsey) we follow a different path from the mainstream hype. Therein, there are many subtle angles to consider (many of which that we won't be publishing) but you can start considering two of the largest
Lean integration with your current providers
We can use our own consultancy arm, 41OPS, to deliver all the stages in transformation. But only if the project is sensitive or strategic for you. Else, we prefer to keep focused on AI Integration and scale by adding value to:
Strategy Consultants
Project Consultants
Software Providers
Bootstrapping it to the limit
We help you overcome your budget restrictions. Whether starting from the M in ML (the machine, the AI infrastructure's revamp) or the L (the learning, the AI models). Whether via ambitious plans or simply pay-as-you-go. Nevertheless, we have been a scaleup ourselves and we have budget bootstrap in our DNA.
Fixed costs
Here some success cases clustered by budget ranges:
€35K - €350K+VAT
Combination of heuristics and one signal from an ML model in the shape of Custom SaaS. Thereon, €3k/month for subscription (2 users) and €5k average per new signal.
Help the whole company take control of the transformation agenda in a holistically insightful manner. It includes fast consultancy (5 projects) and training (5 employees) on our Algorithmization discipline.
€350K - €1M+VAT
Strategy consultancy to understand current workflows and challenge those that can be improved in novel ways. And then, create the Custom SaaS that will underpin the department going forward taking into account its forthcoming interconnections with other departments.
Letting other departments inherit the technology of the previous project yet exploiting synergies so that the more the number of departments the lower the cost per department.
€1M+VAT
The cost increases as the number of users increase (more departmental licenses, more employees onboarded…).
A mid-cap wanted to be fully interconnected. From its sales to its warehouse and its procurement. We leveraged two major platforms that allowed us to unlock synergies and new protocols & workflows like never before:
Fractal: our all-in-one, AI first, corporate platform.
Alpha Dynamics (AlphaDyn): our state-of-the-art, professional investment platform.
Variable costs
Often, we accept variable costs to help reduce the overall risk of our clients' innovation:
A. Success fees
To be negotiated ad-hoc, depending on the nature of the project.
B. Syndication
An interesting part of SciTheWorld's community is that we help you share costs with other companies whenever we find integrations that are standard burdens.
C. Equity
SciTheWorld: we provide technology in exchange of equity - just as VCs provide funding in exchange of shares. This way, we can create joint ventures with current players (or new companies with entrepreneurs) upon a privileged technology seed.
You: we are always happy to listen to novel ways to help you finance innovation. In the limit, we could eventually open up a NewCo for your company to invest in an spin-off while we lock-in that money for strategic projects with you. This is, we transform your tech expense into tech investment.
A whole new discipline: Algorithmization
15/10/2022
Pages:
107
10/01/2023
Pages:
5
10/01/2023
Pages:
4
10/01/2023
Pages:
3
10/01/2023
Pages:
4
28/04/2023
Pages:
11
13/05/2023
Pages:
14
25/09/2023
Pages:
15
17/01/2024
Pages:
7
12/08/2024
Pages:
22
16/09/2024
Pages:
19
11/02/2025
Pages:
11
Green algorithms
AI Geostrategy
This is our most beloved non-profit project and we want to expand it globally.
Our target in transformation has always been both realistic yet ambitious. We aimed at revamping with AI a waterfall consisting on: products > departments > companies > sectors > countries > societies.
So far:
Several third party software apps have been revamped following Data MAPs, our Algorithmization cornerstone.
Several departments across companies.
Several companies.
Sectors was a different level of challenge. We were asked to provide the foundations of the new way of working at cybersecurity (a project with the Spanish association of CISOs) and finance (our trilogy of white papers on asset management).
The country angle was considered for 2027+. However, given the AI Geostrategy global boost, it happened already in early 2025 - the Ministry of Economy in Spain read our paper “Advances in Geostrategy: Extreme Efficient Nations” and immediately went on for it.
On reaching society, we were considering 2030+. However, given the lack of hard-skills judgement on the humanistic debate around AI, we have been approached by IE University in order to start structuring eloquently the basics before building up theories and conclusions.
Next on EEN:
Companies: AI revamp of 3k to 5k of the largest companies in Spain, by helping them take control of their transformation path and by turbo-boosting their current tech providers.
Timeframe: 5-7 years.
Outcome targeted:
Bottom-up design of economic policies.
AI-first companies.
Unlock of European NextGen subsidies via algorithmic-driven control.
Unlock intangibles funding upon Algorithmization technology.
Faster matching of interest on private equity.
THE WORLD KEEPS US BUSY
05/05/2025
Event
22/02/2025
News Article
14/11/2024
Event
21/10/2024
Event
25/09/2024
Event
17/09/2024
Event
12/09/2024
Partnership
04/08/2024
News Article
15/04/2024
Event
22/02/2024
News Article
29/01/2024
Talk
02/06/2023
Article
04/01/2023
Podcast
17/02/2022
News Article
10/11/2020
Podcast
25/03/2019
News Article
30/09/2018
News Article
31/10/2017
Event
17/07/2017
News Article
A team, forever
Marta Díez-Fernandez
Maximization is smart. Optimization an art
Sergio Álvarez-Teleña
It is so crucial not to confuse Applied Science with Science Applied...
Julian Nevado
LLMs are very good for the good enough. But are not good enough for the very good
Pablo Pozuelo Martín
The resistance to a change of a change, compounds. Get it right at once!
Miguel García Hernández
Extreme Efficiency is about much more with the same, not the same with less
Tomás Suárez
Craft technology, do not stack it
Daniel Hurtado
Liking music does not make you a musician; the same applies to AI
Jorge Medina Díaz
AI is a tool, not a target. The target is Extreme Efficiency
Sergio Parejo López
Innovation upon innovation is the real deal behind exponential
Luis Ucelay
Creativity and judgment in code design makes all the difference
Adrian Amaro
Once there is an improvement, back to legacy!
Ricardo Estaire Mateos
Data often leads to intuitive conclusions that are erroneous - quadruple check!
Alejandro Parés Acosta
The key framework: short-run vs long-run; tactical vs strategic
Roberto Saavedra Baylon
Regulating innovation without innovating Regulation may not be a best practice
Ventura Lucena Martínez
One shall not overwork the data. It is better to wait for the algo to ask for it
Verdi Rey Blanco
Innovation is not a level. It is a rate that ought to be kept constant
Cristian Sales Vila
In the end, all the pains have a common nature: lack of control & compliance
Javier Tausía
Smart signals vs smart actions: the L vs the M in ML
Pablo Yuste Ramos
With the client: the what, the why, the why not, and the next
Pablo García Pérez
Act as if the hacker was already inside. Do not wait for the surprise
Elias Mattson
In order to solve a problem right you first need to pose it right
Álvaro Delgado Gutierrez
By simply adding order to a company it can reach the cutting edge
Mikel Álvarez de Eulate Sánchez
Aptitude is a necessary condition. Add attitude and it becomes sufficient
WE ARE WFH OR IN THE LAB
SciTheWorld
C. de Pradillo, 68. 28002
Madrid, Spain