Thursday, 09/04/2026 (NZ time) was the longest day. I couldn't make a
single mistake, so I turned off the phone and the internet to concentrate
from 8am to 9pm, then slept like a log. I was a walking zombie yesterday.
I'm glad this will never need to be done again.
This is the successful culmination of months of mentally challenging
preparatory work for the new Pipi Core data centre.
A typical day for me is
50% learning
40% thinking
10% doing
The less I do, the more productive the solutions. The faster the
progress.
Speed will come from Pipi-driven automation, not Mike working harder.
Mike is the inventor and architect of Pipi and the founder of Ajabbi.
Inside the Pipi Core data centre, the careful migration and
reorganisation of hundreds of thousands of Pipi-related files began and were
completed yesterday. Each file or directory was placed in only one Nestspace, and then the historical Nestspaces were converted into zip archives
It is a good solution, but it took a month of trial and error to figure
out. It had to be 100% correct to enable rapid, reliable data centre
automation, self-managed by Pipi.
It was fascinating to watch the recent Lex Fridman interview with
NVIDIA CEO Jensen Huang, who described NVIDIA's large-scale
problem-solving process.
Problems to solve
A large number of problems had to be solved in parallel. Each problem
affected the others.
Where to put self-organising Pipi swarms.
Pipi instances
Naming
Self-evolving
Versioning
How to use DevOps automation with all of the above.
The knock-on effects on
Namespaces
Backups
Replication
Accounts
Workspaces
i18n
Web URLs
Developer documentation
Training
etc.
Securing 100% security and privacy
Cross-platform and environment portability
This has led to changes in the underlying Pipi
System Engine (sys) data model, which I will write about
tomorrow.
Nestspace
The fundamental organising principle is to use a uniquely named directory,
now named as a "Nestspace".
Now, I can start configuring the files and settings for each
Nestspace. The necessary code has already been successfully tested.
Eventually, this process will be completely automated.
Application.cfc
server.xml
Datasources
OS environment
Java
Application server
Cloud platform
REPL
Speed is King
Then, each Pipi will be back in business and can be left running 24x7 in
its own Nestspace, thereby increasing DevOps speed by at least
10x. The priority now is to increase Pipi's Data Centre speed by
1000x using automation and keep going. So what previously took a year can be
done in an hour. The over-optimistic complete deadline is June 2026.
But nothing happens as expected
It is hard to make completion predictions when this architecture is completely novel, and everything is uncharted territory and hellish difficult. Everything is a back-of-the-envelope guess. I'm slowly getting there ...
"like a man riding a drunken donkey facing the tail, two steps forwards, one step backwards"
... enjoying the whole journey and it's a hell of a lot of fun.
Future customers
The increase in speed will also directly benefit all future customers using
the SaaS workspace applications. Deployments, configuration, and
updates will also get the same
1000x speed increases at no extra cost.
I was greatly influenced by a recent article by Gennaro Cuofano in The Business Engineer about how Apple ensures Privacy.
Gennaro wrote ..."
Apple’s response is not to win training. It is to dominate inference
Apple’s strategy is internally coherent:
Tier 1: On-device inference
Small local models handle personal and contextual tasks.
These run without network dependency and with minimal privacy leakage.
Tier 2: Private Cloud Compute
Apple Silicon-based servers handle workloads beyond device capacity.
The architecture is stateless and privacy-preserving.
Tier 3: Third-party frontier models
Apple relies on external model providers such as Google and OpenAI for world knowledge and advanced reasoning.
These models are treated as backend commodities underneath Apple’s interface layer.
..."
Update 18/06/2026
Use Data Diodes. Prices start at $5K
A Data Diode is a physical device that allows data to travel only in one direction.
How it works: Inside the device, a fibre-optic LED transmitter sends light to a receiver on the other side. There is no return fibre cable physically.
Security benefit: It is physically impossible for a hacker to send a command back or steal data through a write-only data diode because the hardware cannot transmit in reverse.
Home > Ajabbi Research > Library > Subscriptions > The Business Engineer
Home > Handbook >
Last Updated
30/07/2026
Pipi three-data-centre model revisited
By: Mike Peters
On a Sandy Beach: 29/03/2026
Mike is the inventor and architect of Pipi and the founder of Ajabbi.
Data Centres
Because of its unusual architecture and the priority it places on privacy and security, Pipi needs three separate data centres that work together in a chain.
Rendering > Staging > Cloud
Rendering: Build enterprise applications.
Staging: Send updates, localise and deploy enterprise applications.
Cloud: Hosting of enterprise applications for scaling and integration.
I like the way Apple ensures privacy for people using AI on their iPhone. Any extra AI work is offloaded to Apple's cloud servers for processing, which are stateless and store nothing. That got me thinking.
Could part of Pipi be made stateless to add extra privacy like Apple?
The way Pipi is designed;
Rendering: Unable to see any customer data.
Cloud: Unable to see any Pipi config data.
This is to be done by using separate databases hosted in separate data centres.
Possible data centre model
Rendering: This is stateful; each agent-engine has its own database. Each enterprise customer has a separate physical server to host a digital twin.
Staging
Inwards-server: Receive anonymised logs
Outwards-server: Send updates, localise and deploy enterprise applications.
Cloud: This would be stateful using customer-eyes-only databases.
Would that work? How?
Network Security Design
It seems that maximum security can be achieved by using physical "air-gaps" initially, followed bydata diodes. Use shielding and Faraday Cages to protect against EMR and acoustic leakage. Disable or remove Bluetooth and WiFi.
Home > Ajabbi Research > Library > Subscriptions > Simons
Foundation
Home > Handbook >
Last Updated
19/03/2026
These New AI Models Are Trained on Physics, Not Words, and They’re Driving
Discovery
By: Elizabeth Fernandez
Fatiron Institute: 09/12/2025
Elizabeth Fernandez is a science writer specializing in science and
society, science and philosophy, astronomy, physics, and geology.
While popular AI models such as ChatGPT are trained on language or
photographs, new models created by researchers at the Flatiron Institute and
other members of the Polymathic AI collaboration are trained using real
scientific datasets. The models are already leveraging the knowledge they
learn from one field to address seemingly completely different problems in
another.
The Walrus AI model simulates fluid motion.
Walrus/Polymathic AI
While most AI models — including ChatGPT — are trained on text and images,
a multidisciplinary team of scientists has something different in mind: AI
trained on physics.
Recently, members of the Polymathic AI collaboration presented two
new AI models trained using real scientific datasets to tackle problems in
astronomy and fluidlike systems.
The models — called Walrus and AION-1 — are unique in that
they can apply the knowledge they gain from one class of physical systems to
seemingly completely different problems. For instance, Walrus can tackle
systems ranging from exploding stars to Wi-Fi signals to the movement of
bacteria.
That cross-disciplinary skill set is particularly exciting because it can
accelerate scientific discovery and give researchers a leg up when faced
with small samples or budgets, says Walrus lead developer Michael McCabe, a
research scientist at Polymathic AI.
“Maybe you have new physics in your scenario that your field isn’t used to
handling. Maybe you’re using experimental data, and you’re not quite sure
what class it fits into. Maybe you’re just not a machine-learning researcher
and just can’t burn the time working through all the possible models that
might fit your scenario,” McCabe explains. “Our hope is that training on
these broader classes makes something that is both easier to use and has a
better chance of generalizing for those users, as the ‘new’ physics to them
might be something another field has been handling for a while.”
Using cross-disciplinary models can also improve predictions when data is
sparse or when studying rare events, says Liam Parker, a Ph.D. student at
the University of California, Berkeley, and a lead researcher developing for
AION-1.
The Polymathic AI team recently announced Walrus in a preprint on arXiv.org
and presented AION-1 on Friday, December 5, at the NeurIPS conference in San
Diego.
Walrus and AION-1 are ‘foundational models,’ meaning they’re trained on
colossal sets of training data from different research areas or experiments.
That’s unlike most AI models in science, which are trained with a particular
subfield or problem in mind. Rather than learning the ins and outs of a
particular situation or starting from a set of fundamental equations,
foundational models instead learn the basis, or foundation, of the physical
processes at work. Since these physical processes are universal, the
knowledge that the AI learns can be applied to various fields or problems
that share the same underlying physical principles. Foundational models have
a host of benefits — from speeding up computations to performing well in
low-data regimes to finding physics shared across different fields.
AION-1 is a foundational model for astronomy. It is trained on data from
astronomical surveys that are already massive in their own right: the Legacy
Survey, the Hyper Suprime-Cam (HSC), the Sloan Digital Sky Survey (SDSS),
the Dark Energy Spectroscopic Instrument (DESI) and Gaia. All in all, that’s
more than 200 million observations of stars, quasars and galaxies totaling
around 100 terabytes of data. AION-1 uses images, spectra and a variety of
other measurements to learn as much as it can about astronomical objects.
Then, when a scientist obtains a low-resolution image of a galaxy, for
example, AION-1 can extract more information about it, learned from the
physics of millions of other galaxies.
Walrus’ domain is fluids and fluidlike systems. Walrus utilizes the Well —
a massive dataset compiled by the Polymathic AI team. The Well’s data
encompasses 19 different scenarios and 63 different fields in fluid
dynamics. All in all, it contains 15 terabytes of data describing parameters
such as density, velocity and pressure in physical systems as wide-ranging
as merging neutron stars, acoustic waves and shifting layers in Earth’s
atmosphere.
Such foundational models can be powerful. AION-1 and Walrus can utilize
physics seen in a different case and apply it to learn about something new.
It is similar to our senses. “Multiple senses together — rather than one at
a time — gives you a fuller understanding of an experience,” the AION-1 team
explained in a blog post about the project. “Over time, your brain learns
associations between how things look, taste and smell, so if one sense is
unavailable, you can often infer the missing information from the
others.”
Then, when a scientist is performing a new experiment or observation, they
have a starting point — a map of how physics behaves in other similar
situations. “It’s like seeing many, many humans,” says Shirley Ho,
Polymathic AI’s principal investigator and an astrophysicist and machine
learning expert. Ho is a senior research scientist at the Flatiron Institute
and a professor at New York University. When “you meet a new friend, because
you’ve met so many people before now, you are able to map in your head …
what this human is going to be like compared to all your friends before,”
she says.
Foundational models make scientists’ lives easier by streamlining data
processing. Scientists will no longer have to create a new framework from
scratch for every project or task; instead, they can start with an already
trained AI to use as a foundation. “I think our vision for some of this
foundation model is that it enables anyone to start from a really powerful
embedding of the data that they’re interested in … and still achieve
state-of-the-art accuracy without having to build this whole pipeline from
scratch,” says AION-1 lead researcher Parker.
Their goal is to make tools that scientists can use in their day-to-day
research. “We want to bring all this AI intelligence” to the scientists who
need it, Ho says.
Other Highlights From the NeurIPS 2025 Conference
CosmoBench: CosmoBench is a multiview, multiscale, multitask
cosmology benchmark for geometric deep learning. Curated from the
state-of-the-art cosmological simulations, CosmoBench is the largest
benchmark of its kind, with over 34,000 point clouds and 25,000 directed
trees. CosmoBench features challenging evaluation tasks from cosmology and
diverse baselines, including cosmological methods, simple linear models and
graph neural networks. This presentation will show how CosmoBench is pushing
the frontiers of cosmology and geometric deep learning.
Lost in Latent Space: Physicists model and predict the behavior of
physical systems using their understanding of the laws of physics. However,
these calculations require significant computing power. Flatiron Institute
scientists and other members of the Polymathic AI collaboration studied
whether a less taxing form of computing can still yield accurate results.
Known as ‘latent diffusion modeling,’ this computational model utilizes
artificial intelligence to generate high-quality images at a lower
computational cost while accurately capturing physical behavior.
Neurons as Detectors of Coherent Sets in Sensory Dynamics: Our
perception of touch, taste, sight and pain is mediated by neurons that carry
signals from peripheral receptors to the brain. This work shows that these
neurons can be understood as detecting ‘coherent sets’ within the sensory
stream — groups of stimulus trajectories that evolve together over time and
therefore share a common past or a common future. By distinguishing these
coherent sets, some neurons predominantly encode what has just occurred,
while others reliably signal what is likely to happen next. Traditional
classifications of sensory neurons can thus be reinterpreted as reflecting a
division between past-focused and future-predictive processing.
Understanding how the nervous system separates and transforms sensory input
in this way may offer new routes for treating mental illness and may also
guide the development of biologically inspired artificial
intelligence.
Predicting Partially Observable Dynamical Systems: Scientists can
predict the motion of a falling object or the evolution of fluids using
deterministic models that compute a single future outcome from past
observations. But this approach breaks down for physical systems where much
of the state is hidden. A prominent example is the sun: We can observe the
activity on its surface, but the processes deep inside remain largely
invisible. Without access to those internal conditions, there isn’t enough
information to forecast a single ‘correct’ future. Researchers at the
Flatiron Institute, together with collaborators in the Polymathic AI
project, have developed a probabilistic approach that can infer these hidden
solar processes. By incorporating information from the distant past into a
diffusion-based generative model, their method produces an ensemble of
plausible futures, offering a clearer understanding of how past sunspot
activity shapes its future evolution.
This is the revised Pipi roadmap now that the data centre is
running. The recent Ajabbi Research report, "The Workspace Issue," has also been revised to reflect these changes.
Update 25/03/2026
A very nice chap from Google contacted me to assist with applying
to join one of the Google AI Accelerators. I started the
application, then stopped when I realised that the 2 years of support and
generous free credits started as soon as it was approved. I need to complete all Stage 1 steps in the roadmap below,
and the Stage 2IaC connection to the
GCP free tier to deploy Pipi open-source before
applying, to make the best use of the opportunity.
I'm requesting support
from Google DeepMind to experiment with
connecting Pipi via MCP to DeepMind Gemini and
to find a way for Pipi closed-core to use Google TPU.Stage 3 will be highly experimental with unexpected
results.
Pipi is a non-generative multi-agent system with no tokens. It doesn't
need tokens to work and its 27 layers deep so far and counting. I expect it to get very
barnacled and crusty over time.
Update 31/03/2026
Setting up data centre automation has revealed that workspace deployments
need to be performed in this order of account types due to the permissions
cascade.
Agent Accounts (to admin Pipi)
Researcher Accounts (to edit UoM, ontologies and physical
laws)
Mike is the inventor and architect of Pipi and the founder of Ajabbi.
The long-planned migration of Pipi 9 to its own data centre has been completed. It took 2 weeks to execute. The existing setup was split into an office network connected to the internet and an isolated data centre that is not connected to the internet.
Starting from zero
The initial data centre consists of a single 45U rack and some other shelving, with mainly older equipment. It will do for a start and can grow as more racks are added, equipment upgraded, and more servers are added, etc.
External hard drives being used in the shift
Issues
Terabytes of data on backup hard drives to shift
Clean reinstalls of many operating systems
14 machines to configure
Adobe CS4 does not like Windows 11
Making do with what is available now
Go slow, think twice and get there faster
Opportunities
Pipi on 24x7x365
All systems can be turned on using multiple servers
Mike is the inventor and architect of Pipi and the founder of Ajabbi.
Long planned, Pipi 9's migration to its own data centre is now underway. The job has top priority and should be easily completed in a week.
First, the existing Ajabbi computer network was split into two.
A small start on the Pipi Core Data Centre to be built out over time, with an attached Mission Control. It is completely isolated from the internet, with all wifi and Bluetooth disabled. It is largely housed in the first 45U rack and some open shelving.
The start of the Ajabbi Office Network. It is connected to the internet for email, Zoom/Meet/Teams calls, office work, writing documentation, minor development, testing, graphics, video editing, office servers, accounts, printing, phones, etc.
Racks
45U Rittal welded rack frames will be added as required. Each with the following standard modular fitout.
UPS
Individually switched PDU
Network switches
Monitoring
Servers
NAS
Cooling
DevOps
Developer laptops attached to the isolated rack for server work have the following minimal developer stack.
Acrobat
BoxLang
CFML Server Dev edition
DBeaver
Dreamweaver
GeoServer
Grammarly
JRE 21
MS Access
MS Excell
NoteTab Light
PostgreSQL+PostGIS
Protege
Python
QGIS
VS Code
etc
The servers have a very different stack.
Initially, the data centre will often be turned off. It will be turned on to allow hundreds of Pipi Agents to run batch jobs autonomously. Eventually, many racks will be running 24x7x365.
Moving Pipi 9 to a data centre now opens the door to Pipi 10 in 2027.
Mission Control
The future Mission Control UI, connected to the data centre, could be shared via a live video feed of a monitor, ensuring security against hackers. It could even be a YouTube Live Stream of probes for those who don't like to sleep. 😀 Though my cat tells me it is much better to watch squirrels on YouTube. 😸😹😺😻😼😽
I'm looking for a solution to a problem. Some working notes.
"The first Data Centre
Once scaling begins, Pipi 9 will then need a data centre to use as a render farm to automatically create customised SaaS enterprise applications based on user requirements. The data centre will be completely isolated from the internet to maximise security. It can be expanded in stages if it is planned appropriately.
Each industry and each enterprise customer will get a dedicated server to store a mirrored copy of their deployment configuration and parameters, including localisation. No user data will be stored." - On A Sandy Beach
Mike is the inventor and architect of Pipi and the founder of Ajabbi.
The future Pipi 9 data centre must be fully automated and physically isolated from the internet.
Power
19" rack-mounted intelligent PDU will distribute power to servers.
According to Server Room Environments
Intelligent Rack PDUs
Intelligent PDUs add a level of sophistication to power distribution within IT server racks and include:
PDU and Outlet Metering: metered PDUs provide power-related information locally and remotely, which can include Amps (A), Volts (V), Frequency (Hz), Watts (W), Energy (kWh) and Power (kVA). The power readings can be for individual outlets and the total PDU usage. The information can be used by IT managers for capacity planning, the prevention of circuit overloads, client and cost centre billing (with +/- 1% accuracy) and efficiency calculations, including Power Usage Effectiveness (PUE).
Switched and Outlet Switched: a switching PDU will include metering and add the ability to remotely control (ON/OFF) to the PDU and the outlets. Outlet-switched PDUs provide a way for IT and data centre managers to remotely reboot or power down connected loads and allow for cascading power-ups to manage load inrush currents. Switching PDUs adds a layer of security to a server room or data centre power plan in terms of controlling unauthorised access to rack-level loads and their power connections. Costs are also reduced, as an onsite engineer visit is removed in most instances.
Remote IP Monitoring: the PDU provides connectivity for remote monitoring and can include HTTP/HTTPS, iPV4 and iPV6, Telnet, SSH, Virtual Serial, SNMP (v1, v2c, v3), JSON-RPC, LDAP, FTP/SFTP and RADIUS for secure login. The monitoring provides a way to view the status of the PDU and its individual socket outlets using a browser, monitoring software or a data centre infrastructure management (DCIM) software suite. The PDU may also offer a RESTful API for bespoke communications applications. Dual Ethernet ports can provide communications redundancy, and the communications module will typically be a ‘hot-swap’ type.
This means that each individual power outlet can be remotely switched on and off via IP. That could be controlled by another server.
A server can power up automatically when the power outlet is turned on. The server can be made to run a program (Pipi 9), do some work, create a backup, and then shut down.
My question is:
"Could the PDU detect that the server has shut down and then turn off the power outlet?"
According to a post on the Schneider Electric forum, this can be done with scripting using Net-SNMP.
Net-SNMP
It runs on Linux and Windows and is open-source.
"Net-SNMP is a suite of software for using and deploying the SNMP protocol (v1, v2c and v3 and the AgentX subagent protocol). It supports IPv4, IPv6, IPX, AAL5, Unix domain sockets and other transports. It contains a generic client library, a suite of command line applications, a highly extensible SNMP agent, perl modules and python modules." - Wikipedia
Pipi 9 in production
Pipi 9 is large but also very power-efficient (it's not an LLM). Each copy of Pipi 9 requires its own server. In this data centre, each enterprise customer has its own backend server running a customised Pipi 9 as a digital twin. I'm experimenting to see whether a small-form-factor refurbished PC could do the job.
There would need to be hundreds of these PCs in racks, autonomously coming online and offline as required to run batch jobs. Massive redundancy is provided by having spare PCs synced, NAS storage, VMs, etc.
This means that single-phase PDUs can be used. Maybe 1 PDU per 10-15 PCs per shelf, with only a few PCs running at any given time. 1 UPS per cabinet at the bottom, supplying multiple shelves.
Mechanical air-lock
Data needs to pass between this isolated data centre and customer deployments hosted in the cloud. My thought is to add a staging area (or several) between the two and pass data back and forth via network switches that are mechanically cycled (analogue). A bit like a double air-lock in space.
This would make it impossible for an external attacker to breach the system.
Robots in control
Pipi 9 is designed to serve as the system administrator for all customer cloud deployments. Coming on only as needed and in quiet times to minimise any disruption.
Number of data centers worldwide 2025, by country or territory
By:Petroc Taylor
Statistica: 19/11/2025
Petroc Taylor is a researcher with Statista's Technology and
Telecommunications team. His research focus is global developments in the
use of data, including trends in big data, analytics, and storage, as well
as the impact of emerging data technologies across industries and sectors.
He also supports the team's coverage of operating systems,
telecommunications, and the technology industry in Africa..
As of November 2025, there were a reported 4,165 data centers in the United
States, the most of any country worldwide. A further 499 were located in the
United Kingdom, while 487 were located in Germany.
What is a data center?
Data centers are facilities designed to store and compute vast amounts of
data efficiently and securely. Growing in importance amid the rise of cloud
computing and artificial intelligence, data centers form the core
infrastructure powering global digital transformation. Modern data centers
consist of critical computing hardware such as servers, storage systems, and
networking equipment organized into racks, alongside specialized secondary
infrastructure providing power, cooling, and security.
AI data centers
Data centers are vital for artificial intelligence, with the world’s
leading technology companies investing vast sums in new facilities across
the globe. Purpose-built AI data centers provide the immense computing power
required to train the most advanced AI models, as well as to process user
requests in real time, a task known as inference. Increasing attention has
therefore turned to the location of these powerful facilities, as
governments grow more concerned with AI sovereignty. At the same time, rapid
data center expansion has sparked a global debate over resource use,
including land, energy, and water, as modern facilities begin to strain
local infrastructure.
A stable platform is required for the isolated Ajabbi Research Data Centre, which is not connected to the internet. The platform O/S needs to be stable and not require weekly updates for Pipi to run on. Ideally, the platform and the major Pipi version would be updated simultaneously every 2-5 years.
Mike is the inventor and architect of Pipi and the founder of Ajabbi.
I like using Windows. It's because I'm very visual and need a GUI.
I have no problem with Linux, but that would require others who can spell and are familiar with the command line.
Pipi can use the command line autonomously because every command is carefully loaded into a database, ready for use in a CFML template.
The biggest challenge with using Windows O/S in the isolated data centre is dealing with O/S updates. It turns out that Microsoft offers a version of Windows specifically designed for embedded applications, such as those found in hospitals and factories. Perhaps that would be a better option for creating initial platform stability. I need to do some experiments.
Later, as Ajabbi grows and more people are involved, switching to a minimal version of Debian Linux could be easily done. Maybe as part of the next major version, Pipi 10.
LTSC
It seems to be slimmed down.
Doesn't have CoPilot, OneDrive, the Microsoft Store and Cortana.
I sent draft notes to Ortus Solutions yesterday to discuss future use of
BoxLang. I then edited the notes and shared them with the growing Pipi
community. I will add the answers as more is discovered about what
needs to be done to make it possible for Pipi 10 to use BoxLang.
Training is available on BoxLang Academy.
Can you spot any errors in my assumptions? Do you have other ideas or queries?
Contact me.
Update 04/06/2024, 05/06/2025
Cristóbal Escobar Henríquez from Ortus Solutions provided excellent answers to
the questions sent via email. These answers have been added to the
questions at the end of this article, which are slightly different
from the original ones (all my fault). Thanks :)
Update 09/06/2025
Luis Majano from Ortus Solutions provided additional answers to my follow-up
questions, which I sent by email. They have been added to the post
below. Thank you, Luis.
Update 12/06/2025
Luis Majano from Ortus Solutions provided additional answers to my follow-up questions, which I sent by email. They have been added to the post below. Thank you, Luis.
Update 13/06/2025
Luis confirmed that BoxLang cannot use CommandBoc modules. It appears that BoxLang can currently handle 90% of my needs. The only real gap is that Pipi cannot currently autonomously configure BoxLang itself via the command line, which is needed for scaling.
This is sufficient to make a start, with the hope that Ortus will address these issues in the future.
I expect Pipi 10 to be a 12-month build after Pipi 9 is in full production.
Update 14/06/2025
There will be an online meeting with people from Ortus next week to discuss and confirm details regarding the migration plan, paid support, and sponsorship of Ortus' open-source.
Mike is the inventor and architect of Pipi and the founder of
Ajabbi.
Draft roadmap
This is a draft, so expect it to change
This roadmap is probably for the next 2-3 years at least
The finalised plan will be published for the Pipi Community.
No money is available right now.
A plan is needed to avoid rework.
Don't expect Ortus to do work for free.
Realistic expectations of what is possible.
Good income will come later from paying customers for solving complex
problems.
Focus on the core by outsourcing other work to experienced
contractors.
Open-source all modules on GitHub.
Ajabbi
The common name of
A company to handle payments
Non-profit foundation to support users (to be established)
Research Institute for ongoing development (to be
established)
Description
Solved a huge, expensive problem with no existing competitors
Pre-revenue start-up with no $$$
Growing community
Purpose
Act like other foundations (Wikimedia, Apache and Linux) and use the
income surplus to support
Minority languages i18n
Open-source efforts (especially CFML, including Ortus)
User groups
User conferences
Book publishing
Open research grants
Etc
Pipi
The name of the software
Mike has been the architect since 1997
Versions 1-4 were in production eventually with 300K lines of code, an
800-table database, ESRI Web GIS, 500 classes, 3K methods, a 5 FTE team
and a 3-rack data centre
Versions 6-8 were working, non-production, undergoing refactoring
Version 9 works, moving to production (current)
Version 10 (planned)
Current Pipi 9
Works great
Runs in a dev environment
Unique architecture with 20+ layers and hundreds of
autonomous agents
Uses nature-inspired algorithms
Going from headless to a Pipi-generated UI
Migrating to a 42U server rack completely isolated from
the internet.
Using old Windows server boxes will do for now.
The CFML code runs on ColdFusion Server 2021 Developer Edition.
The CFML code could run on ColdFusion MX, Open Blue Dragon, or
Lucee.
Uses <CF> tags as in Ben Forta, et al. (don't tell Michaela, I
don't want to die)
Uses 300+ embedded config databases (MS Access)
Uses PostgreSQL for production data
Writes CFML code and then executes it
Writes SQL and then executes it to create and populate tables
Static HTML generator (90K page websites)
Uses XML, XML Schema, etc, internally
Mostly self-managing
Self-documenting
Currently rendering a 20,000-page static website for developers
The code base will self-grow 10x larger over time
Planned Pipi 9
Some money is coming soon
Generate revenue from niche batch render jobs
Code generators and the generated code must be rewritten to be BoxLang
compatible, so no refactoring by Ortus is required.
Pipi 9 will create Pipi 10
Pipi 10 to be BoxLang native
BoxLang (+++++++) paid support
Pipi 10 to become geodatabase capable
BoxLang will enable the future use of Python, PHP, etc
Planned Pipi 10 Systems
Render System
Functions as an automated AI-assisted development environment
Isolated from the internet
Uses VM
Run on BoxLang installed on Windows on new 64 hardware
JDK 21+
Ortus to supply a ready-to-run VM image.
100% self-managing via CommandBox dynamic CFML scripts
Renders working enterprise system modules (database, code, UI, API,
user documentation)
MS SQL edition
MySQL edition
Oracle edition
PostgreSQL edition
Exports modules to the Staging System via manual hard drive
transfers
Staging System
Connected to the internet
Uses VM
Runs on BoxLang installed on Debian
JDK 21+
Ortus to manage the BoxLang platform VM on a long-term contract
Uses PostgreSQL
Adds i18n to modules
Creates updates
Integrates i18n modules into customised enterprise systems for paying
customers
Exports
Github
i18n modules to GitHub for open-source
i18n user documentation
Cloud
Customised i18n enterprise systems
System updates
Cloud System
Connected to the internet
Uses VM
Runs on BoxLang installed on Debian
JDK 21+
Ortus to manage the BoxLang platform VM on a long-term contract
VM hosted
AWS
Azure
Digital Ocean
GCP
IBM
etc
private cloud.
The customer chooses their production RDMS database(s) for their
data.
MS SQL
MySQL
Oracle
PostgreSQL
Questions for Ortus Solutions
Answered by Cristóbal Escobar Henríquez, and Luis Majano.
Is a list of supported (or not supported) CFML tags available?
Yes! We maintain a detailed inventory of supported tags and
functions:
Yes, BoxLang fully supports CFML tags when running in
CFML compatibility mode. This ensures legacy CFML
applications can continue to function with minimal changes.
Is there a performance hit from using CML compatibility mode?
There is of course a penalty to pay if you are transpiling on the
fly. It's negligible to the point that we are faster in CFML
mode than all Adobe engines even in transpile mode. However,
you can transpile to BX as part of your build process and then this
won't exist.
Does BoxLang use CommandBox modules?
No. BoxLang Modules are something new. Eventually maybe, but for now, no.
Can BoxLang accept (CFML) command-line modules generated by Pipi without needing to be restarted?
Yes.
The ModuleService has the necessary API for you to load modules a-la-carte. All of our integration testing uses this approach.
You can also uninstall them
Will these CF tags become compatible with BoxLang in future?
cfftp
This is done in our bx-ftp module
cfregistry
This can be sponsored if needed.
cfschedule
This is in progress, but can be accelerated by sponsorship
Could BoxLang use MS Access via JDBC on the isolated Render System,
and how much would it cost to create a module to do this?
We can enable MS Access integration by building a custom module
using the UCanAccess JDBC driver.
Estimated time: 10–15 hours (2–3 business days)
Note: This estimate covers driver integration only and does
not guarantee full MS Access compatibility. As an alternative, we
recommend migrating the database to SQLite, Derby, or MySQL for
improved stability and long-term support.
What kinds of modules can be developed by Ortus?
Ortus can develop any module you need, whether it's for BoxLang,
application features, integrations, or infrastructure. We tailor
modules to your requirements, from core language enhancements to
platform-level tools.
What is the cost for Ortus to provide dedicated platform
support?
We offer dedicated platform support through our BoxLang+ and
BoxLang++ plans, which include guaranteed response times,
SLAs, priority bug handling, and access to our core engineering
team.
We’d be happy to tailor a plan based on your service level and
usage expectations.
Can patched VM images be supplied?
Yes, we can provide patched VM images as part of a BoxLang++
subscription. If you're not on a ++ tier, this can still be
delivered through a separate consulting plan.
What about the time zone differences between New Zealand, the US,
and Spain?
Our team operates mainly on Central Standard Time, but we
can easily align with your schedule. If needed, we can also assign a
developer located in the EU (CET) for better time zone
overlap.
Would a 100% dedicated Ortus developer be possible in the
future?
Yes, we offer team augmentation services, including
assigning a dedicated developer to your project. We can adapt
the engagement based on scope, timeline, and availability.
Is a 24/7/365 platform supported by Ortus possible in the
future?
Yes, that’s absolutely possible. We can support a 24/7/365
production environment using our teams in El Salvador and Spain.
This would require a custom service agreement or license tier beyond
the standard BoxLang offerings, but we’re happy to scope and price
this accordingly based on your needs in the future.
Questions for the Pipi Community
Would support also be available from independent CFML
contractors?
Where is the boundary between community contributions and paid
contractors?
What other new BoxLang Modules would be helpful, including other
coding languages?
