Platform
Build the interesting part. We handle the rest.
The platform (The AI agents we build for your firm run on our platform. We host the platform in Switzerland; external models may process data elsewhere.) carries what we build for you, on our own infrastructure in Switzerland.
The platform comes with the Setup and the Business seat (The Business seat is one person’s access to our platform (open alpha); Pro Setup includes three months, and with Operator System it is six per person.); its parts are not sold separately. It is still in open alpha (Open alpha means the platform is at an early stage: we use it ourselves, and it changes all the time.), and we run our own company on it.
How it fits together
Your workspace, your projects, your call.
The platform in six answers, before the detail.
One workspace for the whole workday.
The Vault (The vault is a folder of your working files on your computer; we create it from the vault template.) is one person’s folder on their computer: projects, notes, documents and the instructions for recurring tasks in one organised place. Local, our local app (A local app is a program installed on your own device, unlike a web app, which runs in the browser.), connects it to the platform. So your computer does not become a graveyard of half-configured experiments.
Your code is yours.
Code you write belongs to you. Shared work lives in the shared project repository (A Git repository is storage for files that records every change with its date and author; GitHub is a well-known service for it.), which you control, for example on GitHub or on your own Git server. If your access ends, everything on your device stays there.
The AI harness leads the project.
In the AI harness (An AI harness is the workspace on your computer in which AI reads files and carries out agreed steps.), the AI reads a project’s files and carries out agreed steps. AI agents (An AI agent carries out several steps of a defined task. A person decides before it sends anything outside your firm or spends money.) take on parts of the work, each with a task of its own. You decide which harness to use; we recommend Claude Code (Claude Code is an AI harness made by Anthropic: it runs on your computer, reads files and carries out steps you describe in plain language.).
The platform keeps the state.
The platform records tasks, runs and results beyond any single chat window. So tomorrow does not start from zero, and the log (A log records what a program or an AI agent did and when, so that afterwards you can trace who triggered what and what happened.) shows what changed.
Humans and agents, each doing their part.
AI agents do the tasks you give them and propose changes; a change goes in only once a person accepts it. A person decides on anything that leaves your firm or costs money. And the log shows you every change.
Private workshop, public result.
A public link publishes only what you choose for it, an app or a read-only view of your data, for example. Your private vault is never published with it.
How it is built
Nine planes, one spine.
Your business on top, then the entry points, control, capabilities, execution and your workspace. On the left the path every project takes, on the right governance and the services the platform works with.
Point at a plane or tap it: its parts come forward.
- Idea
- Requirements
- Design
- Plan
- Build
- Review
- Deliver
- Operate
Every project takes the same steps, from the idea through plan, build and review to operation.
More on this plane: Build lifecycle- Services
- Templates
- Platform capabilities
One part tailored, the rest sensibly reused: nobody needs to rediscover the login form.
More on this plane: Business and delivery model- Dashboard
- Local
- Command centre
- Claude Code and other AI harnesses
- Installation
- Operations console
You work in the web dashboard, in the local app Local or with the AI harness of your choice.
More on this plane: Interfaces and entry points- Sign-in
- API
- Project registry
- Tasks and workflows
- Runs and scheduling
- Capability gateway
- Git and vault registry
- Publishing
This is where it is settled which projects exist, who may do what and which task runs where.
More on this plane: Control plane- OCR
- Routing
- Messaging
- Scraping
- Document processing
- Scheduling
- Knowledge search
- Integrations
- Secrets and configuration
What has proven itself in one project is available to every project, from OCR to scheduling.
More on this plane: Capability plane- Your computer
- A machine on your premises
- Our infrastructure in Switzerland
- Cloud workers
A workflow runs on your own device, on a machine on your premises, on our platform in Switzerland or on cloud machines; the quote sets where.
More on this plane: Execution plane- Vault (PARA)
- Project repositories
- Templates and components
- Logs and artifacts
Your work lives in your vault and in project repositories you control. Your code is yours.
More on this plane: Workspace, projects and assets- Identity
- Approvals
- Security
- Observability
- Activity log
- Support
Agents propose changes; a change is applied only once a person accepts it. What goes out or costs money is decided by a person, and the log shows every change.
More on this plane: Governance and operations- Code hostingGitHub: project repositories, Git and vault registry
- Publishing and deploymentVercel, as an option
- The platform’s database
- Object storagelogs and artifacts; backups at Exoscale
- Secured remote connection
- Third-party APIsfor integrations
- Model providersfor example Anthropic, OpenAI, Ollama, OpenRouter: for Claude Code and the capability gateway
The platform works with services you may already use: Git hosting, publishing, storage and the model providers of your choice.
More on this plane: External systemsPlane by plane
What each plane does for you.
In the drawing’s order, with the details for everyone who wants them.
Build lifecycle
Every project takes the same steps, from the idea through plan, build and review to operation.
- The steps match our ways of working: Figure It Out (Figure It Out is one of our five ways of working: turn uncertainty into a decision you can act on.) clarifies at the start, Build It (Build It is one of our five ways of working: build what does not exist yet, around the way your work actually happens.) builds, Own It (Own It is one of our five ways of working: someone who stays on and sees how the pieces fit.) runs it afterwards.
- How much checking a step needs depends on the risk: a small script does not need the same acceptance as a system your company runs on.
- An AI saying it is done is not enough. It is done when it has been checked.
Business and delivery model
One part tailored, the rest sensibly reused: nobody needs to rediscover the login form.
- On top are the services you get from us: the Setup and our five ways of working, from Figure It Out to Own It.
- Beneath them the templates (A template is a prepared structure, for a document, an email or a folder for example, which you only complete for each case.) projects start from, and the capabilities all projects share. How they come together is further down, under ‘How a project starts’.
Interfaces and entry points
You work in the web dashboard (A dashboard is an overview page, usually in the browser, that shows the state of tasks, results or figures at a glance.), in the local app Local or with the AI harness of your choice.
- Your staff start a workflow (A workflow specifies what information goes in, what happens to it and who checks the result.) with a short command (A command starts a defined workflow. In Pro Setup, we set up three to five commands for your work.) and check the result wherever suits them: in the AI harness, on the command line (The command line, also called the terminal, is a window in which you give the computer instructions by typing text instead of clicking.), in a dashboard or command centre (In a command centre, your staff start workflows and check results without using the command line.), or in a desktop or web app.
- Local connects your device to the platform, and there is an Android app for when you are out.
- The platform is not tied to any one agent; we recommend Claude Code. We choose the harness, model and provider with you. Windows and macOS are supported.
- For now we set the platform up with you rather than leave you alone with a download.
Control plane
This is where it is settled which projects exist, who may do what and which task runs where.
- Each project describes itself in a project manifest: what belongs to it and what it needs from the platform.
- Each run follows a run contract. It sets the terms a run happens under and is the same on your computer, on your hardware and on our servers (A server is a computer that runs all the time and provides services to other devices, such as email, files or a website.); not every place offers the same facilities.
- The manifest and the run contract are the technical contract. The commercial one is separate, in a binding written quote before we start: scope, price, payment, acceptance and licence.
- A person approves every new account on the platform.
Capability plane
What has proven itself in one project is available to every project, from OCR to scheduling.
- The platform includes an MCP gateway (MCP (Model Context Protocol) is an open standard for connecting an AI harness to other software and data sources.), a messaging bus (The messaging bus passes messages between the AI agents, the platform and your phone.), an API (An API is an interface through which one program uses the data or functions of another, without a screen or a mouse.) for document and data processing, OCR, data ingestion, documents generated from your templates, the choice of model, tool and integration (An integration connects your setup to software you already use, such as your email inbox.) for each task (routing (With routing, a system picks the right language model, tool and integration for each task.)), speech to text and back, web scraping (With web scraping, a program reads details from web pages and stores them in order.), scheduling, search, a database with structures you define (A database whose tables and fields you define, for suppliers or orders for example.), Local as the local app, an Android app and a web dashboard.
- Each AI agent has its own task and works alone or together with other agents.
- What proves itself in one project can become a template or a shared capability. The capabilities come with the platform, not as a product of their own.
Execution plane
A workflow runs on your own device, on a machine on your premises, on our platform in Switzerland or on cloud (The cloud means computers in a provider’s data centres on which programs run and data is kept, instead of on your own device.) machines; the quote sets where.
- A workflow installed on your work device runs only while the device is on. If it needs to run without that device, we agree the required setup in the quote.
- On the platform, user agents run in sandboxes (A sandbox is a sealed-off area in which a program runs without reaching the rest of the system.), sealed-off areas of their own.
- Our services run on our own infrastructure in Switzerland. Language models (A language model is the actual AI behind an AI assistant: it has learned from a great deal of text and writes an answer to an input.) run in the cloud with providers in the US or the EU, or as open-weight models (An open-weight model is a language model whose weights are freely available, so it can run on your own hardware.) on your own or our hardware.
- Third-party providers, such as AI model providers, do not process inputs in Switzerland by default.
- We quote a local or hybrid (Locally, the language model runs on your hardware; in a hybrid setup the working files stay local and only the excerpts needed go to a cloud model.) solution separately.
Workspace, projects and assets
Your work lives in your vault and in project repositories you control. Your code is yours.
- The Vault is one person’s folder on their computer. It is organised by PARA (PARA organises storage into four parts: projects, areas, resources and archives.) and opens in Obsidian (Obsidian is a note-taking program that shows a folder of text files on your computer as linked notes.) by default; other tools work too.
- What the team shares lives in the shared project repository, which you control, for example on GitHub or on your own Git server.
- Your files live on our infrastructure or on yours, whichever you choose.
- We keep backups (A backup is a copy of your files in a second place, from which you restore data that was lost or damaged.) locally on your device, on network storage (NAS) or hosted in Switzerland at Exoscale (Exoscale is a Swiss cloud provider with data centres in Switzerland. It rents out servers and storage in its own data centres.).
- You receive a licence to use the workflows, skills (A skill is a written instruction for a specific AI task, such as checking invoice details.) and commands we build for you; the rights remain with Jules Solutions. The written quote sets the terms.
- The written quote sets out what the platform stores for you, how files are backed up, what availability you need and how you can retrieve your data.
Governance and operations
Agents propose changes; a change is applied only once a person accepts it. What goes out or costs money is decided by a person, and the log shows every change.
- Nothing we build sends a message outside your firm or spends money before a person decides.
- Each AI agent sees only its own project, and different clients’ data is kept apart on the platform.
- Every agent run is logged. The log records requests and changes, each change with the state before and after.
- Each person has their own access: one Business seat covers one person and their main device, plus their phone and a shared drive.
- In a team, everyone who works in the system needs at least a Business seat. Anyone who does not set up their own machine adds a Pro Setup (Pro Setup is our recommended package at CHF 1'490.–: vault, workflows with AI agents, first integrations, an introduction session and three months of Business seat.).
- If someone leaves the team, you revoke their access to the vault, the AI provider (An AI provider runs language models and sells access to them, for example OpenAI (ChatGPT), Anthropic (Claude) or Google (Gemini).) and the platform. The handover document (The handover document records what we set up, how you work with it and how you back it up and restore it.) shows you how.
- We keep credentials out of chats, for example in 1Password (A password manager such as 1Password stores passwords and other credentials encrypted in one place.) or a designated local file.
- We work on your device on site or remotely, and we run remote sessions (With remote access, someone works on your device over the internet without being there, to set up a program for example.) only from Switzerland. We have no standing access: every remote session needs your consent. We agree in advance which access we need.
- At handover, we remove our access. You then replace any credentials known to us.
- If your setup touches personal data you are responsible for, we sign a data processing agreement (A data processing agreement (DPA) sets out how a service provider handles personal data it receives on your behalf.) with you before work begins, as standard. Otherwise we sign one on request.
- We confirm your message within 2 hours and start working on it within 24 hours, seven days a week.
- Jules Solutions does not hold a data protection certification.
External systems
The platform works with services you may already use: Git hosting (Hosting means that a provider runs the computers on which a website, a program or data runs and can be reached.), publishing, storage and the model providers of your choice.
- Your project repositories live with a Git host such as GitHub or on your own Git server.
- Vercel (Vercel is a US provider that runs websites and their server functions.) is also available for hosting if you prefer.
- You choose the model providers with us, for the AI harness and for the platform’s capabilities alike.
- The quote records which work content goes to which providers.
- Your data, files and accounts with your chosen providers are yours.
Same room. Different work.
Before: the team already uses AI, but without a system, and the paper keeps piling up. After: the workflows run, and you have time for clients and strategy. The robots stand for software you approve.
Drag the slider.
Scroll, and the room transforms. Use the slider to compare before and after.
What holds
Six things you can count on.
They hold for every project, wherever it runs.
Each person has their own vault.
The Vault is one person’s workspace. What a team shares lives in the shared project repository.
You control your repositories.
Code you write belongs to you. Changes from AI agents arrive as proposals and go in only once a person accepts them.
One contract for every place.
Every run follows the same authorised run contract, on your computer, on your premises, on our platform or in the cloud.
A person has the last word.
You decide which AI harness to use. Anything that leaves your firm or costs money is always decided by a person.
Only what you choose is published.
A public link shows what you release for it and never opens your private vault.
Each agent stays in its project.
An AI agent sees only the project it works for, and different clients’ data is kept apart.
How a project starts
One part tailored. The rest sensibly reused.
Every project combines custom work, proven templates and the platform’s capabilities. Your solution gets the parts that make it different. Nobody needs to reinvent the login form.
From a template
A proven project structure with tests and design is the starting point. It is copied and developed further for your project.
From your repository
You bring an existing project as your own repository, and we carry on working in it.
Built to measure
What exists only for you: your own integrations, apps, dashboards, AI agents and business logic.
Either way, the project uses the platform’s capabilities. It is built in the Build It way of working. More on Build It
At a glance
The platform in facts.
- Name
- Jules Solutions Platform
- Run by
- Jules Solutions, Meggen
- Hosting
- Our own infrastructure in Switzerland, Vercel on request
- Stage
- Open alpha; we run our own company on it
- Build
- Separate services instead of a single program, since April 2026
- Tests
- 2,616 passing, none failing, 52 per cent of the code covered (release log of 30 September 2026)
- Runs on
- Your device, a machine on your premises, our platform or cloud workers, per quote
- AI harness
- Your choice, we recommend Claude Code
- Systems
- Windows and macOS
- Access
- Business seat, CHF 199.– per person per month, CHF 169.– a month billed yearly (CHF 2'028.– a year)
- Account
- On our account page; a person approves every account
There is deliberately no list here of features that might arrive one day.
Access and cost
Access through the Business seat.
A Business seat gives one person access to the platform. Pro Setup includes three months, and with Operator System (Operator System is our package for teams at CHF 3'500.– to 5'000.–: the shared base with workflows for the team and connected systems.) it is six per person. After that it costs CHF 199.– per person per month, or CHF 169.– a month billed yearly, which is CHF 2'028.– a year.
- Without renewal, access to the platform ends. Your working files remain in the vault on your machine. What keeps running without the Business seat depends on the setup; the written quote states it.
- The quote states the notice period.
- You create your account for the platform on our account page.
How to start
Four ways in, from the free interview to a written request. You choose how much you tell us first.
Where you stand
A few questions about your working day, and you see where you stand with AI and where a first step pays off. Free.
Start the interviewTalk to Jules
20 minutes with Jules in person. You tell him where things get stuck, and he tells you honestly whether and how he can help. Free.
Pick a timeAccount
Sign in with a link by email, no password needed. Pick up saved answers, download your details or delete them.
Go to your accountRequest
Rather put it in writing? Describe what you need in the form. We confirm within 2 hours and start working on it within 24 hours.
Go to the form
Or write to Jules directly in the chat on the contact page.Go to the chat
Rather roll up your sleeves? You can build the basic setup yourself, too. Do it yourself
FAQ
Questions about the platform
What might I pay for after setup?
You pay once for the setup, at a fixed price. We put its scope and price in writing before work begins.
Running costs may include a Business seat, if your package includes one and its included months have ended (CHF 199.– per person per month, or CHF 169.– a month billed yearly, which is CHF 2'028.– a year), model usage (Model usage means accessing an AI provider’s language model. The cost is not included in the setup price.) (charged according to usage) and separately agreed support (CHF 500.– to 2'000.– per month). Other licences may cost extra.
For model usage, here are two examples of business plans for two or more people, as of October 2026, depending on plan and billing: ChatGPT Business: CHF 15.– to 19.–; Claude Team: USD 20.– to 125.–, each per person per month. Usage beyond what a plan includes costs extra. You choose the provider. If you prefer, we manage your account with the AI provider or provide API keys (An API key is a secret access code that lets a workflow use a service such as a language model.) where the provider’s licence allows it. We invoice you for the cost.
For an example of first-year costs, see ‘Additional costs’. ‘What does AI automation cost an SME? Setup and running costs’ shows how to calculate your costs using a worksheet.
Where will our working files be after setup?
The working files in the installed vault are on your machine. If you use a cloud model, its provider receives the content you send to it. We discuss which data you can use with it.
For confidential files, we assess whether a local or hybrid setup fits. We host our platform in Switzerland, but a cloud AI provider may process content you send it elsewhere. Before setup, we establish what goes to which provider. See ‘Data and processing’.
The written quote sets out how files are backed up, what availability you need and how you can retrieve your data.
Can we keep using the setup after handover without you?
The AI harness runs on your device. Your working files are in the vault on that device, and the accounts with your chosen providers belong to you. The handover document records what we installed. So another IT professional can carry on with it.
Pro Setup and Operator System also include platform access through a Business seat for a limited period. If you do not renew the Business seat after the included period, platform access ends. Your working files in the vault remain on your machine. What keeps running without the Business seat depends on the setup; the written quote states it.
What happens to the vault if a device fails or someone leaves the team?
The vault is a folder containing ordinary working files. The backup lives wherever you want it: on your own infrastructure, on ours or with a cloud provider of your choice. We document how to restore the vault on another device.
For Operator System, shared work lives in the shared project repository, on GitHub or on your own Git server, for example, or wherever you prefer; the vault stays each person’s own folder. We agree with you who can access the repository.
If someone leaves the team, you revoke their access to the vault, the AI provider and the platform. The handover document shows you how.
Must we use Claude Code or a particular provider?
No. The choice of AI harness, model and provider remains yours. Claude Code is an option, not a requirement. Your working files are in the installed vault on your machine. If you use a cloud model, we discuss what content goes to its provider.
Start with the interview
Find the work your team keeps doing by hand.
The free interview asks about your work, one question at a time. Your report shows where your time goes and where it pays to start.
No technical knowledge needed, no account. You can stop any time and continue later.
