The AI platforms we build on, and how we pick one for your workflow
You bring the workflow and we choose the platform that fits it, whether that is n8n, Microsoft Foundry, Sim, Kestra, Claude, ChatGPT or SAP. The same team scopes it, builds it and runs it after launch.
Already paying for one of these? We build on the platform you have.
- When something happens in one app, update the othersn8n
- An agent that works through a queue of tasksSim
- An assistant that answers from your own documentsMicrosoft Foundry
- Custom agents IT and security can sign offMicrosoft Foundry
- Scheduled data and AI pipelines, defined as codeKestra
- AI for everyday work across every teamClaude or ChatGPT
- Agents inside your SAP processesSAP Joule Studio
The first four jobs are scored across eight platforms in our platform comparison, checked 4 October 2026.
Clients include
Every platform we build on, by the job it does
Platforms in the same group do the same kind of job. Each page shows where that platform fits, what it costs to run and where we would build on something else.
A trigger runs the same defined steps every time, with AI inside the steps, so every run is predictable, testable and auditable.
You describe a job and an agent works a queue of tasks with the tools you give it.
- SimOpen-source agent builder with tools, memory and human approval steps.
Agents built inside your Microsoft estate, under your identity, data and compliance rules.
- Microsoft FoundryAgents built in your Azure tenant under Entra ID, billed on your Azure account.
Assistants your whole company uses for everyday work, set up with the connectors and permissions each team needs.
- Claude CoworkAnthropic's agent for knowledge work, with the connectors and skills your teams use.
- ChatGPT for workChatGPT rolled out to your company with connectors, custom GPTs and admin settings.
AI inside the SAP systems your finance and supply chain teams already run.
- SAP Business AIJoule and SAP's AI features, connected to the processes you run in SAP.
- SAP Joule StudioCustom Joule agents and skills built for your own SAP workflows.
Score eight platforms against what matters to you: where it runs, who builds it, how many runs and how much AI it needs.
Open the platform comparisonSix questions that decide the platform
We ask these in the scoping call, before any platform is named. The platforms that score highest on each come from our platform comparison.
Where must it run?
If the workflow has to run on your own servers, n8n and Sim both self-host. Microsoft Foundry does not, but its Standard setup keeps agent data in your own Azure resources.
Who will change it after launch?
If your developers will maintain it, we use the platform they already know. If business users will, we pick one they can edit and train them on it.
How much custom code does it need?
n8n and Microsoft Foundry score highest when the workflow needs custom code, its own models or full control over hosting.
How many runs a month?
At millions of runs, billing decides the cost. n8n bills once per workflow run however many steps it takes, which keeps high volume cheap.
A few AI steps, or an autonomous agent?
Sim and Microsoft Foundry score highest for agents that choose their own next step. A workflow with a few AI steps fits n8n.
What do you already run on?
Teams on Microsoft 365 and Azure get the most from Microsoft Foundry. Teams that run finance and supply chain in SAP start with SAP Joule Studio.
What teams ask about AI platforms
Ask us something elseWhich AI platform should we use?
The one that fits the job. Workflows that move data between apps on a trigger fit n8n. Agents that work a queue of tasks fit Sim or Microsoft Foundry. Assistants your security team must approve inside Azure fit Foundry. The platform comparison scores eight platforms against your own priorities.
Do we have to choose a platform before we talk to you?
No. We scope the workflow first: the systems it touches, who will maintain it and where it has to run. The platform follows from those answers, and we tell you why before you commit.
Can one workflow use more than one platform?
Yes, and it often should. A common pattern is n8n moving the data between your apps and calling a Microsoft Foundry agent for the reasoning step.
Which of these can we self-host?
n8n has a self-hosted Community Edition plus Business and Enterprise plans. Sim is open source under the Apache License 2.0 and runs with Docker Compose or Kubernetes. Kestra is open source. Microsoft Foundry runs in Azure only.
Who runs the system after launch?
The engineers who built it, in weekly sprints that you direct, with error alerts going straight to them. See AI managed services.
Where should we start?
With the workflow that has the most volume and the clearest rules. If you already know which one, book a discovery call. If you have several candidates, the AI Audit ranks them by value, effort and risk over 4 to 8 weeks and ends in a 90-day roadmap. It costs $10,000 to $100,000 depending on scope, and the fee is credited toward your first build.
Put your first workflow on the board
Answer a few questions about your team and what you want automated, then Jason scopes it with you on the call.
