NederlandsORQ: The partner for AI Observability & Orchestration — without Microsoft lock-in.
KODIFY is one of the first ORQ.ai partners in the Netherlands. We install, configure and manage the observability layer that gives you control over quality, cost, safety and bias of your AI agents — across Claude, GPT-4o, Mistral, Gemini and LLaMA.
Kodify — one of the first ORQ partners in the Netherlands. Production deployment at PostNL.
ORQ.ai is a Dutch AI Observability and Orchestration platform: an independent layer above any LLM — Claude, GPT-4o, Mistral, Gemini, LLaMA — where you centrally monitor and control quality, cost, latency, bias and PII exposure.
KODIFY was one of the very first partners of ORQ. Our AI Consultant Wouter Sligter installed ORQ and took it to production at PostNL — including routing engine, RBAC, vaults and the OTAP pipeline their teams now release on.
ORQ founder Sohrab Hosseini is a long-time friend of KODIFY and a regular speaker at our AI Power Sessions. His core message: 85% of AI initiatives fail not on the model, but on the organisation around it. That shared focus is where we work together.
The six questions you must be able to answer about your AI.
Without an observability layer, AI in production is a black box. ORQ.ai gives KODIFY — and your team — answers to the six questions every CIO, security officer and finance lead asks, all in one dashboard.
What is the quality of my agents?
Continuous evaluation of output — hallucinations, tone, language quality and correctness per use case. ORQ scores every run and routes regressions back to your team.
How are employees using the AI?
Which prompts, which agents, which output — per team, per user. Find the power users and the departments still falling behind.
Which LLM works best — and when?
Compare Claude, GPT-4o, Mistral and LLaMA on quality, latency and cost. ORQ's routing engine automatically picks the right model per request.
Which agent costs the most?
Token usage and € cost per agent, per team, per LLM. Set budget caps, alerts and cost fallbacks before a cost centre spirals out of control.
Is there bias or PII leakage?
PII detection on input and output. Bias evaluations on sensitive decisions. Guardrails that automatically block personnel files and business-sensitive data.
Where are my bottlenecks?
P95 latency, retries, fallbacks and tool failures visible in real time. Find what is slow, what fails too often and where caching is needed.
Build → Deploy → Evaluate → Improve. In one platform.
Agents don't fail on the model — they fail on the loops around it. ORQ orchestrates those four stages as one product, with shared accountability between Product, Engineering, ML and Operations.
Build
Studio for agents, prompts, tools and workflows. Every change version-controlled, with OTAP.
- Prompt & workflow editor
- Tool & MCP library
- RAG-as-a-Service
- Multi-LLM playground
Deploy
Safely to production with OTAP, canary releases, PII control and RBAC — without bypassing your IT team.
- OTAP / DTAP pipelines
- Canary & A/B releases
- PII control on input/output
- Vaults, RBAC, IP allow-list
Evaluate
Every run scored on quality, cost, latency and bias. No opinions — dashboards and regression tests.
- Automated & human evals
- Bias & safety monitoring
- Quality regressions per release
- Cost per agent / team / LLM
Improve
Reversible changes based on data. Switch LLMs, rewrite prompts or add tools — and measure the impact immediately.
- Routing engine per use case
- Prompt improvements from data
- Roll-back on bad release
- Shared accountability dashboards
One stack — from model to governance.
ORQ.ai sits as an observability and orchestration layer between your applications and the LLMs. Below are the four rings we set up and manage for you.
Wouter Sligter took ORQ to production at PostNL.
Wouter — KODIFY's AI Consultant and one of the first ORQ implementors in the Netherlands — installed, connected and launched the platform at PostNL in production. From vault and RBAC configuration to routing engine, OTAP pipelines and the evaluation loops the team now releases on.
No sandbox pilot: PostNL runs ORQ live, with real volumes and governance at the level an enterprise security team demands. It is the blueprint we use to bring similar implementations at other clients to production in days — not months.
PostNL — ORQ in production
Not on a Microsoft-first philosophy? Then ORQ is stronger.
Microsoft Purview is excellent if your entire world sits in Azure and M365 — it is already there, and you use it accordingly. But most organisations we speak to do not want to be tied to one cloud and one LLM vendor.
For those organisations, ORQ.ai is our preferred supplier for observability: vendor-neutral, broader in AI-specific evaluations, stronger in routing and lifecycle control — and owned by a Dutch team that KODIFY can work with directly.
We are pragmatic — where Purview wins (M365 DLP, document classification), Purview wins. For the AI-specific work, ORQ wins.
Sohrab Hosseini, CEO of ORQ.ai, on stage at KODIFY's AI Power Sessions.
Sohrab is a well-known face at KODIFY — and has spoken at our Partner Event among others. His keynote covers what he sees daily at enterprises: the organisational disconnect that blocks real AI production, and how a collaborative platform resolves it.
"Most AI agents fail in production — not because the model is weak, but because teams can't collaborate on how they're built, evaluated and improved."
Why 85% of AI initiatives fail — silos, not the model.
An open talk for data, AI and operational leaders. Concrete, with live examples from Sohrab's own clients.
Senior consultants — all ORQ-trained with production experience.
We assign the right consultant to lead each engagement. Everyone below has been trained by ORQ and has taken one or more implementations to production.
Ready to get your AI
production-observable?
30 minutes with Wouter Sligter or a senior consultant. We map one agent or use case onto the ORQ stack and show you what we can set up in days — not months.
More about the platform →