WITTIGONIA Agentic AI Governance
Human-architected. AI-accelerated. Disclosed, not hidden.
We use AI, including agentic AI, to do the work faster and go deeper. Every system we build stays governed, auditable, and clearly labeled, so you always know what a human decided and what a machine assisted with.
Why this page exists
AI without governance is not leverage, it is risk moved somewhere you can't see it. WAI is how WITTIGONIA keeps that risk visible: a plain statement of how we use AI in our own work, how we govern it, and how we disclose it, published where anyone, a client, a partner, a regulator, can check it.
Our commitment to transparency
The EU AI Act's Article 50 transparency obligations took effect on 2 August 2026. They require AI-generated and AI-manipulated content, text, image, audio, and video, to be disclosed and, where the provider is responsible, marked in a machine-readable format. Systems already on the market before that date have until 2 December 2026 to complete machine-readable marking.
We are not waiting for the deadline. Content and media that WITTIGONIA produces with meaningful AI involvement carries the WAI mark:
AI-assisted - learn more at wittigonia.net/ai/
Three rules govern how we apply it:
- Human-led, always. AI drafts, researches, and accelerates. A named person at WITTIGONIA reviews, edits, and takes responsibility for what gets published or delivered.
- Marked, not buried. Where AI involvement is substantial, we say so, on the page, in the image metadata, or in the post itself, not in a footnote nobody reads.
- No unlabeled synthetic media. We do not publish AI-generated images, audio, or video under WITTIGONIA's name without disclosure.
How we govern AI and agentic AI
Agentic AI, systems that take multi-step action rather than just answering a prompt, changes what governance has to cover. Our approach rests on four points we hold ourselves to before we ask a client to trust them:
Pillars | What it means in practice |
|---|---|
Data privacy | Client data stays client data. No training on proprietary inputs; EU data residency where the platform supports it. |
Bounded action | Agentic workflows run inside defined permissions and schemas. An agent doesn't get access it doesn't need for the task in front of it. |
Traceability | What an AI system did, and when, is logged. If something needs to be reconstructed later, it can be. |
Human checkpoints | Decisions with real consequences, financial, contractual, personnel, get a human confirmation step before they execute. |
Foundation first: infrastructure and the Zoho ecosystem
AI built on a fragmented data setup produces fragmented results. Before we add AI or agentic capability to any client environment, we make sure the underlying system can support it: one system of record, clean data flow, logic that holds together end to end.
This is where the Zoho ecosystem, Zoho One, CRM, Analytics, and the automation layer connecting them, does the heavy lifting for most of our clients. We hold individual Zoho certifications in Analytics, Creator, and CRM today, and are working toward formal Zoho partner status. We configure and extend the ecosystem, connect it to agentic workflows and frontier AI models where that adds real capability, and build the dashboards that let you see the system working, not just its output.
Technology and platforms
We work across leading AI and business platforms and choose deliberately, rather than commit to one vendor for everything.
Frontier AI: Anthropic, Google AI, OpenAI, and Zoho Zia.
Agentic AI: Zoho Zia Agents for automation embedded inside Zoho, and Anthropic's Claude Code for agentic engineering, research, and content workflows, including the one that built this page.
Business infrastructure: the Zoho ecosystem, Zoho One, CRM, Analytics, and Creator.
AI across our 3D practices
Digital. Campaign and channel work gets an agentic layer where it earns one, lead triage, intent routing, cross-channel consolidation, always landing in a clean CRM record, not a parallel spreadsheet. Explore Digital
Data. Dashboards and reporting built to be stock-and-flow consistent, so an executive dashboard shows the state of the system, not just a snapshot. AI accelerates the ETL and querying; the modeling discipline underneath is what makes it trustworthy. Explore Data
Dynamics. System dynamics and simulation work, built in Vensim, applied to real decisions under real uncertainty. We increasingly use coding AI, including Claude Code connected through MCP, to automate the workflows around a model and build the interfaces that make it usable. The causal thinking still comes from the model, AI speeds up everything around it. Explore Dynamics
AI governance is a moving target
This field, and the regulation meant to govern it, is moving fast, and the rules are still catching up to the technology. WAI is a living framework, not a fixed policy. We update it as our own practice, the available tools, and the regulatory picture evolve. We deploy on ourselves first, which is why this page is live today rather than waiting for a finished standard that may not arrive for years.
Explore further
- Our services
- Client projects and references
- Insights blog
- WITTIGONIA Lab
- YouTube - Podcast
- The 3D framework
- About WITTIGONIA
Get in touch
We'd rather you trust what we build because you can see how it's governed, not because we told you to. If you want to talk through what agentic AI, done properly, could do inside your Zoho environment, get in touch.