ATAMYA AI Strategy

AI that collaborates with you. Instead of merely supporting you.
Most PIM systems treat AI like a drawer full of tools: you take out whatever you need at a given time. In case of ATAMYA, AI becomes an integrated part of your system. You use it in workflows, in the chat, and in background monitoring – with clear roles, a mind of its own, and full traceability.

That’s our path.

Experience Agentic PIM

AI executes. AI answers. AI watches.

Before talking architecture, models, and governance, let’s cover the essentials first: how AI actually operates in ATAMYA. Three intelligent modes that improve product data exactly where efficiency, quality, and speed shape decisions.

 

AI executes

Workflow Intelligence. The AI enriches, classifies, translates, analyzes for compliance – directly as part of the process. Automatically triggered, rule-based, controllable, and completely traceable.

AI answers

Conversational Intelligence. The sidekick answers questions revolving around product assortments in natural language. It recognizes your data model, your domain, your rules. Without exports, no SQL, no detours.

AI watches

Proactive Intelligence. The AI scans your product data continuously to check for anomalies, missing values, and rule violations – so that it can report back before a problem reaches your channels.

AI Only Becomes Valuable, Once It’s a Component of the Process

Almost every PIM offers an AI button nowadays. A text generator. A few automation mechanisms. It helps, yet it does not really drive companies forward.

The difference between a productivity tool for individuals and an intelligent infrastructure lies not in the model but the architecture. When AI operates external to the process, island solutions and workarounds pile up that are completely unconnected to the workflow where the actual decisions are being made.

ATAMYA is designed differently from scratch: AI is a native component of the process, not an addon. Our approach is grounded upon five design decisions, some of which are already productively used today, while others will be built as part of the next releases. We make transparent what is already up and running today and what’s next.

AI within the Process, Not Next to It

Every AI call in ATAMYA is a native BPMN service task. This sounds technical – but it constitutes the very core of our approach.

More concretely, this means:

  • AI results flow directly into the decision-making process
  • multiple AI steps can run at the same time or iteratively
  • retries, timeouts, and fallback paths apply automatically when a provider is momentarily unreachable
  • automatic transfer to user tasks in critical cases
  • human-in-the-loop as a native step, not an emergency exit

The AI creates a product description. The workflow evaluates the result. If it matches the defined rules, the process continues on its own. If not, then a task is sent to a real person — with context, reason, and audit trail.

Full Freedom with Your Model of Choice

No lock-in. No hidden dependencies. You choose the model that matches your requirements best and are free to change whenever you want to.

  • Free Provider Choice: OpenAI, Anthropic, Google, Azure, AWS Bedrock, or On-Prem.
  • Your Own API Key: Full transparency for use and costs.
  • Change Models Anytime: Without the need to readjust your workflows.
  • Private Enterprise LLMs: Your data stays within your organization

Governance First: The AI Only Does What You Allow it To Do

LLMs are powerful. And it is precisely for this reason that we handle them with clearly defined boundaries and controlled processes.

  • Separate Layers: Model and tools are strictly decoupled
  • Permission-Aware: The AI only has permission to do what users or workflows specify
  • Guardrails: Limits for tool calls, response size, and runtime per execution
  • Audit Trail: Complete traceability for all model calls
  • Human-in-the-Loop: As a configurable process step, not an emergency solution

Trust is not a setting. It’s an architectural decision.

ATAMYA Knowledge Layer: A Mind That Grows with You.

Most AI solutions in the PIM section start from zero with every prompt. Each interaction begins anew. This is like working in a new product manager every monday – just to let them leave again on friday.

What we build: the ATAMYA Knowledge Layer

Persistent. Provider-independent. Growing over time.

  • Object Memory: What are the lessons already learned concerning this product?
  • Channel Memory: What works on Amazon and what works on the B2B platform?
  • User Memory: How does this team work, what are individual preferences?
  • Domain Memory: What about industry-specific knowledge?
  • Event Memory: What happens and in what context?

The objective: AI that grows with your company.

No Chatbot. An Agent with Roles, Rules, and Contexts.

Our ID: Agents are not chatbots. They are configurable roles – with a name, their own system prompt, their own tool access, and their own contexts.

You can already configure a dedicated AI integration per use case today: with the model, rules, tool access, and your user-defined system behavior. The next step: Agents with permanent contexts and consistent behavior – continuous across sidekicks, workflows, and external systems.

The objective: One and the same agent for each use case. For chatting. In workflows. Or via our MCP server, in tools such as Claude Code or n8n.

See What Happens When AI is Integrated Into Your Product Data.

AI at its core. Product data in its flow: Experience how our automation functions – from smart import, through workflows, to an MCP server, as well as your own LLM integration.

Request AI Demo