Iteractive.ai

Enterprise AI, implemented with control.

We implement enterprise AI: Anthropic Claude and Claude Code, OpenAI Codex, SAP Joule. We build MCP servers over SAP and other enterprise systems so Claude works with your data, establish governance, and put production use cases into operation. Remote, worldwide, with the model hosted in the EU.

EU data residency · GDPR & EU AI Act · every request audited

What we implement.

  • A secure company AI assistant: A company-wide AI assistant on Claude with sign-in through your existing accounts, personal data masked before it leaves your network and every request audited. Deployed in the EU. How it works and what it costs to run.
  • Claude Enterprise implementation: How we implement Claude Enterprise: company sign-in, user management, audit, contractual no-training, and when AWS Bedrock or Google Vertex AI in the EU is the better route. Delivery in five stages.
  • Claude Code and Codex for software teams: A governed rollout of Claude Code and OpenAI Codex for engineering organisations: policy, secrets handling, code review gates and measurable adoption instead of unmanaged licences.
  • Claude on SAP through MCP: We build and run MCP servers over SAP ECC and S/4HANA so Claude can read SAP data and prepare transactions through OData and RFC/BAPI. No S/4HANA migration, model hosted in the EU, sign-in with your accounts, every call audited. The server is open source.
  • AI on SAP ECC without the cloud commitment: SAP ties Joule for ECC and on-premises systems to moving at least half of maintenance spend to the cloud. Custom Joule skills on SAP BTP or a Claude layer over SAP data via MCP give AI on ECC without that commitment.
  • AI agents for invoices, quotes and reporting: Business-process agents that prepare the work and leave the decision to a person: incoming invoices matched to orders, quotes drafted from ERP data, customer replies with CRM context, month-end reporting. Every step audited.
  • AI governance and the EU AI Act: An AI policy your legal team will sign, data protection impact assessments, EU AI Act classification of each use case and role-based training. What the August 2026 obligations mean for a company already using AI.

How it works over SAP and Microsoft 365.

The agent sees only what the employee sees. Access is decided by the permissions in your systems, not by the model.

  1. Ask: An employee asks in Claude Desktop, or in Excel or Outlook.
  2. Plan: Claude in Frankfurt reads the request and decides which tools it needs, for example late deliveries in SAP.
  3. Fetch: The gateway checks the role and calls SAP and Microsoft 365 under the employee's own account. Steps 2 and 3 repeat until the agent has what it needs.
  4. Answer: Claude prepares the answer from the returned data, with sources, and sends it back to the employee.
  5. Approve: Only after the employee confirms does the agent write to SAP or send the email, under their name. Everything is logged.

your company AI

  • Agents by role: purchasing, production, sales, finance, quality, maintenance
  • Tool gateway: allows tools by role and logs every SAP call with the user's name
  • Approval: writes and sends only after a person confirms

Claude on Amazon Bedrock

Frankfurt: your AWS account in the EU, no prompt retention, sees only the request and what the tools return

SAP

our MCP connector · the employee's SAP account · your roles apply

SAP MM: Purchasing & inventory

  • stock and reservations
  • late deliveries
  • open purchase orders
  • purchase requisition (write: Every write happens only after a person confirms, under their name.)

Which suppliers are late with confirmations this week?

SAP PP: Production

  • production order status
  • missing components
  • capacity check
  • rescheduling proposal (write: Every write happens only after a person confirms, under their name.)

Can we release order 4711, or is a component missing?

SAP SD: Sales

  • order and delivery status
  • delivery dates for customers
  • customer history
  • inputs for quotes

When will the customer's order ship?

SAP FI/CO: Finance

  • costs per order
  • plan versus actual
  • open receivables
  • report for management

Where is order 4711 over plan?

SAP QM: Quality

  • inspection lots
  • complaints and notifications
  • 8D inputs
  • quality notification draft (write: Every write happens only after a person confirms, under their name.)

Which complaints are still open this month?

SAP PM: Maintenance

  • breakdowns and machine history
  • maintenance plan
  • maintenance notification draft (write: Every write happens only after a person confirms, under their name.)

How often has press 3 failed this year?

Microsoft 365

official connector · the employee's own access

Outlook

  • finds and summarises mail
  • finds the thread for an order
  • draft reply (write: Every write happens only after a person confirms, under their name.)

Summarise what the supplier wrote about the delay.

Teams

  • summarises chats and meeting transcripts
  • finds what was agreed
  • message to the team (write: Every write happens only after a person confirms, under their name.)

What did we agree on Tuesday's call?

SharePoint & OneDrive

  • finds contracts, standards, tenders
  • compares two documents
  • saves output to the library (write: Every write happens only after a person confirms, under their name.)

Which tender requirements can we not meet?

Excel, Word, PowerPoint

  • analysis and charts from SAP data
  • reports and minutes
  • presentation for management

Build the monthly overview for management.

audit

The gateway logs every SAP call with the user's name. The logs go to your SIEM together with the Microsoft 365 and AWS audit logs.

Every write happens only after a person confirms, under their name.

Get this architecture drawn for your company.

A two-page proposal for your SAP and Microsoft 365: the architecture, what the agent will do in each module, and what one year can bring. Request the proposal

Same team, more output.

AI is more than saved time. With it, people prepare more bids and analyses, serve more customers and take on work they would otherwise need a colleague for. Set your team.

More output per person with AI: Latest measurements: 26% more completed tasks (field trials, 2024) and 33% more output in every hour worked with AI (St. Louis Fed, 2025).

What the trials showed

Where the freed time goes

  • More bids: Tender analysis goes faster, so the same team answers more tenders.
  • Fewer stoppages: Late deliveries and missing components show up early, not on the day of production.
  • Hands untied: The agent does the searching, retyping and reports. People spend the time on customers, offers and decisions.
  • Faster answers: Delivery dates from SAP without searching, a morning overview for management.

This applies to work AI helps with: bids, analyses, documents and communication. The trials ran in other industries. Revenue grows when the extra output goes into sales and there is demand for it. A pilot measures the real number in your company.

Technology and controls.

A data-handling map

Where prompts, files and logs are processed, how long they are retained, which subprocessors are involved. Written down, per component.

An audit record you can inspect

Source data, model and version, proposed action, who approved it, what changed. Your auditor sees the same trail we do.

A documented exit

How agents are stopped, credentials revoked, configurations exported and processes returned to manual operation. Agreed before go-live.

How we deliver.

  1. Assessment: Workshops with leadership and IT, platform choice, first processes.
  2. Architecture: The AI layer with sign-in, audit and data masking, in your environment.
  3. Implementation: Two to four processes with measurable impact.
  4. Enablement: Role-based training on your team’s real tasks.
  5. Operations: Monthly oversight, evaluation, new automations.

The team.

We are a senior collective with decades of combined experience across artificial intelligence systems, machine learning, enterprise architecture, cloud infrastructure, process automation, ERP and SAP.

We have designed, built and operated systems where reliability, security and change control are non-negotiable. Senior experts stay close to every engagement, from the first decision to stable operation.

Talk to us.

A 30-minute consultation on your scenarios. You will leave with a concrete assessment, whether we work together or not.

info@iteractive.ai · +421 905 452 325