AI and automation for business processes

Version 1.0 — Last updated: October 29, 2026

How can a company use AI safely in its own processes? Turkey-based Gramoni builds AI applications that run on your own data and processes, with every result going through human approval. Nothing is written to your systems without sign-off, and Workcube is not required. The first step is a free opportunity assessment that produces 3–5 concrete automation scenarios.

Gramoni is a Turkey-based software and technology company. It builds AI applications that run on your own data and processes, with every result going through human approval. The application works inside the business process rather than on a separate screen: it opens records, processes documents and reports status. Gramoni is a Workcube partner (Workcube is a Turkish ERP platform), but this service does not require Workcube.

What we build today

Everything below has been delivered before and can be repeated.

• Agents that write records with human approval. They don't just read: they open records, add notes and log time in your system. Every action that writes to the system goes through your approval.
• An identity-bound WhatsApp assistant. Employees ask about status or send documents over WhatsApp. The system verifies who the sender is and does not answer unknown numbers.
• Records from document photos. A photographed document becomes a structured record linked to the right parent record. The same document is not entered twice.
• A decision log from meeting recordings. Meeting and phone recordings become a decision log written into the project's own documents, not a pile of transcripts.
• A searchable knowledge base. Company documents are scanned, tagged and synced incrementally as they change. The knowledge base can be queried in chat or in a structured way.
• Multi-channel notifications. An event reaches each person through the channel they chose: in-app, browser notification, WhatsApp or email. The record is kept even when a channel is off.
• Screen automation and regression checks.
• Monitoring and alerts. Logs and system status are watched and turned into events someone can act on.

Why human approval is part of the design

An AI application can be wrong. That is why every action that writes to your system is tied to a person's approval. The application prepares the record; the decision and the responsibility stay with the person you authorise. The approval step is a design decision, not a limitation.

We don't treat this as a software-only job. Opening a process to AI means redesigning that process: which step is handed over, which stays with a person and who approves is agreed in writing.

Why ERP depth matters

Gramoni's core work is ERP, banking, production and warehouse processes, so we don't have to learn your processes and data from scratch. If you run Workcube, the application works directly on your Workcube records; see Workcube ERP consulting and Workcube custom development for our Workcube services.

If you don't use Workcube

We connect the application to your existing systems. There is one condition: the target system must be reachable, through an API or through database access you authorise. If neither exists, the scenario cannot be delivered, we make no promise, and we tell you so at the start.

Who it is for

• Companies that run their own ERP, production system or industry-specific software and have built up data.
• Operations with high-volume, rule-based work that is still done by hand.
• Companies where information is retyped, that is copied by hand, from one system to another.
• Processes where a decision waits for someone to read something and summarise it.

Who it is not for

• Processes with no data, or with data that cannot be accessed.
• Cases where the target system has neither an API nor authorised database access.
• "We just want some AI" requests whose output cannot be measured. We don't take these on.

What we don't do

• We don't train models on your data. We do not train models on your data.
• We don't promise to automate everything. We work out together which tasks can be handed over, and we tell you which ones cannot.
• We don't match data by name similarity. Accounting-grade data needs a key value.
• We don't write records without approval. Every action that writes to your system sits behind human approval.

First step: a free opportunity assessment

We look at your processes together in one session. The output is 3–5 concrete automation scenarios that your data supports today. For each one it states in writing which work goes away, where the data comes from and how the application will run.

We don't discuss price at the first contact. The commercial conversation starts after the assessment, with a pilot whose scope is written down; duration and assumptions are settled there.

Book a call for an opportunity assessment

How widely do companies in Turkey and the EU use AI?

According to the Turkish Statistical Institute (TÜİK), the share of Turkish enterprises with 10 or more employees using at least one AI technology rose from 7.5% in 2025 to 14.0% in 2026. It rises with size: 12.8% for 10–49 employees, 17.0% for 50–249 and 37.1% for 250 or more. Among enterprises using AI, the most common purpose was marketing or sales (51.0%), followed by R&D or innovation (46.3%) and production or service processes (43.4%).

In the European Union, 20.0% of enterprises of the same size used AI in 2025, up from 13.5% in 2024 (Eurostat). The most common use in the EU was analysing written language (11.8%).

enterprises using AI, Turkey 2026 (TÜİK)

14.0%

enterprises using AI, Turkey 2025 (TÜİK)

7.5%

enterprises using AI, EU 2025 (Eurostat)

20.0%

reason for not using: lack of expertise (TÜİK 2026)

72.3%

The reasons for holding back are clear too. Among Turkish enterprises that do not use AI but are considering it, the main reason is lack of relevant expertise (72.3%), followed by legal uncertainty about who is liable if AI causes harm (66.4%) and concerns about data protection and privacy (65.4%), according to TÜİK. The obstacle is mostly not the technology. It is that nobody has defined who does what, who approves and who is responsible.

Why human oversight is becoming the norm

The main international frameworks agree on one point: an AI system should not take the decision away from people.

• EU AI Act, Article 14. Regulation (EU) 2024/1689 requires high-risk AI systems to be designed so that natural persons can effectively oversee them while they are in use. The person doing the oversight should be aware of the tendency to over-rely on the system's output (automation bias), be able to disregard, override or reverse that output, and be able to stop the system in a safe state. The article applies to high-risk systems; its principle is a sensible guide for any AI application.
• OECD AI Principles. Adopted in 2019 and updated in May 2024, the principles have 47 adherents. They ask AI actors to keep a capacity for human agency and oversight, appropriate to the context and including risks from uses outside the intended purpose, and list accountability as a principle of its own.
• NIST AI Risk Management Framework. Released by the US National Institute of Standards and Technology on 26 January 2023, the framework is intended for voluntary use and is built around four functions: Govern, Map, Measure and Manage.

In Gramoni's applications, the rule that every action writing to a system goes through human approval is this approach turned into day-to-day practice.