Atefe Nejatian Ghorashi

04

I am CEO of Innovative Creative AI Kish (ICAI). The company is registered in the Kish Free Zone and works in three connected fields: platform development, AI advisory, and practical training.

I take an executive part in the work. I carry out the system architecture and the product decisions myself — from infrastructure that serves several organisations, to connecting the model's answers to reliable sources, the design system, and putting the product into operation.

The decisive part of the work, for me, is this: identifying where a model genuinely helps, refusing a confident answer that has no basis, and testing a claim instead of merely stating it.

Kish Island · Innovation Tower

Role: CEO and Chair of the Board
Company: Innovative Creative AI Kish (ICAI)
Location: Iran, Kish Island — Innovation Tower, Unit 412
Registration: Kish Free Zone
Focus: Applied AI platforms · serving several organisations from one system · AI strategy
Languages: Persian (native) · English

Capabilities

TypeScriptNext.jsReactNestJSNode.js PostgreSQL / PrismaDockerCaddy / VPS ops Applying language models and designing prompts Connecting model answers to an organisation's sources Cloud software architecture for several organisations Multilingual and right-to-left (Persian, English, Arabic) Design system Product strategy

01

Work, from idea to a working system

Two platforms in development, and one course underway.

A platform for tourism on Kish Island. It learns from tourist behaviour, business activity and events on the island, and offers trip suggestions, planning and a travel assistant. Businesses have an analytics dashboard, and AI use on the platform is controlled.

A business has no access to a tourist's financial or behavioural patterns. This limit is part of the platform's architecture and cannot be changed in the settings.

Infrastructure: Next.js 15TypeScriptTailwind 4PostgreSQLPrismanext-intl
Scope: Three languages (Persian, English, Arabic) · three in-house design systems: Aurora, Atlas, Nexus
Link: kishpulse.ir ↗

One platform serves both the hotel guest and the manager. A separate version is not built for each hotel; the same engine is configured with that hotel's capabilities. There are two separate applications: one for the guest and one for the manager.

A guest's answer comes only from that hotel's information. If a subject is not in the hotel's data, the system does not produce it.

Infrastructure: TypeScript monorepoNestJSNext.jsDockerPostgreSQL
Scope: Three capability levels: Professional, Enterprise, Ultimate
Link: cnshotel.com ↗

A staged, project-based course. Each stage ends not with an exam but with something usable: first a personal assistant, then a tool for solving a problem, and finally the building of a system. The topics include prompt design, research method, data literacy, and recognising where ready-made tools are no longer enough.

Practice is done on the participant's own problem. The outcome of the course is not a ready-made classroom example.

Stages: Foundation (4 hours) → Explorer (24 hours) → Creator (30 hours) → Builder
Format: Individuals and in-house teams

02

Advisory: where to build, and where not to

The result of a collaboration is an opportunity map. Ideas are set side by side by value and feasibility, so it becomes clear what should be built today, what needs data that does not yet exist, and what is better left unbuilt.

In many cases, using an existing tool is preferable to a custom build. When that is the conclusion, I say so plainly.

Sometimes the right result of advisory is: do not build it at this stage. That, too, is a valid outcome.

Feasibility → Value →

03

Approach: the order matters more than the list

01
Understand the problem first, then propose
If the problem is unclear, two or three specific questions are asked: volume of work, channels, where time is lost, and what data actually exists. Ten questions are not asked at once.
02
Connect the problem to work already done
If the problem resembles work already solved in KishPulse or CNSHotel, that is said, and exactly what it does there is specified. The difference between a precise answer and a general one lies in this.
03
State the boundary, plainly
What can be carried out today is specified, along with what depends on data that may not yet exist, and where AI is not the right tool. In many cases, an existing tool is preferable to a custom build.
04
The next step
Usually a short conversation, not a price proposal. The scope of the work is set after the problem is agreed, not before.

If the assessment shows that AI is not worthwhile for this business in its current state, that answer is stated plainly.

05

Tell me about the business, not the technology.

I am available for advisory, for taking part in platform development, and for speaking. The first conversation is short and costs nothing. If AI is not the right answer for that business, I say so plainly.

Direct line+98 915 510 0769

EmailNejatian@aikish.com

Office+98 76 4456 0987

Companyaikish.com

Send an email Call directly →