Skip to content
hevsel

03 / ai

We automate business processes with AI

We build AI assistants and automations that take repetitive work off your business's hands. What matters stays your decision: the system prepares, you approve.

01 / use cases

What we can automate

  • Customer assistant

    Answers questions that come in through your site or WhatsApp and takes appointments and requests. Anything it cannot answer is handed to you.

    customer question → an answer or a structured request

  • Quotes and pricing

    Structures an incoming request, drafts a quote from your pricing rules and puts it in front of you for approval.

    free-text request → draft quote

  • Document and invoice processing

    Reads the fields you need from a PDF or a photo of a document and records them in your system.

    email + PDF invoice → accounting entry

  • Email and request routing

    Classifies incoming emails and requests by topic and routes them to the right person or queue.

    incoming email → category + owner

  • Stock and order alerts

    Notices stock that falls below a threshold and orders that run late, and notifies the person responsible.

    stock and order data → alert notification

  • Reporting

    Prepares daily or weekly summaries on its own and sends them to the people you choose.

    business data → summary report by email

  • Content and product data

    Drafts product descriptions, translations and search engine copy. You approve everything before it goes live.

    product attributes → text draft, published after approval

  • System integration

    Moves data between systems such as accounting, e-commerce and CRM, so nobody types the same information twice.

    e-commerce order → accounting ledger entry

02 / process

How a fully automated process works

A process can run from start to finish on its own. At every stage it is clear what belongs to the AI, what belongs to a rule and what belongs to you.

  1. trigger

    An email arrives, a form is filled in, a scheduled time comes or a record changes, and the process starts on its own.

  2. understanding

    The AI reads and structures the data: it pulls fields out of text, parses the document, classifies the request.

  3. decision rules

    Pricing and business rules live in code. The AI does not make up numbers or conditions.

  4. human approval

    Where it matters, the action is put in front of you and does not go ahead until you approve it.

  5. action and record

    The email, PDF or notification goes out, the record is created, and everything that happened is written to a log.

03 / live example

A live example: this site's quote system

This very site is an example of automation. When someone asks for a quote, this is what happens:

  1. request

    The form, or the optional "describe it in your own words" assistant, turns what the customer wrote into a structured request.

  2. price

    The draft price is not computed by the AI but by a rule-based price engine working from a catalogue.

  3. approval

    The draft goes to the site owner, who edits the items, approves it and sends it.

  4. customer

    The customer gets a private quote page and a PDF, and can accept the quote, make a counter-offer or ask a question.

  5. follow-up

    If there is no reply, reminder emails go out automatically after 3 and 7 days. A quote whose validity has run out is closed automatically.

04 / principles

Our principles

  1. The AI never sets a price or makes a commitment

    Numbers and terms come from rules you define.

  2. Critical steps need a human's approval

    Anything that reaches a customer or cannot be undone goes through your approval.

  3. Your data stays in your accounts

    Data is kept in the customer's own accounts and processed in line with the Turkish data protection law (KVKK).

  4. Every automated action is logged

    What was done, when and why can be traced afterwards.

  5. The assistant does not pass itself off as a person

    The person on the other end knows from the start that they are dealing with an AI.

05 / getting started

How we start

  1. process review

    We look at how the work runs today and decide together which steps are suited to automation.

  2. pilot automation

    We build one chosen process end to end and try it on real work, then review the result together.

  3. roll-out

    We extend what works to other processes, reviewing approval steps and logs together.

quote

Let's pin your project down in a written quote.

Tell us briefly what you need, and we'll send you the scope, timeline and price in writing.