SysaiQ
FA Start a project

[ SYSAIQ—SERVICE / AI-AGENT ]

AI agents and chatbots

An assistant that answers from your own data and documents, lives inside your site or system, and takes over repetitive work.

Start a project

Who this is for

  • Businesses that receive many repeated customer questions every day and spend staff time answering them.
  • Organisations with large document sets, regulations or catalogues where staff waste time looking for answers.
  • Stores, clinics and agencies that want customers guided at any hour, with their requests recorded.
  • Owners of systems who want an agent to perform defined tasks such as logging requests, searching or summarising.

Problems it solves

  • Wrong or made-up answers: with RAG, the model answers only from your approved documents and data, and the source of each answer is visible.
  • Answers limited to office hours: the assistant is always available and hands off questions beyond its scope to a person.
  • Repeated staff questions: an internal assistant answers from regulations and documentation, with references.
  • Repetitive manual tasks: an agent can work with your systems — log a request, fetch information, produce a summary.
  • Concern about data: the data given to the model is scoped and documented, and sensitive information stays outside it.

Deliverables

  • A chatbot or agent with a RAG knowledge base over your documents and data
  • A chat widget for your site, or connection to your system and channels
  • Knowledge-base admin panel: add and edit content without a developer
  • Conversation logging, hand-off to a person, and lead or request capture
  • Behaviour rules, answering scope and answer-quality tests
  • Architecture, data-flow and privacy notes
  • Full source code, documentation and every credential handed over in your name

How the work runs

  1. 01

    Discovery

    We gather the real questions customers or staff ask, the documents you have and the tasks the agent must perform, and define its answering scope.

  2. 02

    Written proposal

    You receive the architecture, data sources, channels, quality criteria, milestones, schedule and cost in writing.

  3. 03

    Contract

    The contract is signed in person in Qazvin or electronically; work starts after signature and deposit.

  4. 04

    Staged build

    First the knowledge base and a test version; you try it with real questions and we refine the answers together. Then connection to your site or system.

  5. 05

    Handover and training

    The assistant goes live on your server, and you are trained on adding content, reviewing conversations and managing API cost.

  6. 06

    Support

    Bugs are fixed free of charge during the warranty period; answer improvements and new capabilities continue under a separate agreement.

Timeline

Duration depends on the volume and quality of source documents, the number of channels and systems to connect, and whether the agent only answers or also performs actions. A significant share of the time goes into testing with real questions and refining answers; your involvement at that stage sets the pace.

The milestone schedule is stated in the written proposal.

How cost is calculated

Cost is calculated from:

  • The volume and type of source documents and data
  • Answering only, or performing actions (agent)
  • The number of channels and connections to existing systems
  • Knowledge-base admin panel and conversation reports
  • A bilingual assistant
  • The level of support and ongoing improvement after handover

Included: architecture design, knowledge-base build, development, quality testing, deployment, training and the bug-fix warranty.

Not included, paid separately in your name: language-model API usage (such as OpenAI), which depends on traffic, plus server and domain.

The final figure is stated only in the written proposal.

Read more about how cost is calculated →

Frequently asked questions

With RAG, the relevant parts of your documents are retrieved before every answer and given to the model, which is instructed to answer only from them. If the answer is not in the documents, it says so or hands off to a person. Quality tests verify this behaviour before handover.

Documents and the knowledge base stay on your server. To produce an answer, only the parts relevant to that question are sent to the language-model API. What data is in scope is defined in discovery and written into the data-flow document; sensitive information stays outside it.

Yes. Current language models handle Persian well, and the assistant can be bilingual. Persian answer quality is tested and tuned with your real questions.

This cost is in your name and depends on usage, so it has no fixed figure. At handover we show the usage dashboard and cost controls (daily limits, shorter answers, caching).

Yes — that is what makes it an agent. It can work with your systems' APIs: log a request, fetch an order status, book an appointment or produce a summary report. Every action follows defined rules and, where needed, a human confirmation.

Yes. The SysaiQ chat assistant runs on this site's own knowledge base with the same architecture; you can try it right now.

Let us talk about your project

Write a few lines about your business and the problem you want solved; you receive a written proposal before any commitment.

Start a project All services