GoldenrozBook a call

A customer service chatbot that answers from your knowledge base and your order data

A customer service chatbot is software that answers customer questions in chat, using your help content and the customer's own order or account data, and passes the conversation to a person when it cannot resolve it. Goldenroz builds custom support chatbots for web chat, WhatsApp and email, connected to your help desk and back-office systems. Projects start at $25,000, fixed price per phase after a free one-hour scoping call.

From $25,000, fixed price per phase

Updated 2026-11-14. By Shai Goldenberg

What usually goes wrong

  • Three channels, three different answers
    Web chat quotes the help center, the WhatsApp line quotes whoever is on shift, and email quotes a saved reply from last year.
  • The bot cannot see the order
    "Where is my order" gets a link to the shipping policy. The bot has no access to the order system, so every question about a specific purchase still becomes a ticket.
  • Tickets arrive empty
    When the bot gives up, the agent gets a ticket with no order number, no category and no transcript.
  • Support in one language, customers in several
    Messages arrive in Spanish, Portuguese or Hebrew, and the team answers in English through a translation tab.

What is included

  • Contact reason review
    We read a sample of your real tickets and chats and rank contact reasons by volume and by how safely each can be answered without a person.
  • Knowledge base answers
    Replies drawn only from approved help content and policies, with the source article recorded for each answer.
  • Order and account lookups
    After the customer is identified, the bot reads their order, delivery, subscription or invoice data through your APIs and answers about that record.
  • Ticket creation in your help desk
    When a person is needed, the bot opens a ticket with category, customer ID, transcript and a summary, in the queue your routing rules expect.
  • Web chat, WhatsApp and email
    One set of answers and rules across the channels you use. Email replies can go out directly for simple cases or be drafted for an agent to send.

What a customer support chatbot should and should not handle

A customer support chatbot earns its place on questions with a known answer: delivery status, return eligibility, plan details, how to change an address, a copy of an invoice.

It should step aside on complaints, billing disputes, anything involving a safety or legal issue, and any customer who asks for a person. Those triggers are written down before the build starts. The customer can always reach a human, and the bot never pretends to be one.

This page covers bots that answer and route. If you want software that also issues refunds, edits orders or works tickets on behalf of your agents, that is a wider scope with approval steps, described on the AI agents for customer service page.

An AI customer service chatbot, measured honestly

Vendors quote resolution percentages. We do not, because the number depends on your contact mix and on how much your systems expose through APIs. An AI customer service chatbot can only resolve what it has the data to answer.

What we do instead is define the measure with you. A conversation counts as resolved when the customer got an answer and did not come back on the same issue within a period you choose. A conversation the customer abandoned is not a resolution. We report resolved, escalated and abandoned separately, and your team reviews a sample of transcripts each week.

A multilingual chatbot without a separate bot per language

Current language models read and write most major languages, so a multilingual chatbot does not need a copy of the flow for each one. The bot replies in the language the customer used, from one knowledge base.

Two things need care. Policy wording with legal weight should be translated and approved by your team, not improvised by a model. And handoff has to follow the language: a Spanish conversation goes to a Spanish-speaking agent, or arrives with a translated summary.

How the work runs

  1. Scoping call

    Before the project

    We review real transcripts, pick the intents worth automating and agree what the bot must never do. You get a fixed price.

  2. Prototype

    About 1 to 2 weeks

    A working bot on the top few intents, tested against your past conversations in a private channel.

  3. Production build

    Roughly 4 to 8 weeks

    System integrations, fallbacks, handoff, logging and channel setup. The range depends on integrations and on WhatsApp account approval.

  4. Limited launch

    At launch

    The bot handles a share of conversations or selected hours first. Your team reviews transcripts daily.

  5. Widen

    After launch

    New intents are added in order of volume, each tested against the conversation set.

What it costs

Each phase has a fixed price agreed before work starts. Model usage and WhatsApp messaging are billed by the providers to your own accounts.

See full pricing

  • Support or ordering bot
    Fixed price per phase
    From $25,000
  • Multi-channel or multi-brand platforms
    Scoped per project
    $100,000 and up
  • Scoping call
    Ends with a written scope and a price
    $0
  • Model and WhatsApp usage
    Billed to your own accounts
    Paid to the providers

Who it is not for

  • Bots designed to keep customers away from a person. Every bot we ship has a visible handoff.
  • Medical, legal or financial advice with no human review.

Who does the work

CTO and technology partner

Shai leads the technology side: business software, SaaS and mobile apps, payments and escrow systems, Postgres security, AI agents built with Claude Code and MCP, and IT and infrastructure projects with senior infrastructure engineers.

Questions buyers ask

How much does a custom customer service chatbot cost?

Projects start at $25,000 and are fixed price per phase after a free one-hour scoping call. Multi-channel or multi-brand platforms run $100,000 and up. Model usage and any WhatsApp messaging fees are paid to the providers from your own accounts. Details are on the pricing page.

Which help desk systems can the chatbot connect to?

Any help desk with an API can be connected, and the scoping call confirms what yours exposes. The bot creates and updates tickets there. We are independent of help desk vendors and hold no partner status with them.

How do you stop the chatbot from giving wrong answers?

It answers only from approved content and live data from your systems, and it declines questions those do not cover. Prices, dates and policy limits are read from the source and not generated. A test set of real conversations runs before every release.

How long does it take to launch?

A working prototype on the top few contact reasons typically takes one to two weeks. Production takes roughly four to eight weeks, depending on how many systems are connected.

Should we buy a support bot or build one?

Buy when your questions are standard help-center content on a web widget. Build when answers depend on your own order and account data, when you need the same bot on WhatsApp and email, or when per-resolution fees are high at your volume. The parent AI chatbot development page covers the build in full.

Guides on this topic

Talk to the person who will do the work

One hour, no sales team. You talk to Shai Goldenberg, the CTO, about your systems or your product, and you leave with a clear next step and a price range.

Or send the details

Book a 1-hour consultationWhatsApp us