Menu
AI Development · Choosing a chatbot type

Which chatbot a business needs: scripted, AI or hybrid

Three types solve different problems and cost differently. Where a script beats a language model, where it's the other way round, and why a hybrid usually turns out to be the right answer. A scripted bot follows branches written in advance: predictable, cheap, and incapable of inventing a discount that doesn't exist — but irritating the moment a question doesn't fit the buttons. A bot built on a language model answers freely and understands how people phrase things, which is exactly why it can state something untrue with confidence unless it's constrained to a knowledge base. A hybrid combines them: a strict script wherever money, deadlines or obligations are involved, free-form answers on reference questions, and a handover to a person wherever the bot isn't sure. This page helps you choose for your own case and shows what to look for in vendor proposals. Building one — with a knowledge base and CRM handover — is a service, described separately.

This is how to choose a type. Building one with a knowledge base and CRM handover is a service.

AI assistant service and pricing
Why this matters

What happens without this work

🔘

Button bots annoy users

Customer asks a question — bot offers 5 buttons 'Pick a topic'. Conversion drops 60%.

🌙

Night inquiries lost

Managers respond only in business hours. Night inquiries go to competitors.

📞

Expensive support

Team answers the same questions 50× a day. Bot resolves them in seconds.

🎯

Unqualified leads

Manager spends 30 minutes realising a lead isn't a fit. Bot qualifies in 2-min dialogue.

📊

No data on questions

Don't know what clients ask — can't improve product and FAQ section.

What's included

What the work consists of

Built on current language models

The bot understands a question asked in plain words, not by template. The exact model is chosen per task and budget — they change every few months.

Company knowledge base

We load all information about products, pricing, FAQ — the bot answers using your data.

Lead qualification

The bot asks the right questions, determines the client type and passes to manager with context.

Multi-channel

One bot works in Telegram, WhatsApp, on the website and other channels simultaneously.

Dialogue analytics

You see what users ask about, where the bot struggles — data for product improvement.

Fast launch

Basic AI bot for a website — from 2 weeks. Without complex integrations, faster launch possible.

Methodology

How it works, step by step

01

Discovery & audit

We map your current AI/SEO landscape: what works, what's broken, where competitors lead. Tools: GSC, Ahrefs, MarketMuse, custom NEXUS agents for AI-mention monitoring.

02

Strategy & roadmap

90-day execution plan with clear KPIs, weekly milestones, and dependency mapping. You see exactly what we do each week and what success looks like.

03

Implementation

Our team executes — content, technical, link building, AI-specific work. Your team focuses on the business. Weekly status calls keep you informed without micromanagement.

04

Measurement & iteration

Real metrics every week: rankings, AI citations, traffic, conversions. We don't hide behind vanity metrics. If something underperforms — we pivot fast.

05

Optimisation at scale

Once a tactic proves out, we automate it (NEXUS agents) and scale across all your assets. Manual work doesn't survive long with us.

06

Strategic review (quarterly)

Every 90 days: full strategy review, competitor benchmark, AI landscape updates, next-quarter planning. You always know where we're heading and why.

Pricing

What it costs

Project cost

after a brief scoped per project

There is no boxed price here: scope depends on which processes we automate, which systems we connect and what has to happen when something fails. We quote after a short review.

Reference figures for one-off work are on the pricing page. Naming an exact number before the task is understood would be guessing.

What the price depends on

  • What exactly gets automated and how many scenarios
  • Which systems have to be connected
  • Whether data and a knowledge base already exist
  • Whether you need support and iteration after launch

We treat the work as finished only when there's an error log and a notification that reaches a person.

Comparison

AI/GEO vs traditional SEO

Both have their place. Most clients need a hybrid. Here's how they differ.

Traditional SEO

  • • Targets Google, Yandex, Bing organic
  • • Keyword-centric: title, H1, meta, content
  • • Link building, technical health, content velocity
  • • Results in 4-8 months, compounds over years
  • • Mature tooling: Ahrefs, Semrush, GSC

AI / GEO NEW

  • • Targets ChatGPT, Perplexity, Gemini, AI Overview
  • • Entity-centric: brand recognition by LLMs
  • • Citation engineering, Knowledge Graph, PR
  • • Results in 2-6 weeks (faster cycle)
  • • Emerging tooling: NEXUS, custom monitors

Most of our clients run both in parallel: SEO for compounding long-term moat, AI/GEO for capturing the new generation of AI-mediated search.

What this work delivers

Scenario and task analysis
Company knowledge base preparation
Bot development and configuration
Website or messenger integration
Testing and refinement
Team training + support

Frequently Asked Questions

How does your bot differ from standard solutions?

We build on GPT-4o/Claude with custom prompting for your niche. The bot understands your business specifics rather than giving generic answers.

Will the bot replace live managers?

No — it complements them. The bot handles 70–80% of incoming questions and passes complex cases to a manager. Managers only deal with qualified leads.

What about data privacy?

We support on-premise deployment with local Llama for sensitive data. For most clients OpenAI/Anthropic API with proper data handling agreements is sufficient and faster to deploy.

Free consultation

Start with Choosing a chatbot type

Get in touch