This website uses cookies to ensure you get the best experience
Leave a request
Fill in your details and we will contact you shortly.
Karcher
Purina
AVON
Purina Friskies
question quiz
7
Breed recognition
messages
2 343
audio messages
78 434
photos
175 435
Alice interactive
Promo code finale
Personalized feeding
CRM integration
Sila Sveta
Schneider Electric
users
1,044
Engineer tool
messages
approved
1 140
997
Typo tolerant
Smart search
Support relief
Action hub
receipts
20K+
Purina Friskies
Chatbot in the widget on the Purina promo site and main site
Goal:
Centralize the Friskies promo campaign flow on the website and reduce the burden on support.
Solution:
A chat widget on the promo page and the main Purina site that guides users through the campaign (quiz → review → receipt upload), answers FAQs, and escalates to an operator when needed.
Outcome:
The bot became the hub for key on-site actions; during the campaign users uploaded 20,000+ receipts
AVON
Chatbot for selecting gifts for men
Goal:
Deliver a fast, shareable gift-picker experience in VK for a women-focused Feb 23 campaign
Solution:
A chatbot featuring Timur Rodriguez: users upload a man’s photo, AI classifies the "type" and recommends a matching Avon product with humorous, personalized commentary.
Outcome:
Strong campaign engagement — 175,435 photos, 78,434 audio messages, and 2,343 messages recorded during the activity.
Schneider Electric
Bot for searching the technical reference guide
Goal:
Speed up handbook search and help engineers quickly find the right products and specs.
Solution:
A chatbot that accepts queries by product name/properties, searches the knowledge base and returns a ranked list of relevant answers; includes typo-tolerant trigram search and a structured table built from the original PDF.
Outcome:
A pilot "search engine" for highly specialized technical content — a practical tool for design engineers.
Sila Sveta
Chatbots for sending messages "into space" (Signal Festival)
Goal:
Let festival visitors send messages to an installation and "into space" while keeping moderation and throughput under control.
Solution:
Two Telegram bots: the user bot collects messages and runs automatic checks, while the second bot routes content for manual moderation in the client’s channel.
Outcome:
In 3 festival days 1,044 people used the bot; users sent 1,140 messages, with 997 approved and "sent to space."
Purina
Dog food selection: breed recognition by photo
Goal:
Support Purina "21 Days" participants and help dog owners transition to a new food with personalized guidance.
Solution:
A VK chatbot integrated with Purina CRM: identifies breed from a photo, collects pet details, builds a feeding plan, enrolls users into messaging, and sends tips (email/push) plus a discount coupon.
Outcome:
A personalized nutrition journey with ongoing touchpoints that keeps users engaged throughout the program.
Karcher
Quiz in Alice about cleaning (Skill)
Goal:
Engage users with an interactive experience, teach cleaning tips, and finish the flow with a discount promo code.
Solution:
An Alice skill built as a quiz: 7 questions with 4 answer options, helpful feedback after each choice, and a compact numbered list format so all options fit in one message.
Outcome:
A light "learn + play" promo mechanic that keeps attention and naturally leads users to the final promo code.
AI agents for your processes
Built into systems and run under metric-based control
Vikulya — corporate RAG agent
Fast search across company knowledge with sources and excerpts. "Average impact values from implementations
EORA Operator
handles thousands of calls and messages simultaneously, integrates with CRM, and takes over routine work
EORA HR Agent
Automatically handles up to 70% of primary HR tasks; speeds up hiring and reduces recruiter workload
EORA Finance Agent (XinData)
Answers based on financial data in real time. Access control and interpretability are mandatory
Monitoring, regular improvements, expansion of integrations.
Commissioning
5
calculate ROI, the cost of a production rollout and support
Impact calculation and scaling plan
4
run the pilot, collect statistics, improve scenarios
Pilot + quality criteria
3
review sources, configure roles and permissions
Connect data and access rights
2
choose scenarios where impact is easiest to quantify (time, workload, conversion, SLA).
Diagnostics and process selection
1
Managed implementation process
Focus on KPIs and economic impact before you move to scaling.
Vikulya shows sources and forms an answer based on your documents. XinData works via authorization and provides read-only access to data.
AI agents work with data and systems, not just text
Sources in answers, access rights, monitoring of quality and costs
Control and improvement
CRM/ATS/financial systems, APIs
Actions via integrations (where needed)
Documents, policies, knowledge base; answers only from your sources
Company knowledge context (RAG)
Chat, voice, corporate messenger
Request from a familiar channel
To ensure AI works reliably, it must be measured and managed like a business process.
Production control and analytics
Link AI spend to business outcomes and KPIs
Understand what drives the cost of each process
Control token usage and infrastructure spend
Know which models and tools are used in each scenario
Track agent actions and performance in real time
We embed into your existing infrastructure
Deploy AI agents in any cloud, channel, or model and integrate them directly into critical workflows of your contact center, CRM, ITSM and other enterprise systems.
