AI · Data · Automation

We turn 1C, industry systems and Excel into OLAP cubes and AI agents.

We deploy unified data marts and digital workers for mid-market business. Manufacturing, restaurant chains, fleets, logistics — without rewriting your accounting systems, inside your own perimeter.

View cases → Discuss a project

production projects

12+

data sources

AI agents on the platform

100%

on-premise / private cloud

Services

A full-cycle IT partner for mid-market business.

From data marts to networks and mail — we cover the whole infrastructure, inside your perimeter, with no vendor lock-in.

Data & analytics

Unified data marts and OLAP cubes from 1C, industry systems and Excel. Role-based dashboards: owner, finance, operations.

Learn more →

AI agents

Digital workers on top of your data: collections, outbound calls, review analysis, HR parsing. Local LLMs — data stays in.

Learn more →

Mail & comms

Corporate mail integrated with Active Directory, migration off ageing platforms, zero-touch client setup, IP telephony.

Learn more →

Networks & access

Site-to-site VPN across offices and branches, resilient uplinks, MikroTik setup, secure access to the services you need.

Learn more →

Infrastructure

Servers and virtualization without excess licensing, Active Directory, monitoring, backups, secrets management.

Learn more →

Infrastructure as code

Your whole infrastructure in git and under version control. Automated operations, runbooks, living documentation.

Learn more →

How we work

From audit to support — in stages.

No big upfront payments, no "rewrite everything from scratch". We start small, prove the value, then scale.

Audit

We map out which systems you run and where the data or infrastructure pain is, and set the scope and cost of the work.

Pilot

A quick result on one narrow area: a cube, an agent or a service. We prove the value in practice.

Rollout

Phased deployment inside your perimeter, with documentation and training for your team.

Support

Monitoring, updates, evolution. You're never left alone with the system.

Trusted by

Cases

Three projects, three industries, one architecture.

Real deployments in the data & AI track. Details anonymized — the substance, stack and outcome are intact.

01 Manufacturing · Distribution production

FMCG manufacturing and distribution

An IT zoo: several 1C databases of different configurations, a mobile app for field staff, operations run in Excel. The finance team kept budgets and cash-flow by hand in spreadsheets. Pain #1 — receipts were entered twice into separate systems.

Incremental OData → SQLite extraction framework across all 1C systems
AI-agent roadmap: outbound calls, priority-based collections, vacancy parser, legal bot, secretary
On-premise LLM infrastructure with no data sent to the cloud
OLAP cube with auto-reconciliation between accounting systems and the labelling system

any 1CODataPython 3.14SQLiteon-prem LLMExcel watchDataLens

pipeline.yml

# one loader across all systems
sources: ["1C", "1C", "1C", "Excel", "…"]
load_mode: "incremental"
target:    "OLAP_cube + AI_agents"

agents:
  - sales_callbot
  - debt_collector
  - vacancy_parser
  - lawyer
  - secretary

status: ● production

02 Fleet · Logistics live

Analytics layer over four 1C databases

A bundle of Fleet Management + Accounting + Payroll + Document Flow. Waybills, fuel, driver payroll, revenue per vehicle, repairs, postings, document movement. Before — manual exports to Excel, with accountants and economists counting the same numbers differently. Now — a single source of truth, scheduled refresh, ready-made marts for BI.

Incremental 1C OData loader with watermark and gap back-fill
5 star schemas: waybills, fuel, payroll, revenue, extra analytics
Full history per vehicle, driver and counterparty in one place
A template reusable across other 1C configurations

1C:FleetAccountingPayrollDocument FlowOData RESTStar schemaPython ETLDataLens

star_schema.sql

-- 5 fact tables
fact_waybills
fact_fuel
fact_payroll
fact_revenue
fact_extras

-- incremental by watermark
WHERE modified_at > :watermark

latency: < 5 min
status:  ● live

03 Restaurant chain · HoReCa pilot

A cube and 9 AI agents for a food-service chain

Three planes — operations (any restaurant/cafe software), finance and HR (any 1C configurations) — merged into one cube with auto-reconciliation per location. Excel is the fourth source: payment calendar, budget plans, price lists. On top of the cube — 9 AI agents: collections, review analysis, HR parser, booking, auto P&L commentary and more.

OLAP cube with marts for budget, food cost, payroll %, checks, suppliers, HR
Two-way Excel integration: watch folder + auto-generated reports
9 AI agents tailored to the client's individual cases
5 role dashboards: owner, CFO, manager, chef, HR

HoReCa softwareany 1CODataPostgreSQLAirflowDataLensLLM Agents

cube.config

cube:
  marts:    8
  network:  "restaurant & cafe chain"
  refresh:  "every 2h"

ai_agents:
  - debt_collector
  - review_analyst
  - hr_parser
  - booking_bot
  - pnl_commenter
  - … +4

status: ● pilot

04 Mail · Active Directory live

Corporate mail migration with AD integration

A group with several mail domains on an ageing, end-of-life platform. Moved to a modern self-hosted stack with single sign-on via Active Directory, anti-spam and zero-touch client setup — no cloud, inside the client's perimeter.

Hundreds of mailboxes migrated with parallel sync and a rollback plan
Hybrid auth: users from Active Directory + a local fallback database
Zero-touch setup for Outlook / Thunderbird / mobile (autoconfig, autodiscover)
Anti-spam, DKIM/SPF/DMARC, auto-renewing TLS certificates

PostfixDovecotActive DirectorySOGoRoundcubeautoconfigDKIM

ad-sync.yaml

# sync users AD -> mail
source:  "Active Directory"
filter:  "OU=Staff"
sync:    ["login", "name", "quota"]
period:  "every 5 min"

auth:    AD -> fallback local
webmail: [SOGo, Roundcube]

status:  ● live

05 Networks · Branches production

Corporate VPN for a distributed group

A group of companies with sites across several regions. Connected offices and branches into one site-to-site VPN with resilient uplinks, policy-based routing and access for mobile staff.

Site-to-site VPN between offices and sites (OpenVPN + WireGuard)
Resilience: backup uplinks, automatic failover on ISP outage
Policy-based routing: the right traffic over the right link
MikroTik at the sites, centralized config and list distribution

WireGuardOpenVPNMikroTikPBRdual-ISPipset

routing.conf

# policy-based routing
tunnels: [wg0, wg1, ovpn]
sites:   10+ locations

uplinks:
  primary:   "ISP-1"
  failover:  "ISP-2"   # netwatch

policy:  fwmark -> table 100
status:  ● production

Stack

Under the hood.

An open stack, no lock-in to any single vendor. Runs inside the client's perimeter, no data leaves the premises.

1C + industry software

OData, REST, direct ODBC. Any 1C configuration, restaurant/cafe software, industry systems.

Python ETL

Incremental loaders in Python 3.14, configured via YAML, no code changes needed.

SQLite / PostgreSQL

SQLite for small setups, PostgreSQL for chains. ClickHouse on demand.

BI / DataLens

Yandex DataLens free in the cloud. Alternatives: Power BI, Superset.

LLM on-premise

Local models on the client's GPU server. Data never leaves the perimeter.

Excel bridge

Two-way integration: watch folder, validators, auto-generated reports.

Airflow / cron

Scheduled jobs, retry on failure, load monitoring.

Git + CI

Project code in Git, versioned SQL models, automated deploys.

Let's build a cube and AI agents for your perimeter.

We start with a short audit: which accounting systems you run, where the data pain is, and which case AI can close fastest. Then — a phased rollout.

Request an audit →

Let's discuss your task and terms · info@leverops.ru