Discuss a task
Independent IT consultancy · telecom · operations · AI

We make your systems observable, automate the routine, and apply AI where it pays off.

A private consultancy with production experience in contact centres, banking and industrial ERP. Delivered as a service or installed entirely inside your network.

Discuss a task
01 — Practice

Four areas of work

Telecom

Telecom and contact centres

Carrier and route quality, traffic balancing, NOC alerting, call-centre operations.

Operations

Monitoring and operations

ERP and licence usage, infrastructure health, incident routing and escalation.

AI

Applied AI

Agents, on-premises models, assistants over company knowledge, transcription, team training.

Finance

Banking and fintech

Lending platforms, ledgers, data marts and analytics for banks and microfinance.

02 — Open source

Contributions accepted by maintainers

21merged pull requests
9projects
197kGitHub stars, combined
13pull requests in review
pola-rs/polarsdask/daskurllib3/urllib3cjpais/Handytw93/Moletrycua/cuaRustAudio/rodiopypa/cibuildwheelcode-review-skill scipy · in reviewxarray · in reviewlitestar · in review
03 — Work

Systems we built and run

SystemDomainWhat it does
Carrier monitoring for a contact centretelecom19 countries, 90+ queues, 32 carriers, 5M+ call records; reaction time cut from hours to minutes.
ERP licence monitoroperationsLicence pool read every 15 minutes, with history and alerts to Microsoft Teams; cloud or on-premises.
Lending platform for microfinancefintechDouble-entry ledger, loan products as configuration, digital onboarding with eKYC.
Knowledge base from chats and meetingsaiTranscription with speaker separation; decisions and tasks extracted and linked to their sources.
Local-first voice assistantaiDictation and meeting notes on the laptop; speech never leaves the device.
Answers from corporate chataiStaff questions answered from chat history and guides, with a link to every source.
Local vision model for agentsaiScreenshot analysis and OCR on a local model for coding and support agents.
Parking access with quotasphysicalPlate recognition at the gate; cars over the owner's quota are not let in.
Media archive pipelinedataDuplicates removed, people and events recognised, years of photos sorted into a timeline.
04 — Approach

Start small, prove it on your data

Step 1

A short call

You describe the problem and how you notice it today. We say what can be measured and automated.

Step 2

A pilot

One system, a few weeks, your real data. You see the result before you commit.

Step 3

Run it your way

We operate it as a service, or install it on your servers and hand it to your team.

Strict security policy? Everything can run inside your network, with no outbound connections beyond the ones you approve.

05 — Contact

Tell us about your task

A few lines are enough: which system, what goes wrong, how you notice it today. We reply by email.

Discuss