Skip to content
agorco

Software studio · Tallinn, Estonia

630,000+ meters · 5+ years in production · national retail POS

We build data-intensive IT products.

Energy, retail, logistics, and lending are where we have shipped. The same stack and domain practice apply in other companies.

Aleksandr Gortujev · Principal — product, architecture, delivery · Tallinn, Estonia Senior architects, developers, and product managers on the bench. Production systems since 2007.

Over-the-shoulder view of architects at a whiteboard sketching an MDM C4 diagram: HES to MDM to TSO and billing, with containers for ingestion, TSO-gw, and billing exchange

Work

Selected work

customers served
630,000+
meter-data intervals
15 min
in production
5+ years
on the market
Before 10:00
DSO · meter data

5+ years

National-scale DSO meter-data management

Elektrilevi · Eesti Energia

  • DSO
  • MDM
  • Estfeed
  • AVP
  • Siemens
  • Landis+Gyr
Meter data path from field hardware to the market before 10:00
  1. Field

    Meters

    15-minute intervals at national scale

  2. OT / MDM

    Hardware and head-end

    Siemens, Landis+Gyr, Estfeed, AVP

  3. Lakehouse

    Same facts

    Iceberg, Trino, operational reports

  4. Obligation

    On the market

    Before 10:00, operators included

Constraint

Fifteen-minute intervals for more than 630,000 customers, held for 5+ years. Vendors, the distribution operator, and Elering still had to share the same facts. The DSO must put that data on the market before 10:00.

Value

  • Settlement-ready series

    A bounded context for DSO meter data: operational reports and lakehouse tables the business can query, hardware and MDM still in the picture.

  • Meter-to-cash

    Complete, timely series cuts the cost of getting from meter read to settlement. Gaps and late facts are what inflate meter-to-cash.

  • Fewer contacts

    When operators close meter issues before the customer calls, you skip the contact. A call is expensive; a silent fix is not.

  • 10:00 obligation

    The DSO is obligated to provide meter data before 10:00. Delivery is met, not hoped for.

Return

Operators and the business see the same fifteen-minute facts — settlement and self-service consumption, fewer costly customer contacts, and data on the market before 10:00.

Medallion lakehouse with Apache Iceberg and Trino; Estfeed and AVP contracts; Siemens MDM and Landis+Gyr hardware; OT on the Purdue Model; Azure hub-and-spoke.

Retail

8 years

Self-checkout kiosks and POS hardware on a grocery store floor

POS, self-checkout, and fiscal hardware

Nixor EE AS

8 years on the floor

Constraint. Store-floor systems for large Estonian retail chains, including the national EEK to EUR change on software and databases without stopping the floor.

Return. Faster queues lift the end-customer experience and let shops take more orders at peak — revenue the line would otherwise drop.

Logistics

National carrier

Logistics operations room with an Estonia routing map and a sorting hall beyond

Addressing and GIS routing

Omniva

National postal network

Constraint. Addressing and GIS routing for Estonia’s national postal carrier — the network already in production, not a greenfield map.

Return. Routes and addresses the floor could use, on the carrier’s live network.

Live sport

In-event

Live-event operations floor with stadium video and real-time charts

Recommendations on the live event

Gamesys / Snowcat Games

Real-time with the event

Constraint. Recommendations on Google Cloud Dataflow for live sport. The pipeline had to keep up with the market as the event moved.

Return. The house did not lose money when the market moved because the pipeline kept up with the event in real time.

Lending

Multi-country

Street ATM at night with a glowing interface

First ATM for a new network

IuteCredit Europe AS

Rolled out across countries

Constraint. First ATM for a new ATM network, with full product ownership — hardware, software, and the customer at the machine.

Return. The same product then went to multiple countries: experience at the machine for the end customer, and money for the company.

Approach

How we engage

We model the domain, ship, and stay until the facts are in production. Delivery can be a product you operate, or a partnership on the product itself — that depends on the work.

  1. 1

    Name the problem

    What the business cannot see or cannot run today, and the context around it — a live system, a new product, or both.

  2. 2

    Agree what done looks like

    A written decision attached to a delivery constraint: scope, ownership, and the production facts you will have.

  3. 3

    Ship and run

    Architecture and delivery until operators can act. Metrics and monitoring are part of the product, not a dashboard after go-live.

Stack

Stack

What we ship with. The same pieces apply when the domain is new.

Contact

If the product has to show the business something

Write the problem and the context. That is enough to decide whether we should take it on together. Replies from Aleksandr at the address below.