AI agents and automation
Autonomous flows that close manual steps end to end, instead of prompting a human to do them.
Open for new projects
We take the routine off your team and hand it to machines
AI agents, integrations and data pipelines. We build the system, ship it to production and keep it running, instead of leaving you with a prototype.
What we take off your team
Six things we do end to end, from the first call to a system your team uses every day.
Integrations and data pipelines
CRM, webhooks, messengers, APIs and databases wired into one working loop.
Parsing and reverse engineering
We pull data from places with no public API. These are the tasks most teams turn down.
Dashboards and analytics
Metrics from scattered services on one screen, with filters shaped around how your business thinks.
MVP and product development
From idea and architecture to live users. We have already taken our own product that whole route.
Payments and billing
Subscriptions, plans and checkout, crypto included. Idempotent webhooks, not the flaky kind.
What we have already built and shipped
Three projects we can talk about openly. Work under NDA is discussed on a call.
Swiftin
AI translator for the browser: pages, subtitles, documents and what you type, in 100+ languages.
Task
The language barrier slows down reading and communication across the web, and ordinary translators lose tone and context.
Solution
One extension that translates web pages, selected text, whatever you type in any field, video subtitles and whole documents, with AI voice on top.
Result
1000+ users and growing, live in the Chrome Web Store, with a free plan and paid subscriptions.
Dashboards and analytics
Custom dashboards with data collection and analytics for business.
Task
Business data sat scattered across services and spreadsheets, with no single place to see what was happening.
Solution
Dashboards that pull from APIs, databases and parsers into one place, with metrics, charts and filters built for that business.
Result
The team sees key metrics in real time and decides on data instead of guesswork.
20+ automations on BAS
Tender parser, marketplace auto-publishing and data collection on Browser Automation Studio.
Task
The business lost hours every day: hunting tenders by hand, filling product cards on marketplaces, collecting data manually.
Solution
A tender parser that monitors platforms and filters by criteria, auto-publishing that fills product cards, plus 20+ other automations.
Result
Tenders are found and sorted automatically, cards go live without manual entry, hours of routine disappeared.
How the work actually goes, step by step
Five steps, in this order. The first one can end the project, and that is the point of it.
01
Scoping
We measure what the process costs you today, in hours and in money. If automation will not pay that back, you hear it on the first call.
02
Architecture and estimate
We show what the system is made of, what it costs and how long it takes, before any code is written.
03
Build
We work in iterations, so you watch progress instead of waiting in silence until the deadline.
04
Launch
We deploy on your data and your accounts, then walk your team through using it.
05
Support
We stay reachable. A system that breaks a month later is not a result.
Why this is safe to hand us
Five claims, all of them checkable.
Our own product in production
Swiftin: 1000+ users, paid subscriptions, live in the Chrome Web Store. We have walked the whole route, not just the client-facing pieces of it.
One contractor for the whole chain
Architecture, code, infrastructure, payments, launch. Nothing gets lost in a handover between teams.
We take the no-API work
Reverse engineering and parsing outside the happy path, and unfamiliar stacks without flinching.
We finish what we start
No quitting halfway. We ship it and then we keep it alive.
Straight about risks
If a task will not fit the deadline or will not pay for itself, you hear that before the work starts, not after.
Who is behind this
Veloris Lab is a small engineering bureau. Small is the feature: nothing is thrown over a wall between departments.
Behind it is Denys Kandyba, a forward deployed engineer: the same person designs the system, writes it, and picks up the phone when something breaks. First code in 2014, automation from 2023, working with AI full time since 2024.
About the founderWorking languagesRussian, Ukrainian, English
BaseBatumi, Georgia, UTC+4
Projects delivered15+
Full time on AI since2024
Contact
We will cost your task free and tell you straight whether it pays off
Write on Telegram or by email. We usually reply the same working day, timezone UTC+4.