AI transformation · Product · Presales
Make teams faster with AI. Build AI into what you sell.
Bring an AI architect into the room when you sell.
Three offers. Pick where it hurts.
What Nur AI does
The offer is deliberately narrow: improve how your company uses AI, build it into the product, or bring credible architecture into the sale.
Your people already have AI licences. Turn them into delivery speed: shared setups, working conventions, training that sticks and governance that keeps adoption from becoming one enthusiast’s side project.
Not tools for your staff. AI in the product itself: features that understand what a customer means, agents that do the work behind a click and applications that make you findable where customers already talk to AI.
Bring Roland into client conversations as your AI capability expert. He translates the problem into what AI can genuinely do, what it would take to build and what it would move—including the honest answer when it is not worth automating.
Before you spend a call
Worth being blunt about, so you can rule us in or out in thirty seconds instead of three meetings.
Proof, not adjectives
No demo theatre. These are the kinds of organisational, product and delivery outcomes the practice is built on.
AI adoption across engineering
AI enablement inside a live engineering organisation: turning tools into shared delivery habits while keeping quality, security and operating responsibility visible.
Groupon published the account on its own careers site, under Roland’s byline as Engineering Director. See it, with the archive citation →
Company-wide LLM knowledge base
A knowledge capability designed for company-wide use: making institutional context easier to retrieve and apply without pretending source quality and governance solve themselves.
10× more campaigns
A proprietary GenAI capability for the marketing team—built around the real campaign process, not a generic content interface.
Measured against the campaign volume the team could run before the tool existed.
8× service capacity
A distressed AI project stabilised under a tight deadline: core functionality rebuilt, system visibility improved, infrastructure secured and capacity scaled beyond the original plan.
Measured against what the platform could carry before the rebuild.
Recommendations from the people accountable
Written on LinkedIn by the people who paid for the work. See the full captures, with dates →
“They inherited a hostage AI project, stabilized it, and rebuilt the core AI functionality to perform as we originally envisioned, all under tight deadlines while our business couldn’t afford delays… They safeguarded our investment when it was genuinely at risk, and then built on it.”
“What we embarked on was truly breakthrough and Roland eagerly tackled the unknown, as well as many obstacles thrown at us. This project wouldn’t have come to fruition without Roland… Roland’s mindset helped us accomplish the seemingly (at the time) impossible.”
“He developed an open-source project, Nur — a knowledge base that later became the foundation for TopAssist, a company-wide tool used at Toptal to interact with LLMs and enterprise documents… those who seek a great manager and hands-on AI expert need look no further.”
Writing
Two pieces that set out the thinking behind the work. Both are readable here in full.
How engagement starts
The service category tells you where Roland helps. This path tells you how much commitment makes sense first.
The three highest-ROI automations in your operation, plus likely EU AI Act exposure. A build/no-build verdict on each.
Written verdict per opportunityOne agentic workflow shipped in your infrastructure, human-gated at consequential decisions and reversible if it does not hold up.
Working software · DeployedPart-time leadership across hiring, architecture, governance and delivery, from someone who has run the organisation.
Part-time · ongoingTeaching your people to get real work out of Claude Code and Codex: shared environment, conventions that make an agent reliable, review around agent-written code.
One workshop, or a running cadenceEngagements run on time and materials against a scope agreed up front. You buy the work actually done, and you can stop at any stage boundary.
How I work
The architecture is designed to leave the client in control from day one.
Nothing you own routes through a platform Nur AI controls. What gets built runs where your systems already live.
Every agentic action that touches a real decision has a person in the loop before launch.
If a pilot does not earn its place, it comes back out cleanly. Walking away should not cost more than the pilot did.
Before a pilot starts we agree what it should move and write down where that number stands today. At the end you take the same measurement, the same way. A pilot nobody baselined can’t be judged, only defended.
EU AI Act readiness
The Opportunity Audit includes a practical exposure read: likely risk class, documentation gaps and controls to review with counsel. This is operational guidance, not legal advice.
EDIH Latvia compatibility
For Latvian and Baltic clients, work can be scoped so you can test whether EDIH Latvia or digitalisation-voucher support applies. Worth checking before you commit your own budget.
About
AI architect and engineering leader—15+ years shipping software, 10+ years running the teams that ship it.
Engineering leadership at Groupon, Core Operations Services & AI Enablement. Before that: Group CIO at Sun Finance and the engineering leader who grew Visma Latvia’s team from 30 to over 100 in under two years.
Nur AI is the practice built on top of that—solo-operator by design, senior by necessity. Roland’s own consulting practice runs on agentic systems he built himself.
You will be talking to the person who does the work
If it is not a fit, you will hear that too.