London / Bucharest Integration-First AI Systems Est. 2025
Nativly · AI Systems
Selected Engagements 2024–25

We design, build, and deliver AI systems wired into the tools and workflows your team already trusts.

AI projects that
never ship cost
SMBs £50K+.

Book a 30-Minute Call → See the work
4–6
Weeks to a live system
01 The Problem

Why most AI projects never reach production.

You are not alone. Most SMBs burn budget on proofs-of-concept that never become working systems. Four reasons it keeps happening.

01

Enterprise pricing, enterprise timelines

Big firms want £200K and six months before you see anything real. By the time a slide deck lands, your team has moved on.

02

No clear first move

You know AI matters, but it is rarely obvious where to start, how to scope it, or how to tell whether it is working.

03

Tools that sit outside the work

Standalone dashboards and chat links do not change behaviour. If it does not live in your CRM, Sheets, or ops tools, it will not get used.

04

Plenty of noise, little delivery

The market is full of people who talk about AI but cannot ship it. Most advice does not survive contact with your real tools, data, and constraints.

02 Who It's For

Teams with no time for experiments.

We work with people who need working systems plugged into real revenue and operations, not another pilot.

i.

Founders & MDs who want AI that pays for itself

You own the P&L and have no patience for hype. You need practical systems that lighten the load and get used.

First system live in 4–6 weeks
ii.

Operations leads and overstretched CTOs

Your backlog is full and the team is small. You need a partner who can design, build, and own a scoped system end to end.

We plug into your stack, not replace it
iii.

Investment firms with complex portfolios

You need live visibility across brands, markets, and operators without forcing everyone onto one tool.

From spreadsheets to live portfolio intelligence
Connected by design
03 The Method

Three stages. Clear scope.

Concrete deliverables at every step. You always know what is coming next.

1

Map

We map your workflows, tools, and constraints, then identify two or three opportunities that will work inside your real systems. You leave with a complete system map and a build roadmap you can use with us or your own team.

1 week · remote
2

Pilot

Using the map, we design and ship the first working version, alongside your team or independently. A scoped, integration-ready build based on the architecture we defined together.

4–6 weeks · fixed scope
3

Scale

Once it is live, we focus on ongoing wins inside your existing infrastructure: strengthening what works, rolling out to more teams, and adding use-cases that fit your stack.

Ongoing · only if it earns its keep

The best systems disappear into the work. They become part of how the day runs, not another tool to manage.

Nativly · Razvan Burlacu & YJ Chen
04 Selected Work

Systems we have shipped.

Not prototypes. Practical systems built around real constraints and real teams.

Flagship · Retail

SNIP · Store Network Intelligence Platform

A live command centre for a multi-brand menswear retailer's 105+ store estate. SNIP consolidates store locations, lease contracts, sales and customer data from a fragmented archive of spreadsheets, scanned PDF leases and BigQuery views into one interactive map, and turns raw signals into recommended actions with deadlines.

105+
stores on one live, multi-layer national map
5 roles
role-based access tiers with financial masking
Bilingual
fully localized operator UI, local language & English
Overnight
ETL refresh plus a morning decision digest

What it does

  • A decision engine turns red, amber and green store flags into recommended actions (consider non-renewal, open negotiation now, re-paper lease), each with a deadline and an accept-or-dismiss audit trail.
  • A lease-runway view tracks every contract through T-180, T-90 and T-30 negotiation windows, so a renewal is never missed.
  • OCR ingestion reads scanned leases written in a foreign language, renders the key facts into English, clusters multi-document packages, and routes every file through review-before-approve before anything reaches the dashboards.
  • Fair pace-to-target scoring and a "Four Wall" new-store ramp view judge each store against where it should be today, not just the month total.
  • Five-tier role-based access with financial masking, a fully bilingual interface, and a daily digest of new flags and lease-window crossings.

Stack

Frontend · Next.js 14 · Tailwind · Mapbox GL
Backend · Node 20 / TypeScript · Express · Prisma
Data · Postgres · BigQuery · node-cron ETL
Intelligence · Document AI · Vertex Gemini · Foreign-language OCR
Cloud · Google Cloud Run · Firebase Auth (SSO)
Further engagements
01
Retail

UK Market GTM Readiness AI

An AI go-to-market engine that scores a retail brand's UK launch readiness and shows exactly what to fix first.

Pitch deck · Web scrape · UK GTM

Capabilities

  • Ingests pitch deck and website to understand the offer automatically.
  • Scores positioning, ICP fit, and category risk for the UK retail market.
  • Outputs a prioritised GTM action list ready for immediate execution.

Stack

Ingestion · Web scraping · Pitch-deck & PDF parsing
Intelligence · LLM scoring · Retrieval
Output · Python · Automated action report
02
Professional Services

Trust & Footprint Intelligence System

A trust platform that turns a consultant's full internet footprint into an evidence-based Trust Index.

RAG · Footprint · Profiles

Capabilities

  • Ingests websites, portfolios, and documents from across the public web.
  • Uses RAG and structured signals to score delivery risk and credibility.
  • Generates clean, comparable consultant profiles ready for matching.

Stack

Ingestion · Web crawl · Document parsing
Intelligence · RAG · Vector search · LLM scoring
Output · Structured profile & Trust Index
03
Investment Firm

Cross-Vertical GCP Data Platform

A cloud data platform giving a multi-market investment firm one live view across every portfolio company.

BigQuery · dbt · Metabase · Chatbot

Capabilities

  • Migrates fragmented on-prem data into BigQuery with dbt modelling.
  • Surfaces portfolio health through Metabase vision-board dashboards.
  • Natural-language chatbot lets the team query everything instantly.

Stack

Data · BigQuery · dbt
Surfaces · Metabase dashboards
Intelligence · Natural-language query chatbot
Cloud · Google Cloud Platform
04
Product · Workshops & Events

Echo Room · AI Decision Environment

A mobile-first platform where teams work through complex tradeoffs, vote on structured options with justifications, and leave with a decision map instead of slides.

PWA · AI Generation · Decision Maps

Capabilities

  • Turns an organiser's short brief into structured quests, decisions and options automatically.
  • Teams of three move through sequential decision points, voting with written justifications.
  • Generates a shareable decision map documenting every choice, tradeoff and commitment.

Stack

Frontend · Next.js 14 · Tailwind · PWA (offline)
Backend · TypeScript · Prisma · Zod
Data · Postgres · Redis rate limiting
Intelligence · OpenAI quest generation
Cloud · Vercel serverless
05
Product · Education

Placement Test & Analytics Portal

A browser-based placement-testing portal for a multi-centre children's language school. Staff launch the right age-banded test and each student is scored, levelled and reported automatically.

Auto-scoring · PDF reports · Analytics

Capabilities

  • Three age-banded test experiences, each with reading, grammar, listening and writing sections.
  • Scores answers in the browser and recommends a CEFR and Cambridge level with a section-by-section breakdown.
  • Generates a branded PDF report per student and streams question-level analytics to the back office, with a training mode that keeps practice data separate.

Stack

Frontend · Standalone HTML / JS · in-browser scoring
Reporting · Client-side PDF generation
Data · Google Apps Script · Sheets · optional BigQuery
Frameworks · CEFR & Cambridge alignment
Hosting · Netlify
05 The Team

Who you will work with.

Senior, hands-on operators. Not a rotating cast of juniors.

YJ Chen
Commercial & Strategy

YJ Chen

Commercial & Strategy Lead

Fifteen years across retail, immersive tech, and market expansion. A cross-cultural operator and advisor to multi-market ventures. MIT AI Executive Programme and London Business School MBA, with a master's in Industrial and Operations Engineering from the University of Michigan, Ann Arbor.

Razvan Burlacu
Systems & Delivery

Razvan Burlacu

Technical Lead

A decade leading engineering across immersive tech, AI-powered SaaS, and real-time systems. A technical consultant to startups and SMEs, known for delivering production-ready MVPs and workflow-integrated tools at speed. Specialised in architecture, infrastructure, and solutions that fit existing environments.

06 FAQ

Common questions.

What happens on the first call?

We map two or three concrete opportunities where AI can help, based on your current tools and workflows. No homework required: you bring context, we bring structure.


What if we are not ready for AI?

Most teams feel that way. As long as you have some digital workflows (Sheets, CRMs, ERPs, dashboards), there is usually a sensible starting point.

A scoped AI system wired into your tools, plus a short enablement session for your team. You keep the system and the documentation, whether or not we continue together.
No. We are stack-agnostic. Recent work spans GCP (BigQuery, dbt, Metabase), custom web apps, and stacks built around Google Sheets, Notion, and CRMs. The right stack is the one your team will use.
We work within your existing security model: VPNs, SSO, service accounts, and least-privilege access. We agree access levels up front and can work with your IT or security partner where needed.
We run strategy and architecture engagements, but our bias is towards systems that ship. If you are not ready for a build, we design a roadmap your team or existing partners can execute.
Get Started

Ready to explore a real use-case?

In 30 minutes we will shortlist two or three sensible AI opportunities for your business, and tell you honestly if now is not the right time.

Book the 30-Minute Call →