Yury Moskaltsov
AI Expert | Independent AI Consultant for UK Businesses | Fractional AI Product & Delivery Lead

Is AI worth building for your business?

As a founder, I know every investment has to earn its place. As an AI and analytics expert, I help you find where AI could make the biggest difference, test it on real work and build what proves its value.

Book a free 30-minute call
Yury Moskaltsov

Yury Moskaltsov

10+ years of expertise in AI, analytics and data science.

LinkedIn profile

Services

AI Strategy & Consultancy

  • AI readiness assessment: a ranked shortlist of use cases and a recommendation on what to pilot first
  • Test feasibility and prioritise use cases
  • Build an AI delivery roadmap
  • Train engineers to use AI coding tools
  • Provide fractional AI product leadership

AI Workflow Automation & Agent Development

  • Integrate agents into existing systems
  • Build AI-native apps from scratch
  • Automate data processing and analysis
  • Support Agents with Machine Learning pipelines
  • Evaluate and monitor agents in production

I can bring in experienced frontend, backend and AI engineers, data scientists and other experts if needed to support your project.

Case Studies

GromBuilt an AI-native business messenger that handles more than 10,000 messages a month

Problem
Field teams agree on jobs in calls and voice notes. When someone needs to check who owns a task or what was agreed, they have to dig through the conversation.

Solution
As co-founder, I built Grom for iOS and Android. It transcribes voice notes and recorded calls, summarises decisions and turns spoken instructions into assigned tasks with reminders.

Impact
Teams now share more than 10,000 messages a month on Grom. They can track a job in the conversation where it began.

Parse LabsBuilt a business analytics agent used by Series B companies to scan 50+ internal tools for insights

Problem
A customer complaint, a product decision and the fix that follows can live in three different tools. Account leaders need to see how those pieces connect.

Solution
As Head of Product, I led the team that built an analytics agent spanning 50+ internal tools, including Salesforce, HubSpot, Jira, GitHub and Stripe.

Impact
Paying Series B companies use Parse to answer questions like “What are my biggest clients complaining about? Is my product team prioritising those issues? Are the engineers shipping fixes in time?”

Rakuten PaymentPrepared an AI roadmap for the CTO of Rakuten Payment

Problem
Rakuten Payment had potential AI projects ranging from cloud cost optimisation to cross-product targeting. The CTO needed a way to compare opportunities with different business goals.

Solution
I identified more than 40 potential applications, developed and validated a framework for comparing them, and used it to prioritise the projects.

Impact
I presented the CTO in Tokyo with a prioritised roadmap showing which opportunities to pursue first.

AstraZenecaImproved a machine-learning algorithm to speed up early drug discovery

Problem
Researchers needed to decide which genes associated with small cell lung cancer merited follow-up studies.

Solution
I improved the machine-learning workflow for identifying candidate genes, using tree-based models and graph neural networks to analyse those associations.

Impact
Selected genes were verified and taken forward into treatment-development studies. This solution helped reduce the time and cost of discovering new drugs.

What People Say about My Work

“He [Yury] combines great technical skills with interpersonal and communication skills at a level not often seen.”
Miguel Pereira, MD PhD

Worked together on the AstraZeneca project.

Read the full LinkedIn recommendation
“Yury is relentless about understanding what customers need. He keeps asking how the work gets done and where it gets frustrating. At Grom, his test for every feature is simple: will it make someone’s working day easier? If the answer is unclear, he keeps listening before he builds.”
Denis Grafski

Co-founder, Grom

“One of Yury’s greatest strengths is knowing what to build first. At Parse, he could look at a dozen competing requests, identify which one would make the biggest difference to customers and explain why the others could wait.”
Sebastiaan Bruinsma

CEO, Parse Labs

Frequently asked questions

Should we automate this workflow with AI or use a simpler integration?

If every new order triggers the same invoice, a standard integration will do. If someone has to interpret a voice note, work out who owns the job and turn it into a task, AI can help. That’s the workflow I built into Grom. Explore the case studies.

Who needs an AI agent?

Look for a task someone repeats often, where the next step depends on what they find. A good example is an account manager checking a CRM, support tickets and payment history to decide which customer needs attention. If the same input always triggers the same action, a simple automation is usually enough.

Can you build an AI app from scratch?

Yes. I can take a product from the first user interviews through design, engineering, testing and launch. With Grom, I built an iOS and Android messenger that turns voice notes and calls into transcripts, tasks and reminders. For a new app, I’d start with one workflow your first users can try.

What happens on the introductory call?

Bring one example of a workflow that takes too long or a pilot that has stalled. During the free 30-minute call, we can talk through who does the work, what you’ve tried and what a useful result would look like. If there’s a fit, I’ll propose a first piece of work.

Can you work with our engineers and existing tools?

Yes. I can join your product and engineering team part-time, pair with engineers on a first implementation, or train them to maintain it. At Parse Labs, that meant building across tools teams already used, including GitHub, Jira and Salesforce.

What if our data is confidential?

We can scope the problem using a redacted example or synthetic data. Before working with live records, we’ll agree on access, where the data can be processed and any NDA your team requires.

How do we know the agent is reliable?

We agree on the metrics, benchmarks and acceptable performance range for your use case. For a task-creation agent, for example, we might measure how often it identifies the right assignee and deadline. Before release, we evaluate the agent on a held-out validation set and run regression tests against the previous version. In production, we monitor for drift in incoming requests and regularly score a sample of outputs. If a metric falls outside the agreed range, we investigate the failure and validate the fix.

How do you price a project?

Pricing depends on the scope, integrations and how much delivery support you need. A defined pilot can have its own project fee. For ongoing work, we can agree a day rate and a weekly commitment. You’ll have the scope and fee before work starts.