SoftTrainer

SoftTrainer is an AI simulation platform for high-stakes workplace conversations: feedback, conflict, delegation, onboarding, and customer-facing dialogue. Managers practise with realistic AI personas, and our proprietary reasoning engine measures how their behaviour changes between an initial assessment and a reassessment on a new case, with deep individual analytics.

The hard part is the assessment, not the conversation. A general-purpose language model can produce fluent commentary on a dialogue, but it cannot tell an organisation whether a manager actually improved. SoftTrainer's engine reasons like a certified human assessor: expert assessment logic is decomposed into competencies, observable behavioural indicators and contextual evidence, so every conclusion is anchored to a specific moment in the transcript and can be read, questioned and defended.

That makes the result explainable. A learner sees which behaviour was observed, where it appeared in their own words, and what changed since their first attempt. An L&D lead sees the same evidence at cohort level, without reading hundreds of transcripts.

Where we are. SoftTrainer is at TRL 5, validated in a relevant environment. The platform has run 4 218 practice sessions with 393 users and 49 211 analysed utterances. Corporate pilots across IoT, enterprise SaaS and IT services.

Research behind the product. We published our own study of how European organisations evaluate soft skills today, drawing on 130+ L&D and HR leaders across 18 countries (DOI: 10.5281/zenodo.21791000). It confirmed what the pilots showed us: soft-skills training is bought on completion rates and satisfaction scores, because nothing on the market measures behaviour change itself. That gap is the market we are building for.

Who we are. DigitalHub LLC is a team of three (2.3FTE).

• Ruslan Kovalenko, CEO/CTO, is a software engineer with 8+ years of experience, launched SoftTrainer's first working version at the AI TON hackathon, then architected the AI simulation and scoring engine, led the team, conducted 130+ interview market research, and took the company from hackathon prototype to paying pilots.

• Kristina Rohovynska, Head of Growth, is a go-to-market strategist with 13+ years of experience who built our outbound channel: 53% outreach-to-call and 44% call-to-SQL. Designs the behavioural logic behind our personas and scenarios, what makes practice transfer to real conversations.

• Antonina Panchenko, Head of Learning Methodology & Assessment, has 15+ years in L&D, 100+ programmes and 50k+ learners behind her, and makes our assessments scientifically defensible.

Responsible by design. SoftTrainer is built for development, not for employment decisions. Individual results stay with the learner; organisations receive cohort-level reporting. All data from European users is processed exclusively on servers within the EU. Users are told they are interacting with AI, and the assessment logic is documented so its conclusions can be reviewed rather than taken on trust. We are preparing for the EU AI Act obligations that apply from December 2027 and designing the product so that compliance is a reason to buy it in Europe, not an obstacle.

What is next? We're currently active participants in a deep-tech grant programme and raising our pre-seed round for market expansion.