Prefer the simpler system
If a workflow change, a search index or a rule solves the problem better than a model, we use that instead.
AI for learning, skills
and work.
Dovichi Labs is an AI-native product company building intelligent solutions, agentic systems and domain-specific AI for the future of learning, skills and work. Three products cover that path today, and we own the technology layer beneath them rather than renting it.
3 products liveUK · NGFounded 2024
Pipeline / Active
FocusLearning · Skills · Work
Learning SystemsSkills & AssessmentCareer IntelligenceStory-Led MediaCreator StudioOwned Infrastructure
01 / The problem we work on
The hard part is no longer learning. It is proving it.
Knowledge has never been more available, yet the path from learning a skill to being hired for it is still slow, unclear and badly evidenced. Learners cannot show what they can do. Employers cannot verify it at scale.
Every product we build works on one stage of that path. We started with the learners and employers we know first-hand, and we build for that route into the global job market.
02 / ProductsThree products.
Three products.
One pipeline.
Learn the skill, retain it, prove it to an employer. Each product owns a stage, and the shared layer underneath means they compound instead of competing for attention.
LearnLive
DoviLearn
Learning and career technology
Learning and career products that take people from a standing start to a job-relevant skill.
DetailsRetainLive
LearningFilms
Story-led learning media
Cinematic, story-led films that make practical and professional knowledge easier to understand, remember and apply.
DetailsProveLive
CVSense
Recruitment intelligence
AI recruitment intelligence helping employers source, screen and identify better-matched professionals at scale.
DetailsScaleIn development
LearningFilms Creator Studio
Creator supply for learning media
A studio opening LearningFilms production to creators, so learning films can be made at a volume no single in-house team could reach.
Details03 / How we buildWe own the layer
We own the layer
beneath our products.
Identity, usage accounting, private history, validated model output and secure cloud access are built once and shared. That is why a small team ships AI features quickly without loosening security or losing cost control — and it is a capability, not a separate business.
Explore the technologyProduct ExperiencesLearning · Career · Roadmaps
Domain LayerSkills · Evidence · Roles
Reasoning SystemsPrompts · State · Validation
Context & Product DataStructured inputs · Grounding
Secure OperationsIdentity · Usage · Private history
Foundation ModelsClaude · Open models · AWS · GCP
Supported by
04 / How we decide
Evidence, not excitement.
Dovichi Labs is the public face of the method. The same rules that govern our roadmap tool govern our own product decisions.
No fabricated returns
We set measurable pilot targets rather than promising outcomes we cannot evidence.
A human owns the decision
Consequential outcomes for a learner, candidate or employer keep a person accountable in the loop.
Exit criteria before scale
Every stage states what evidence would justify continuing, changing or stopping.
Try the roadmap tool the method is built into. It is free and it will tell you where AI does not help.
Enter Dovichi Labs05 / Client engagements
We take on a small number of AI engineering engagements.
Intelligent solutions, agentic systems, domain-specific AI and product infrastructure, for clients who need systems we already build daily. It funds the product work and keeps us close to production pressure. It is deliberately capped and it is not our product line.
See selected engagements