Executive Overview
The intersection of artificial intelligence and the legal sector has long been dominated by incremental efficiencies. For years, legal technology start-ups have focused primarily on document drafting, contract review, and automated form-filling—tools designed to make existing lawyers slightly faster at their jobs. However, a much deeper structural problem has largely remained untouched: the economic viability of lower-value civil disputes.
For millions of individuals and small businesses, the traditional mechanics of civil litigation are prohibitively expensive. When the cost of legal representation approaches or exceeds the value of the claim at stake—and with legal costs typically unrecoverable in small-claims courts—justice becomes an economic impossibility. Viable claims are routinely abandoned, and countless unpaid invoices are written off simply because the delivery model of legal services is fundamentally broken.
Enter Mikhail Yatsuha, CEO and Co-Founder of CaseCraft.AI. A UK solicitor and legal technology entrepreneur with over a decade of hands-on experience in the legal services industry, Yatsuha has experienced these systemic failures firsthand. Rising through the ranks from intern and trainee solicitor to Partner at Sterling Law—where he led the commercial department—Yatsuha repeatedly witnessed the friction points where civil claims fell apart due to prohibitive legal overhead.
In late 2023, Yatsuha co-founded CaseCraft.AI to completely rethink how civil claims are assessed, prepared, and progressed. Rather than building another drafting tool, the company has developed an AI-native platform designed for agentic civil litigation. By integrating case assessment, evidence processing, pre-action correspondence, document generation, procedural tracking, and human legal review into a single, unified workflow, CaseCraft.AI is attempting to bridge the justice gap.
Operating on a no-win-no-fee basis for individual claimants and offering subscription-based bulk-claim and case-management tools for businesses, the company has raised approximately £1.8 million across three funding rounds. As of August 2026, the platform has processed over 4,500 matters with a cumulative value of £14.9 million, achieving a 51% settlement or default-judgment success rate. With enterprise deployments underway and an Employment Law Minimum Viable Product (MVP) currently in testing, CaseCraft.AI is charting a new course for vertical AI in high-stakes, regulated industries.
Detailed Chronology: From Traditional Practice to Legal Tech Disruption
The Evolution of a Legal Entrepreneur
Mikhail Yatsuha’s journey into legal technology was forged in the trenches of traditional law firms. Over more than a decade in the legal sector, he experienced every tier of practice, starting out as an intern and trainee solicitor before eventually earning a partnership at Sterling Law. As the head of Sterling Law’s commercial department, Yatsuha managed a heavy caseload of commercial disputes, contract breaches, and debtor-creditor conflicts.
Yet, it was precisely this traditional vantage point that exposed the deep-seated flaws in how civil justice is delivered. Day after day, Yatsuha encountered lower-value disputes that were legally sound and factually compelling, yet entirely uneconomic to pursue through conventional legal channels. In the UK small-claims landscape, where legal fees are rarely recoverable from the losing party, paying a solicitor an hourly rate to recover a £2,000 debt or resolve a minor contractual dispute makes zero financial sense.
Witnessing clients abandon valid claims—and watching businesses routinely write off unpaid invoices because pursuing them cost more than the debts themselves—planted the seed for CaseCraft.AI. Yatsuha realized that the crisis was not a drafting problem; it was a delivery-model problem. Traditional legal services were built for high-value litigation where hourly billing made sense. Lower-value disputes required an entirely different paradigm—one where technology could shoulder the heavy administrative and procedural burdens to make representation economically viable.
The Founding and Rise of CaseCraft.AI
At the end of 2023, Yatsuha teamed up with co-founders to launch CaseCraft.AI, a UK-based legal technology company engineered specifically for civil litigation. The core thesis was simple yet radical: instead of adding AI features to legacy workflows, build an AI-native platform designed from the ground up to handle the entire lifecycle of a civil claim.
Securing approximately £1.8 million across three funding rounds, CaseCraft.AI moved swiftly from concept to commercial deployment. By August 2026, the platform had achieved significant operational milestones:
- 4,562 matters created by users on the platform.
- £14.9 million in cumulative claim value processed through the system.
- 51% success rate, with matters either successfully settled or won on default judgment.
- 373 active matters currently undergoing live legal work.
Building on this momentum, the company has begun piloting its first enterprise deployments, enabling businesses to manage high volumes of claims with minimal friction, alongside the active testing of an Employment Law MVP designed to expand the platform’s jurisdictional reach.
Supporting Context & Metrics: The Architecture of Agentic AI in Law
Moving Beyond the Chatbot: The Mechanics of Agentic Workflows
The first wave of generative AI was characterized by conversational assistants and text generators—tools capable of writing emails, summarizing documents, or drafting paragraphs. While these tools have utility, they fall dangerously short in complex, multi-step environments like civil litigation.
A chatbot answers a discrete question in isolation. In contrast, an agentic legal workflow must maintain the state of a real-world legal matter over extended periods. It must track what has happened, catalog what evidence exists, identify what information is missing, and dynamically determine the next procedural steps.
CaseCraft.AI addresses this complexity by deploying specialized modular components rather than relying on a single, monolithic AI model. Some components are optimized for collecting and validating user inputs; others are dedicated to grounding legal reasoning or parsing incoming documents. This orchestration layer ensures that the system executes a reliable sequence of decisions and actions, steering a claim toward its optimal resolution without human micromanagement of every intermediate step.
Grounding and Reliability: Separating Inference from Extraction
In a legal workflow, hallucinations or plausible-sounding falsehoods can lead to catastrophic failures. To prevent generative models from inventing legal theories or misinterpreting facts, CaseCraft.AI enforces strict foundational boundaries.
The system is structurally constrained to verifiable evidence and authoritative legal sources. Rather than asking a general-purpose model to rely on its internal memory, CaseCraft.AI integrates with government APIs and authoritative databases to maintain real-time awareness of legislative changes, judicial precedents, and procedural best practices.
Furthermore, the platform maintains a rigorous boundary between extraction and inference. When a user uploads unstructured evidence—such as a WhatsApp chat thread with a contractor, digital invoices, or email chains—the AI extracts raw facts: dates, quoted prices, and verbatim statements. Conclusions regarding the legal significance of those facts are treated as separate inferences. By preserving an unbroken digital chain of custody linking every structured claim back to its original source document, CaseCraft.AI ensures that its case files withstand rigorous legal scrutiny.
Official Statements & Insights from CEO Mikhail Yatsuha
To gain a deeper understanding of the philosophy driving CaseCraft.AI, we examine key insights shared by CEO Mikhail Yatsuha regarding the realities of operating AI in high-stakes legal environments:
-
On the Economic Reality of Small Claims:
"What convinced me was not that lawyers needed a better drafting tool. It was that, for many lower-value disputes, the economics fail before a lawyer can help. The cost of representation can approach or exceed the value at stake, and most legal costs are generally not recoverable in small claims… That is a delivery-model problem, not a drafting problem. AI made it possible to redesign more of the journey around the person bringing the claim rather than simply make the lawyer working on it a little faster."
-
On the Realities of Live Case Management vs. Prototypes:
"Two things stand out. First, generating a document is the easy part; running a matter is hard. A live case means evidence, payments, deadlines, responses and next steps have to stay consistent over months… Second, people behave differently with AI. They can be more candid, which helps surface important facts earlier, but the system also has to provide boundaries and know when a human should step in. We learned that directly when introducing a human call at a trust-sensitive stage improved conversion by at least 15%."
-
On Measurable Human Oversight and Risk Boundaries:
"Measurable human oversight means human involvement is tied to identifiable risk, and we can see where intervention happens and why. ‘Human in the loop’ should not just be a reassuring phrase; we should be able to point to the trigger, the review and the decision that followed… One measure we can already point to is review time: since launch, the time required for human review has decreased fourfold as the system has improved."
-
On Knowing When Not to Act:
"One of our safeguards is counterintuitive: we pay attention not only to cases the AI wants to progress, but also to cases it is inclined to reject. Human oversight is not just about stopping an AI from being too aggressive; it can also stop the system from being too conservative and discarding a legitimate argument. One of the clearest escalation signals today is a lack of evidence… Where the matter is not sufficiently clear or complete, the next step can be human review rather than automatic progression."
-
On the Sustainable Competitive Advantage of Vertical AI:
"Our advantage was never going to be owning a foundation model. Better general-purpose models are good news for us. The difficult part is everything around them: retrieval, the matter database, legal knowledge, workflow infrastructure, integrations, controls and the operational understanding of how a claim moves from beginning to end. That work does not disappear when a new model ships."
Future Outlook: The Horizon of Agentic Automation in Regulated Industries
As foundational AI models continue to advance at a breakneck pace, the broader legal and technology landscapes are forced to confront questions of autonomy, liability, and professional boundaries. According to Yatsuha, the boundary between software automation and regulated legal services will inevitably shift, but at a measured pace. In high-stakes domains, users and regulators cannot rely on the speculative promise that future models will be error-free. True reliability must be continuously demonstrated through authoritative sourcing, robust technical controls, and accountable professional review.
For CaseCraft.AI, the immediate future centers on scaling its core civil litigation infrastructure while expanding into adjacent practice areas. The platform’s enterprise capabilities—which allow businesses to execute batch claims of up to 150 matters with a single click alongside advanced case-management workflows—signal a major shift toward enterprise-grade legal operations.
Beyond general small claims and commercial debt recovery, the company is actively testing its Employment Law MVP and exploring applications in Personal Injury litigation in partnership with specialist legal firms. Looking at the macro picture, Yatsuha believes the foundational architecture built by CaseCraft.AI is readily applicable to other document-heavy, highly regulated industries such as insurance, financial services, and compliance.
Ultimately, the companies that succeed in vertical AI will not simply be those that build the most articulate conversational interfaces. As CaseCraft.AI demonstrates, the true test of agentic AI lies in its operational rigor: knowing precisely when to process data, when to automate workflows, and, most importantly, when to step aside and let human judgment take the reins.
For readers interested in learning more about automated civil litigation and enterprise case management, visit CaseCraft.AI.
