Cognitive performance, turned into actionable insight

How BearTech and Cognitive Healthtech turned a founder-led, lived-experience framework into a working AI product that helps people measure and monitor cognitive performance.

In short

Cognitive Healthtech approached BearTech with two years of proprietary research and a clear, deeply personal vision: develop a novel and unique way to help people measure and monitor cognitive performance, supporting earlier insight for individuals, families and clinicians across a wide range of conditions.

Together, we have turned that founder-led framework into a working AI product, one that learns from people's everyday communication patterns and surfaces meaningful signals across five dimensions of cognitive health. The platform has been piloted with real users, is now extending to native mobile, and Cognitive Healthtech was awarded one of a handful of places on the UKRI STFC Health Mission Accelerator for dementia, with the 2026 cohort running from January to March as part of the government-backed NHS Fit for the Future programme.

This is the story of what we built, how we built it, and what is next.

It started with someone who had lived the problem

Cognitive Healthtech started where the best products often do, with someone who has lived the problem.

Its founder grew up watching family members navigate neurodevelopmental and neurodegenerative conditions including ADHD, Dementia, Autism and Bipolar Disorder, supported by a mother who worked as a clinical lead in mental health. The day-to-day reality was unmistakable: when the right understanding, tools and support were in place, life looked very different. Without them, even small things became hard.

The vision that emerged from that experience is simple to state and ambitious to deliver. Help people measure and monitor cognitive performance through their own data, so they can recognise their patterns, respond earlier, and make better decisions, alone or alongside a clinician.

"This work is rooted in lived experience, so finding the right partner really mattered."

Victoria Burnip, CEO, Cognitive Healthtech

Translating a framework into a product

After roughly two years of research and clinical reading, CHT had built a proprietary framework for thinking about cognitive performance through natural communication patterns. It worked. But it was manual. To make it useful at scale, the framework needed to be translated into automated technology, and into something that could fit into people's lives.

Three things needed to be true:

• Automated. The framework had to be consistent, repeatable, and not dependent on manual interpretation.
• Scalable. It needed to support the company's ambition across pilots, partnerships, clinical contexts and commercial growth, without rebuilding from scratch each time.
• Real-world. It had to work with the conversations people actually have, in the places they have them, not synthetic prompts in a clinical lab.

That last point matters more than it sounds. Many cognitive assessments require people to sit a test. CHT's bet, and one of the founding insights of the product, is that the richest signal is already happening in everyday speech, every time someone joins a meeting, takes a call, or reflects out loud.

Delivered in two phases, each solving a real problem

We delivered the platform in two main phases, each one feeding the next, and each one solving a real problem rather than chasing technical novelty. The focus throughout has been on building AI outcomes that make sense to a non-expert, while doing the underlying engineering carefully enough that the product is ready for clinical and commercial conversations.

Phase 1: foundations, first AI model, and pilot

A working end-to-end pipeline that turned the founder-led framework into measurable software:

• Capture from real life. Secure, consented ingestion of audio from Microsoft Teams, Google Meet, and direct upload. Users connect once; the system listens to what is already happening rather than asking them to perform a test.
• Understand what was said and how. Automated transcription and extraction of dozens of conversational signals: speaking rate, pauses, prosody, fillers, interruptions, and the way people phrase tasks.
• A daily cognitive score. An overall, person-centred cognitive performance score, produced daily, a clear and interpretable summary number that becomes more meaningful with use.
• Five dimensions, not one number. Beyond the overall score, the platform surfaces five interpretable subscales: communication flow, cognitive focus, affective expression, social reciprocity, and memory.
• Confidence built in. Every score is published with a confidence rating, so users know when a reading is robust and when it is not, a critical design choice for an audience that includes clinicians and people with a range of cognitive conditions.
• A dashboard that respects the user. An accessibility-compliant (WCAG 2.1 AA) web interface that visualises trends and lets users tag and reflect on what they are seeing. Feedback language is deliberately supportive and non-diagnostic.
• Privacy and ethics from day one. Pseudonymisation, encryption in transit and at rest, UK-hosted infrastructure, consent management, and full audit logging. The platform was designed against UK GDPR from the first sprint, not retro-fitted later.
• Piloted with real people. The MVP was deployed and tested with a small cohort of real users, producing the first real-world data on how the platform performs, and the body of footage that powers Phase 2.

Phase 2: custom-trained AI, mobile, and commercial readiness

With the MVP validated, Phase 2 takes the platform from a research prototype to a production-ready product:

• AI trained on the platform's own data. The cognitive indicators have been re-trained on roughly 100 hours of real Phase 1 footage, replacing generic pre-trained models with ones that learn the specific patterns the platform cares about.
• Score articulation. When a user sees a score, they can also see why: explanatory logic that links scores to the behavioural and conversational signals that drove them. A deliberate choice for trust and clinical usefulness.
• Behavioural video intelligence. Users, and in time clinicians, can replay the exact moments in a video that influenced a score. Click a dashboard insight and jump straight to the second it came from.
• Reply-video. Users can respond to their own analysed sessions, a natural feedback loop for self-reflection and, eventually, richer training data.
• Mobile-first. Native iOS and Android applications, designed for daily use, with deployment to the Apple App Store and Google Play Store.
• Commercial model. Stripe-powered subscriptions for paying users, plus a voucher system for single-use trial access, important for research partnerships, clinician pilots, and grant-funded cohorts.
• Configurable clinical pathway. A clinical pathway variable that can be tuned to legislative and regulatory guidance as it evolves, keeping the platform deployable across different clinical contexts without rewriting the product each time.

"The underlying engineering matters, but the experience matters more. The product is designed to feel like a tool, not a test."

How we worked together

CHT chose BearTech for three reasons stated up-front: strong technical expertise, a lived understanding of the problem space, and a pragmatic, flexible way of working. All three have mattered.

Practically, the engagement has looked like this:

• A working team, not a remote agency. BearTech has operated as an extension of the CHT team, joining the same conversations, holding senior CTO-level decisions with the CEO, and bringing product opinions alongside engineering ones.
• Patience when it counts. Cognitive health data is sensitive and slow to gather. The cadence of the project has been adjusted around data availability, pushing hard during build sprints, then pausing when the right thing to do was to wait for clean, consented data.
• Built with funding in mind. Throughout, the architecture and roadmap have been shaped by what investors and grant assessors need to see. Pitch deck development, grant-ready product planning, and introductions into relevant networks have all sat alongside the engineering work.
• Other small acts that add up. Help advertising a PhD role, sense-checking decisions out loud, talking through ideas before they hit a sprint board.

"BearTech have felt like a genuine extension of the Cognitive Healthtech team from the very beginning. They've helped us turn a founder-led framework into a scalable product, bringing technical depth, strategic thinking, and a collaborative approach throughout."

Victoria Burnip, CEO, Cognitive Healthtech

Validation, milestones, and what exceeded expectations

The most concrete external validation came earlier in 2026, when Cognitive Healthtech was awarded one of a handful of places on the UKRI STFC Health Mission Accelerator for dementia, a competitive, government-backed programme aligned to the NHS Fit for the Future strategy. The 2026 cohort ran from January to March, supporting multi-site testing and adoption of breakthrough health technologies, an important step toward real-world clinical deployment.

Alongside that:

• An MVP piloted with real users, producing real-world performance data.
• A fully working AI scoring engine across five dimensions of cognitive health, with confidence ratings.
• Custom-trained models built on the platform's own Phase 1 footage.
• Pitch and grant material that has supported the company's wider funding conversations.
• A mobile-ready platform progressing toward public release.

"I've seen other projects stall, miss the brief, or never deliver. Having a partner that delivers reliably, and talks openly when something needs discussing, has been amazing."

Victoria Burnip, CEO, Cognitive Healthtech

The roadmap from here

• Stronger datasets, stronger AI. Train against larger, richer datasets, including open data sources and additional pilot footage, to increase accuracy and confidence ranges.
• Structured user testing at scale. Iterate on the mobile product and dashboards based on real-world usage, across the full range of conditions the platform supports.
• A clinician dashboard. Develop and test a dashboard built specifically for clinical use, included in upcoming grant bids.
• AI-driven recommendations. Evolve the platform from describing patterns to suggesting practical, evidence-based actions that users can take based on their data, the long-term vision.

BearTech continues as a key partner in this evolution: collaborating on product development as requirements mature, supporting user testing and iteration, strengthening the AI and data pipeline as new datasets become available, and advising on scaling and readiness for bids, funding, and clinical deployment.

"Exceptionally talented tech developers with a human approach, who consistently exceed expectations."

Victoria Burnip, CEO, on BearTech in one sentence

About