NU·WRIGHT
Makers
of deeptech & social impact
Nuwright is a remote-co building frontier AI for the teams who do the relational work of care: children's hospitals, mental-health services, schools, and the community organisations holding the people most easily missed. The applications we build are calibrated, evaluated, and shipped alongside the practitioners whose judgement they are meant to support.
Most of what we do is unglamorous and durable: sit with the people doing the work, learn their practice line by line, evaluate before we announce, ship in partnership. We very firmly believe that technology should enhance human judgement, not replace it.
Frontier AI for mental health
Live and in-pilot programmes building frontier AI for the clinical, educational, and community teams who already do the relational work of care. Calibrated, evaluated, and shipped with them, not at them.
- Children & adolescents
- Teenagers & young adults
- High-school students from low-income families
- First-generation university students
- Young people of colour
- LGBTQ+ young people
- Young people in crisis
- New mothers
- Families navigating substance use
- Educators & school staff
- Vulnerable women & girls
the work is made possible by a small number of institutions whose patient capital and standards we are grateful for.
- 1.feelxs
A non-verbal and verbal assessment for hiring the people we trust to care for others. Built for hospitals, clinics, schools, and youth services.
→ - 2.the lift
A multi-role wellbeing platform for children. Clinician, school, and parent dashboards anchored on a child-friendly weekly check-in.
→ - 3.relationalAI
An open-source platform for building conversational agents with the emotional and motivational sensitivity of trained mental-health practitioners.
→with Project Evident · The Gates Foundation - 4.rare impact fund toolkit
A suite of tools non-profits can use to deliver AI-enabled impact safely, from impact-assessment visualisers to evidence-freshness auditing and training adaptation for vulnerable populations.
→with Project Evident · Rare Impact Fund · Google.org - 5.matriculate
Helping Matriculate equip their near-peer advising fellows with emotionally sensitive frontier AI, in service of the high-school students they support.
→with Matriculate · Project Evident
- 6.empower youth
LLM-supported capability tools that help non-specialist practitioners deliver behavioural interventions to vulnerable children.
with Harvard Medical School
Alongside the commissioned work, we also work on personal projects, typically data visualisations of complex datasets, built to make a hard signal legible. Deep space data, SCADA output from wind turbines, or code repositories themselves: we try to make layered complex data easy to understand for any audience.
- 1.ab astris
A deep-space cataloguing project that uses Python to analyse TESS, Kepler and JWST data in search of new variable stars and exoplanet candidates.
→PythonLomb-ScargleNumPyAstroPy - 2.torquescope
Cold-start fault detection for wind turbines from SCADA data. No new sensors, no historical failure data, no ML team required.
→PythonFastAPIPINNThree.jsSupabase - 3.coilmaps
A data-architecture visualiser that reads public GitHub repositories so complex structures can be interrogated by sight.
→PythonThree.jsReactLLM
- 4.ferroviz
A pair of WebGPU visualisations of the planet under pressure: ocean-plastic accumulation traced through real NOAA currents, and the urban heat-island gap drawn against income inequality from NASA satellite imagery. Rust on the server, WebGPU in the browser.
RustWebGPU / WGSLThree.jsNext.jsSupabase