STUDIO
Veres Labs
AI systems architecture & operational practice
Designs, builds, and ships AI systems for operational work. Independent practice.
London, United Kingdom.
founded MMXXVI
Products
| code | project | description | status |
|---|---|---|---|
| VL007 | AI Operations | Scoping an operations assistant for scheduled staff checks, stock routines, follow-up and owner reporting, with tools added in stages. | in scoping |
| VL006 | Thimble | AI audio reader with a reading interface, speech generation and saved listening position. | preparing v1 |
| VL005 | Patient Notes | Bespoke patient records, medical dictation and editable prescription output for a consultant orthopaedic surgeon. | deployedin use |
| VL004 | Taste OS | Personal companion system with deep memory architecture | core researchprivate alpha |
| VL003 | RP Engine | Cinematic narrative system with scene-aware orchestration | prototype |
| VL002 | Job Engine | AI-native recruitment operating system | in development |
| VL001 | Veres Labs | Studio site | shipped |
Writing
| title | summary | date |
|---|---|---|
| Session-Boundary Distillation★ | A memory architecture primitive — substrate paper Memory architecture at the seams — how AI systems preserve what matters across context resets. | 24.VI.2026 |
| Inhabited Architecture | How the novel output happens — architecture, not prompt | 3.VI.2026 |
| Mining for Substrate | A practitioner's methodology for working with large language models | coming soon |
CVs
| AI product & systemsProduct management, systems building and commercial transformation. | View CV |
| Applied AI architecture & researchMemory systems, methodology and technical writing. | View CV |
Research
active threads
| thread | description |
|---|---|
| Self-Aligning Entities | AI systems that organize their own substrate and improve as context accumulates — without re-prompting. |
| Inhabited Systems | Architectures for AI with continuity, voice, and the ability to sense their own state. |
| Asymmetric Compression★ | Differential memory rendering — routine moments compress, high-signal memories persist at full resolution. Same budget, sharper recall. |
Contact
Yasser Jawad · yas@veres.global · linkedin · github