Featured content / 03
Yet Another AI Memory: only 2 of 222 conversations needed a detailed recap
What changed when project memory let an AI agent continue ERP work across conversations, and what a 170-day analysis showed.
Read the article
PRODUCTS · OPERATIONS · AI SYSTEMS
Denis HizhnyakSystems Architect &0→1 Systems Builder
I design systems by changing the architecture around a problem, rather than merely optimizing existing components.
01SELECTED WORK
Different domains. One systems approach.
From marketing experiments and business operating systems to physical interfaces and ERP, these projects share an architectural approach to real problems.
01LANAPRO
A renovation operating system connecting design, estimates, procurement, schedules, and quality control, so clients did not have to coordinate the work themselves.
Client NPS >9002ERP DevelopmentWorkflows tested in Google Sheets and Forms became the basis for an ERP with 11 role-specific portals and project memory.
249,876 linesat a historical snapshot03Mass Marketing ExperimentationA method for finding ad combinations at scale: 80–120 compatible mixes of audiences, channels, strategies, positioning, and conversion in one cycle.
80–120combinations per cycle04Thumb InterfaceA compact pointer worn on the thumb, with a tactile anchor for one finger and cursor control without a desk or separate controller.
Physical prototypeNovember 202402HOW I WORK
From the overall architecture to concrete action.
I start with the goal and the real context: how the work happens, where intent and practice diverge, and what keeps the system from moving forward. Then I connect roles, processes, data, and tools, and bring the solution to a working state.
Principle
Architecture has to stand up to real work. I connect system design with enough implementation to test the solution in practice.
03CURRENT PROGRAM
0→AGI
A new public architectural experiment: I explore how to organize long-running, more autonomous work by AI systems.
01Yet Another AI Memory
The project preserves context across AI conversations.
Published · field study02Work-Centric AI
The user sets work goals; the system coordinates AI sessions.
Next publication · implemented03Accumulating Experience
Work accumulates experience that helps the system learn.
Preparing for publication · working implementation04COLLABORATION
Two ways to work together.
I’m currently open to collaboration on products and business operating systems, as well as research and development of systems for long-running AI work.
Products, companies, and operating systems
I work with founders, owners, and venture studio teams on new products, business models, and systems that need to work in practice.
0→AGI: research, validate, build
I invite people to independently validate the architecture, propose research questions, and build software, infrastructure, ML, product, and UX components.