Education
MBA, MSc Economics, BSc Economics.
Full-Time MBA from Alliance Manchester Business School, MSc Economics from the University of Tehran and BSc Economics from Shahid Beheshti University.
StrategyEconometricsGame theoryMarket entry
Founder & CEO · AI-agent-run company · Agentic workflow specialist
Bez
Noferesti
Agentic AI Workflows
I build and operate AI workforces: specialised agents, context systems, workflow graphs, memory, tool access, evaluation loops and human escalation paths. As Founder and CEO of Torsik, I run an AI-agent-led organisation with myself as the single human operator.
My edge is the blend of agentic AI practice with commercial leadership, economics, management accounting and analytics. I turn messy business work into repeatable systems that agents can research, reason through, execute and hand back to humans at the right moment.
Agentic workflows
Designing agent roles, tool access, task queues, memory, verification and escalation paths for real operational work.
AI workforce orchestration
Managing specialised agents as a coordinated workforce with objectives, handoffs, audit trails and human oversight.
Graph engineering
Structuring knowledge, entities, relationships and decision history so agents can retrieve and reason with continuity.
Context engineering
Building the instructions, source libraries, retrieval ladders and operating context that make AI behaviour dependable.
Torsik
Founder and CEO of an AI-agent-run company.
Torsik is built as a practical demonstration of the future organisation: a company operated by an AI workforce, with one human accountable for direction, judgement, governance and client outcomes.
The operating model combines research agents, writing agents, memory systems, monitoring jobs, structured knowledge graphs, task backlogs, approval gates and executive decision support. The result is a lean organisation that can explore markets, build assets and manage follow-through through agentic workflows.
1
Human operator
AI
Agent workforce
Operating model
AI agents as the team
Specialised agents handle research, drafting, analysis, monitoring, reminders, triage and knowledge retrieval through a managed workflow with clear boundaries.
Human role
Direction, judgement and accountability
Bez sets strategy, designs the system, validates outputs, manages risk and keeps the organisation aligned to useful commercial outcomes.
Use cases
Market intelligence and execution systems
Torsik applies agentic workflows to market exploration, competitor analysis, opportunity mapping, content assets, operating procedures and executive decision support.
Professional foundation
Commercial leadership, economics and analytics behind the agentic AI work.
The AI work is grounded in board-level pricing leadership, quantitative analysis, market strategy, process design and finance discipline.
Finance
Part-qualified CIMA management accountant.
CIMA Dip MA and CGMA progression, with financial modelling, DCF, NPV, scenario analysis, sensitivity analysis, forecasting and management-accounting discipline.
CIMA Dip MAModellingForecastingDecision support
Commercial background
Pricing and business intelligence leadership.
Head of Pricing experience across a multi-brand portfolio, senior pricing leadership at BSS, business intelligence at Johnson Controls, and earlier consulting and startup work in market entry and technology.
Pricing strategyBIProcess designStakeholders
Agent design
Clear roles, tools and boundaries.
Each agent needs a job, source context, accessible tools, verification criteria, memory rules and a path to escalate judgement.
Context architecture
Knowledge that agents can use.
Strong retrieval, clean source hierarchies, graph structures and reusable instructions give agent workflows continuity.
Human oversight
Automation with accountability.
The human operator sets goals, approves meaningful decisions, monitors quality and owns the outcome.
Dogs · Household board members
Billy & Ash, the real leadership team.
Every serious AI workforce needs a strong governance layer. Mine happens to be fluffy, opinionated, permanently unpaid and highly motivated by snacks.
MoraleAnomaly detectionSnack governanceAsync compute
Chief context officer
Billy
Billy is a white Bichon Frise with the presence of a tiny aristocrat and the hairstyle of someone who absolutely has a stylist. Born in 2017, he brings seniority, softness and a firm belief that every room is improved when he is in charge of it.
Specialisms: supervising humans, maintaining household uptime and ensuring every workflow includes a morale checkpoint.
Junior agent evaluator
Ash
Ash is a Kerry Blue Terrier, born in 2025 and already operating like an autonomous exploratory agent: curious, fast, overcommitted and convinced every object deserves a detailed investigation. He brings junior-agent energy with senior-level confidence.
Specialisms: stress-testing household systems, edge-case discovery and keeping Billy’s leadership layer adaptive.
Connect
For agentic AI workflow, AI workforce and context-system roles.
I am interested in roles, advisory work and introductions around agentic AI operations, AI workforce design, context engineering, graph-backed knowledge systems and practical AI transformation.