# Tigran Hakobyan > Tigran Hakobyan is an AI product leader, builder and consultant in Yerevan. He teaches teams to build AI agents and makes free things for the community. I turn complex AI into products people use, and teach teams to build their own AI agents. On the side, I experiment a lot and share what I learn with the community. Based in Yerevan, working globally · English, Armenian, Russian. ## At a glance - 6+ Years in AI products - 25+ Products & capabilities shipped - 4 Applied-AI programs designed and delivered ## Experience - Work Your Agent, Founder (Aug 2026 to Now). Founded an agentic AI practice for professionals and businesses, from hands-on training to workflows running in their own systems. Design and implement agents for recurring reporting, research, document processing and operations. Build in human checkpoints and teach teams to maintain what we make together. - Ultralytics, Lead Product Manager, AI Platform (Feb 2025 to Now). Lead product and a team of 10 on Ultralytics Platform, an end-to-end computer vision SaaS built around the YOLO models, reporting to the founder and CEO. Promoted to Lead in July 2026. Launched the Platform across multiple product verticals, leading product end to end. Drove enterprise readiness, contributing to ISO 27001 and SOC 2 Type I. Built an AI-native product system with Claude Code, Linear and GitHub. - Prequel, Senior Product Manager, AI (Apr 2023 to Feb 2025). Established the AI product function as the company's first AI Product Manager, leading research-to-product delivery across 20+ specialists. Launched 15+ AI products and service pipelines across four product lines generating $7M in annual revenue. Reduced generation costs by 40%, delivering $450K in annual savings through vendor strategy. - Picsart, Product Manager, AI (Aug 2020 to Apr 2023). Owned a portfolio of 10+ AI products across five business units, taking computer vision models into consumer products and business APIs. Increased Background Removal and Object Removal conversion by 55% and 70%. Scaled Video Background Removal to 1.5M monthly users and productized AI for 50+ business customers. - KPMG Armenia, Advisory Consultant (Sep 2016 to Mar 2018). Provided business advisory and management consulting for leading local and CIS companies. Delivered industry, market and strategy research for valuations and due diligence. Analyzed business processes and management performance to identify improvements. ## Work with me, through Work Your Agent Website: https://workyouragent.com/ - Training: Build your own AI agents. Build an agent on a real task. Learn hands-on with guidance. Keep the skills to build more. - Consulting: Find the right next step. Map the work and its pain points. Test where AI could help. Decide what is worth building. - Implementation: Put a workflow into operation. Connect approved sources. Automate with human review. Help your team run the system. - Ongoing advisory: Keep the work moving. Get support after delivery. Review and extend workflows. Work through what blocks you. ## Teaching - TUMO Labs, Instructor, Agentic AI (April 2026 →). Created Build Your Own AI System and AI Studio. Working professionals build agents with Claude Code and MCP integrations. - Russian-Armenian University, Adjunct Lecturer (2020–2024). Taught Product Management to 200+ Master's students through workshops and real product cases. ## Things I build - [Financial Literacy for Kids](https://dramaget.pages.dev/): Eight playful lessons that teach kids in Armenia how money works, with games, polls and quizzes. No accounts, no paywall. - [Avo the Coach](https://github.com/thakobyan08/avo-the-coach): A cheerful avocado coach in your Mac menu bar that nudges you to stretch while you work. ## My approach Most people do one or two. I do all three. I ship AI products, build with agents and teach people how, all at the same time. Each one keeps the others honest. - I ship. 25+ AI products and capabilities led, from first model to market, at Ultralytics, Prequel and Picsart. So you get advice from someone who has seen what breaks after launch. - I build. My own product work runs on agents: discovery notes, roadmap updates, reports, specs and tickets. So what you learn has already been tested on real work. - I teach. 200+ Master's students, and 4 applied-AI programs designed and delivered at TUMO Labs. So your team keeps the skill after I leave. And underneath it all: a PhD in Finance and consulting at KPMG. My first question is always whether it's worth building. ## Resources - [Answer-first working instructions](https://thakobyan.me/resources/answer-first-working-instructions/): Makes any AI assistant lead with the answer, stay short and stop padding replies. - [Customer feedback triage](https://thakobyan.me/resources/customer-feedback-triage/): Separate product gaps from support needs and turn feedback into recommendations supported by evidence. - [Agent workspace starter](https://thakobyan.me/resources/agent-workspace-starter/): A project instructions template that keeps status, facts and history apart so agents stop working blind. - [Meeting to tasks](https://thakobyan.me/resources/meeting-to-tasks/): Turn meeting notes into grounded task drafts with owners, deadlines, and a concise team recap. - [Learn from every correction](https://thakobyan.me/resources/learn-from-every-correction/): Makes your AI workspace save each correction as a rule, so you stop repeating yourself. - [Product agent instructions](https://thakobyan.me/resources/product-agent-instructions/): Give an AI assistant clear product context, evidence rules, and working boundaries. - [Slide copy standard for AI-made decks](https://thakobyan.me/resources/slide-copy-standard/): Stops AI slide generators from writing label headlines, slogans and filler slides. - [Weekly product update](https://thakobyan.me/resources/weekly-product-update/): Explain weekly delivery, user value, and blockers without overstating progress. - [Deck prompt in four gates](https://thakobyan.me/resources/deck-prompt-in-four-gates/): Turns a deck request into a stand-alone prompt for an AI slide generator, with approval before writing. ## Find me - [LinkedIn](https://www.linkedin.com/in/thakobyan/) - [GitHub](https://github.com/thakobyan08) - [Contact form](https://thakobyan.me/#contact)