The Reference Point

Close Your Eyes made the case that the bottleneck was never code but seeing. The visualization loop (think, talk, see, understand, refine, ship) was the development methodology used to build every product in this portfolio. Proves understanding of human-AI interaction at the cognitive level.

The Positive Loop identified the first productive dopamine cycle in internet history. AI tools collapsed the gap between idea and shipped product from months to hours. This is the foundational observation behind everything else: the reason all of this work was possible.

The Human Sensor Layer extended the thesis into labor economics. Three-tier economy, node labor, belief-based investing, resistance assets. These demonstrate the ability to think at the macro level about how AI reshapes economic structures.

Preference for Legible Risk (2024) was the earliest piece. Game theory applied to why people prefer transparent negative-sum games over opaque positive-sum systems. It establishes the analytical lens that runs through all subsequent work.

Hive is the flagship. A daemon, dashboard, and coordination layer for running multiple Claude Code agents simultaneously. Auto-discovery, real-time status, auto-pilot, advisory locks, compound learning. Built using four agents iterating on each other's output while a human directed the architecture. This is the proof of concept for the entire thesis.

Crawler takes any GitHub repository and produces flowcharts, layer breakdowns, and context blocks explaining how the project works. Reverse Figma for code. Demonstrates system thinking applied to developer tooling.

Nudge is life maintenance on autopilot. Smart reminders for health, car, home, finance. Demonstrates product direction: identifying a problem space and building the minimum viable solution.

Booksby is personalized book recommendations that learn reading taste over time. Demonstrates iterative product development and recommendation system design.

The Sleepless Rishi is a fully automated YouTube channel. AI-generated Hindu scripture narration, visuals, and content end to end. Nine deities, 59 planned videos, seven-step pipeline from script to upload. Demonstrates pipeline engineering, content automation, and the ability to build and operate complex multi-step AI workflows at scale.

Stotram is the web companion: sacred hymns of India, readable and downloadable. Built alongside the YouTube pipeline, sharing the same data layer.

rohitmangtani.com is the public portfolio. Clean, light theme, curated. Houses the research articles, projects, and resume. Built with Next.js, Tailwind v4, deployed on Vercel.

The site itself is evidence. It was not outsourced to a designer or a development team. It was built by one person directing multiple AI agents, iterating through the visualization loop, shipping a production-quality web application. The medium is the message.

HBR definition: Understanding how AI agents work, what they can and cannot do, how to evaluate their output.

Evidence: Hive is a production system for managing AI agent fleets. Every product in the portfolio was built by directing AI agents, not by writing code manually. The articles demonstrate deep understanding of agent capabilities, limitations (context degradation, hallucination, coordination gaps), and operational patterns.

Resume language: "Built and operate production infrastructure for multi-agent AI systems. Daily operational experience directing four simultaneous Claude Code instances across independent projects."

HBR definition: Deep domain knowledge that allows you to evaluate whether AI output is correct in your field.

Evidence: MBA in Business Analytics provides the finance and operations foundation. RBC fixed income operations provides daily experience with reconciliation, risk controls, and settlement systems. The research articles demonstrate ability to evaluate AI output against economic theory, labor market data, and technology trends. A rigorous career audit demonstrates the ability to apply analytical frameworks to your own credentials and products with honesty.

Resume language: "Combine MBA-level business analytics with daily financial operations experience and independent research in AI labor economics."

HBR definition: Seeing how parts of a system interact and identifying leverage points.

Evidence: Hive's architecture (daemon, dashboard, hooks, coordination API, learning system) was designed as a complete system before any piece was built. The article series itself is a system: each piece builds on previous ones, cross-references create a lattice, the writing and building compound into a single proof of concept.

Resume language: "Designed and built multi-component system architecture including real-time daemon, WebSocket dashboard, coordination API, and compound learning layer for AI agent management."

HBR definition: Ability to adapt as AI capabilities evolve rapidly.

Evidence: The entire body of work was produced within twelve months using tools that did not exist eighteen months ago. Each quarter, the AI models improve. The workflow adapts. The meta-skill is regime detection, knowing when conditions shifted. The portfolio itself is a demonstration of building in a rapidly shifting environment and adapting in real time.

Resume language: "Produced twelve published research articles and five shipped products within twelve months by continuously adapting to evolving AI capabilities."

HBR definition: Ability to communicate intent to AI systems effectively.

Evidence: Every product was built through iterative prompting across four simultaneous agent instances. The visualization loop described in Close Your Eyes is a prompt craftsmanship methodology: describe what you want, see the result, identify the gap, describe the correction. Iterative direction is the core operational skill.

Resume language: "Developed and refined multi-step AI prompting pipelines for content generation, code development, and product iteration across five shipped products."

HBR definition: Designing workflows that combine human judgment with AI execution.

Evidence: Hive is literally this. Auto-pilot handles routine decisions. The dashboard surfaces only what requires human judgment. The coordination layer prevents AI agent collisions while the human bridges context between independent agents. The two roles that make hybrid workflows function, the technical bridge and the extraction layer, came straight from running these systems. This is not a theoretical capability. It is daily practice.

Resume language: "Designed and operate hybrid human-AI workflows where autonomous agents handle execution while human oversight focuses on context-holding, taste evaluation, and cross-system coordination."

Lead with a one-line positioning statement below the name. Not an "objective" or "summary." A declarative sentence that tells the reader what this person does.

"I build the coordination layer between AI systems and business outcomes." This works for enterprise roles. For frontier AI PM roles: "I build infrastructure for directing AI agent fleets." For consulting: "I help companies turn deployed AI into measurable business value."

RBC is the strongest traditional credential. But "Fixed Income Operations Analyst" undersells the relevance. The actual work includes: identifying automation potential in reconciliation workflows, categorizing processes as rules-based vs. judgment-dependent, and evaluating where systematic reduction of manual overhead is possible. This is exactly the skill set required for AI deployment in financial operations.

Reframe: "Evaluate operational workflows for AI automation potential. Categorize processes as rules-based (automatable) vs. judgment-dependent (human-required). Scope systematic reduction of manual overhead across reconciliation, settlement, and reporting cycles."

Fidelity Technology Risk adds: data integrity across multiple platforms, dashboard creation, audit process improvement. Reframe through the AI lens: "Built cross-platform data integrity monitoring and visual dashboards for business unit audit visibility."

The current resume has an empty projects section. This is the largest gap. Add a section titled "AI Products & Research" that includes:

• Hive: Production daemon and dashboard for managing multi-agent AI fleets. Real-time status detection, auto-pilot for routine approvals, coordination API for cross-agent communication, compound learning system. Used daily to direct four simultaneous Claude Code instances.
• Automated Content Pipeline: Seven-step AI workflow from script generation to video upload. 59 planned videos across nine content categories. Demonstrates pipeline engineering and content automation at scale.
• Developer Tooling: Tools that reverse-engineer repositories into flowcharts, layer breakdowns, and architectural context blocks.
• Published Research: Ten-article series on AI labor economics, agent management, and human-AI coordination. Topics include agent fleet management, residual human inputs in automated systems, the visualization development loop, and hybrid workflow design.

Each bullet should include a result or metric where possible. "Used daily" is a metric. "59 planned videos" is a metric. "Ten-article series" is a metric. "Four simultaneous agents" is a metric.

The current skills section lists: Financial Modeling, Analytical Skills, Agile Methodologies, Strategic Thinking, Communication, Collaboration. These are generic. Every MBA resume lists these words.

Replace with skills that reflect actual demonstrated capability:

• AI Operations: Multi-agent coordination, prompt engineering, AI output evaluation, pipeline design, compound learning systems
• Product: System architecture design, hybrid workflow design, product direction, user experience iteration
• Technical: Next.js, TypeScript, Python, Node.js, WebSockets, Express, Tailwind, Vercel, FFmpeg, edge-tts
• Business: Financial modeling, reconciliation and controls, data analytics, PowerBI, Tableau, SQL
• Research: Published AI labor economics research, cross-domain analysis, framework development

Target companies: Anthropic, OpenAI, Google DeepMind, Meta.

Lead with: Hive as proof of product thinking in the agent space. The Future of PM as evidence of understanding the landscape. The visualization loop from Close Your Eyes as a development methodology. Shipped products as velocity evidence.

De-emphasize: Finance operations details. Traditional business skills.

Key language: "Designed and built agent management infrastructure." "Identified the coordination gap between agent capability and fleet operations." "Published research on the operational layer missing from current AI tooling."

Portfolio link order: Hive, The Future of PM, The Five Inputs, Close Your Eyes.

Target companies: Citi, JPMorgan, Goldman Sachs, Visa, Caterpillar, Salesforce.

Lead with: The coordination layer between AI deployment and business outcomes. RBC experience reframed as automation potential evaluation. MBA as credibility signal. Research articles as thought leadership on enterprise AI adoption challenges.

Key data point: "95% of enterprise AI pilots show zero P&L impact" (MIT). Position yourself as building the coordination layer that converts deployment to value.

Key language: "Evaluate operational workflows for AI automation potential." "Built coordination infrastructure for multi-agent AI systems." "Research-backed understanding of why enterprise AI pilots fail and how to fix them."

Portfolio link order: The Human Bridge, The Future of PM, Hive, The Five Inputs.

Market context: AI consulting market at $11B in 2025, growing 26% annually. GenAI specialists command $350-$700/hr.

Lead with: The full portfolio as the pitch. Articles prove thinking. Products prove execution. The daily operation of Hive proves that the workflow works. The career audit proves you apply the same analytical rigor to yourself.

Entry point: One company in your network with deployed AI that is not generating value. Two-week engagement at $150/hr ($12,000 plus a case study). The case study becomes the next article. The article becomes the next pitch. The compound keeps running.

Key language: "I help companies close the gap between AI deployment and business outcomes." "Portfolio of shipped AI products, published research, and production agent management infrastructure." "Direct operational experience running multi-agent AI systems across independent projects."

Portfolio link order: The Human Bridge, Hive, The Future of PM, rohitmangtani.com/writing (full collection).