AI-First Product Marketing · GenAI Solutions

I build the systems that turn AI capability into brand narrative.

8+ years translating complex AI/ML and GenAI solutions into business-ready narratives, sales assets, and go-to-market campaigns. I ship agentic pipelines and multi-agent LLM workflows, then package the outcome into strategy that converts.

View the work Get in touch
70%asset-time reduction
$450Kpipeline influenced
43%organic traffic growth
core.competenciesRESUME
01
Technical-to-Business TranslationTurning AI/ML capability into value narratives sales can use
02
AI/ML & GenAI Product MarketingGo-to-market for complex, technical AI products
03
Agentic Automationn8n · Zapier · Claude Code · MCP pipelines
04
Multi-Agent LLM OrchestrationBot-stacking & context-passing agent frameworks
05
Content & Thought LeadershipEditorial systems and brand-voice governance at scale
06
Sales EnablementBattle cards, decks, and nurture sequences that convert
07
Brand StrategyPositioning, narrative architecture, and activation
Systems shipped

Tri ValueMappers — the AI solution translator

Technical teams build brilliant AI/ML solutions; sales teams struggle to say why they matter. ValueMappers ingests technical specs plus customer intent signals and generates five customer-centric sales assets in minutes — elevator pitch, battle card, LinkedIn authority post, nurture sequence, competitive positioning — each mapped across three business-value dimensions (operational efficiency, customer experience, revenue growth) and passed through brand-alignment and clarity quality gates.

Claude APITri-Value frameworkQuality gatesMulti-asset generation
70%reduction in asset creation time
5asset types generated per run
2–3 minper feature (beta-tested)
12+industries supported
SYS-01 / AUTOMATION

YouTube MCP Pipeline

End-to-end MCP orchestration linking Claude, Google Drive, and YouTube. Cut per-scene production from 45 to 18 minutes with zero manual handoffs and OAuth-secured API access.

SYS-02 / SCALE

HSB AssetFactory

A three-tier AI-augmented content production system: automated research → Claude generation → editorial review. 3–4x weekly output at 95%+ brand-voice consistency.

SYS-03 / STRATEGY

Goa F&B Bot-Stacking

An 11-bot orchestration framework — research, strategy, and execution tiers chained with context-passing — producing a full category strategy in 2 weeks instead of 8–12.

SYS-04 / CONTENT OPS

AI ThoughtLeader

A dual-founder voice engine that turns one long-form blog into a week of on-brand LinkedIn content, with a human-in-the-loop quality gate. 8 hours of work down to under 20 minutes.

SYS-05 / HEALTHCARE

HealPonda — ID Hospital

A role-based staff dashboard plus a patient-facing "Health Passport" app for a community hospital in Ponda, Goa — built as a design-thinking healthcare infrastructure initiative.

Original IP · Solo systems build

AkshaAstra — an AI-native comics studio, built solo

An original Indian-mythology-fused sci-fi noir webcomic universe, produced almost entirely solo using AI production tools — approached not as a shortcut, but as a systems-design problem: replicating the consistency, governance, and discoverability a full studio's art department, dev team, and SEO desk would normally provide.

Next.js / SSRCharacter-consistency pipelineOpenArtStudio Command CenterAEO / GEO strategy
Visit AkshaAstra ↗ Read the full case study →
SYSTEM 01 / PRESENTATION

Website rebuild

Rebuilt on Next.js with true server-side rendering, replacing a client-side shell that was invisible to Google and AI crawlers alike.

SYSTEM 02 / OVERSIGHT

Studio Command Center

A live, filterable dashboard unifying assets, production tracking, and launch planning across four disconnected creative tools.

SYSTEM 03 / GENERATION

Character-consistency pipeline

A canonical-reference locking technique in OpenArt that eliminates AI visual drift across 100+ generations of the same character.

SYSTEM 04 / DATA LAYER

Image manifest architecture

A structured JSON manifest connecting the generation pipeline directly to indexed, production-ready pages — one governed path from generated to live.

Trajectory

Experience

Jan 2026 — May 2026

Creative Director, Marketing

DRAF Club Goa · Contract

Led brand repositioning for a founder-led wellness retreat: a three-track narrative strategy (founder mental health, Goa's wellness ecosystem, AI-driven cognitive-compatibility matching), grounded in psychographic research and FSSAI/ASCI compliance mapping, with a multi-channel distribution plan ready for execution.

Jun 2023 — Oct 2025

Content Marketing Manager

Moon Technolabs · Ahmedabad

Architected an AI-first content strategy: 60% faster production cycles, 43% organic traffic growth, 70–80% top-10 SERP rates across 8–10 monthly technical blogs, 37% lift in lead conversion via NLP-driven persona automation, $450K in influenced pipeline, and 35% faster time-to-close. Nominated Best Manager (Non-Tech).

2016 — 2023

Content & Editorial Roles

Blackboard Radio · Bloop Communications · 9series · Writopedia · Infibeam

Seven years across ed-tech, agencies, and e-commerce — writing for Paytm Mall, Bajaj Finserv, Paisabazaar and international audiences in Japan and South Africa. The editorial craft underneath every AI system's output.

2024 — 2026

PGDM (In Progress)

Goa Institute of Management

Strategy, finance, marketing, and design thinking — the business layer under the creative-technical stack.

Open channel

Let's build something that ships.

Looking for AI Generative Producer and Head of Brand roles — or a project that needs someone who can hold both the creative vision and the technical pipeline.

iampragya17@gmail.com LinkedIn ↗ Download résumé ↓
← Home
Work

The work, in full.

Seven systems — sales enablement, production automation, brand strategy, healthcare infrastructure, and the AI-native comics studio behind all of it. Each section includes the full written case study as a downloadable document.

SYS-00 · Featured · AI Product Architect & Systems Designer

Tri ValueMappers

An AI-powered technical-to-business translation engine — turning a technical feature description into four to five customer-centric sales assets in minutes.

70%faster asset creation (weeks → minutes)
80%per-feature time cut (4–6 hrs → 2–3 min)
12+industries supported in beta
4.8/5beta user satisfaction

The problem

A technical feature like "LSTM-based demand forecasting using temporal pattern recognition" means nothing to a CFO. Sales teams couldn't articulate why a feature mattered; marketing lacked the technical depth to extract the real innovation. Creating sales assets took weeks of back-and-forth between engineering and marketing.

The Tri-Value framework

Every technical feature is mapped across three dimensions of business value: Operational Efficiency (cost, time, manual work reduced), Customer Experience (satisfaction, retention), and Revenue Growth (new or expanded revenue).

What it generates

  • Elevator pitch — a 2-sentence value proposition in a fixed for/struggling-with/delivers/resulting-in structure.
  • Battle card — Hook, "So What?", and Proof Point, formatted for direct sales use.
  • LinkedIn authority post — a 200-word founder-voice post: hook → problem → solution → CTA.
  • Email nurture sequence and competitive battle card for pipeline and positioning.
  • Every asset scored for brand alignment, clarity, and customer-centricity before it ships, with an "Auto-Scout Intent" mode that researches customer pain points automatically.

Why it matters

The differentiator isn't simplification — it's translation to business outcomes. Technical teams focus on building; sales gets what it needs, when it needs it; the framework is reusable across any AI/ML solution.

SYS-01 · Automation · AI Systems Architect & Technical Lead

YouTube Automation & MCP Pipeline

End-to-end orchestration of Claude, Google Drive, and YouTube via Model Context Protocol — removing every manual handoff from AkshaAstra's motion comic production line.

MetricBeforeAfter
Time per scene45 min18 min (−60%)
Manual handoffs3 per scene0 (100% eliminated)
Production velocity~5 scenes/week~15 scenes/week (3x)
Quality consistencyVariable95%+

The build

  • Custom MCP connectors bridging Claude, Google Drive, and YouTube APIs with OAuth 2.0 authorization.
  • Multi-round prompt sequences maintaining character-arc and visual-direction consistency across 200+ scenes.
  • A Node.js orchestration layer with trigger-based automation, retry logic, and monitoring.
  • FFmpeg integration for automated video assembly, encoding, and thumbnail generation.
  • Direct YouTube API publishing with automated metadata, playlist organization, and analytics tracking.

Why it matters

Proof of deep technical fluency connecting LLMs to external systems — systems thinking applied to a real production bottleneck, not a demo. The MCP framework is now reusable for other automation projects, including blog-to-LinkedIn repurposing.

SYS-02 · Scale · Content Operations Architect & Systems Designer

HSB AssetFactory

A three-tier, AI-augmented content production system for a B2B SaaS company — scaling output velocity 3–4x while holding editorial quality through human-in-the-loop checkpoints.

MetricBeforeAfter
Assets / week2–38–10 (3–4x)
Time / asset8–10 hrs2–3 hrs (−70%)
Brand voice consistency~70%95%+
Cost / asset$500–800$150–200 (−70%)

The three tiers

  • Tier 1 — Research: AI agents auto-analyze customer data, competitor content, and market trends into a structured Content Brief.
  • Tier 2 — Generation: Claude drafts multiple angle/tone variations against the brief.
  • Tier 3 — Editorial review: senior editors select, refine for voice and accuracy, and approve before publish.

Why it matters

Operational excellence — scaling production without sacrificing quality, and positioning AI to augment human editorial judgment rather than replace it.

SYS-03 · Strategy · AI Strategy Architect & Multi-Agent Systems Designer

Goa F&B Bot-Stacking Project

An 11-bot orchestration framework producing a complete healthy-snacks category strategy for a Goa-based F&B brand — market research through campaign execution, chained end to end.

11specialized agents orchestrated
2 wksvs. 8–12 wks traditional consulting
70%cost savings vs. external consultants
95%+cross-dimension strategy coherence

The 11-bot ecosystem

  • Tier 1 — Research & Intelligence: Market Research, Ingredient Research, Consumer Psychology bots.
  • Tier 2 — Strategy & Positioning: Brand Voice, Positioning, Pricing Strategy bots.
  • Tier 3 — Execution & Optimization: Copy Generation, Visual Direction, SEO, Campaign Strategy, and Refinement bots.

Each bot passes structured context to the next, with quality gates validating outputs before hand-off, and a final Refinement bot resolving contradictions across all eleven outputs.

Why it matters

Multi-agent orchestration applied to strategy, not just content — proof that decomposing a business problem into agent-sized tasks is itself a strategic skill, delivered at consultant-level output for a fraction of the cost and timeline.

SYS-04 · Content Ops · AI Systems Architect & Full-Stack Developer

AI ThoughtLeader (BlogToLinkedIn Autopilot)

A full-stack app that converts long-form blog posts into a week of LinkedIn content while preserving two distinct founder voices through agentic orchestration.

60%time reduction (8 hrs → <20 min)
2distinct founder voice personas
100%human editorial control retained
Liveused for real proof-of-work demos

Core features

  • Blog input & auto-parse — Claude extracts insights, data points, and "LinkedIn-worthy" hooks from any URL or text.
  • Dual-founder Voice Radar — maps content to "The Architect" (technical, data-driven) and "The Storyteller" (narrative, human-impact) personas.
  • Quality Gate review panel — side-by-side draft review with engagement-prediction scores and a More Human ↔ More AI refinement slider.
  • One-click publishing bridge — exports approved posts as a structured CSV for scheduling tools.

Why it matters

Proof of building a functional AI product, not just prompting — a working demonstration of the "AI-first" operating model: automation for the repetitive work, human judgment where it adds the most value.

SYS-05 · Healthcare Infrastructure · Design Thinking Initiative

HealPonda — ID Hospital Care Platform

A two-part system for a community hospital in Ponda, Goa: a role-based staff dashboard, and "Health Passport," a mobile-first companion app for patients.

Product 1 — ID Hospital Dashboard

Three role-based views switched by login identity — Medical Officer (Care & Admin), Camp Organizer (Coordination), and Resident/patient (Menu). Core features: a facility status board (Working / Limited per service), referral tracking to prioritize which services to restore, live-simulated emergency blood-request dispatch with donor matching and ETA, ambulance dispatch tracking, a patient-journey timeline (Registered → Examined → Referral → Transit → Admission → Follow-up), and full English / Konkani / Hindi localization.

Product 2 — Health Passport

A standalone, mobile-first patient app built around one resident's record: a personal Health Passport home screen, smart health records aggregated from the hospital and community health camps, a health-camp calendar run with Rotary and CSR partners, the same emergency blood/ambulance flows written for a patient audience, and the same three-language translation layer.

Design notes

Both apps share a warm, editorial visual system — Hanken Grotesk for UI text, Newsreader (serif) for greetings and headlines — deliberately calm rather than clinical for a stressed audience. Both deployments are now publicly accessible (Vercel Authentication disabled). All names and figures in the current build are placeholder/demo data; this is a product prototype, not a system processing live patient data.

Original IP · Creative Director / Solo Systems Builder

Building AkshaAstra: An AI-Native Comics Studio, Solo

Four integrated systems — website, dashboard, generation pipeline, and data layer — solving, as a team of one, what a full studio's art department, dev team, and SEO desk would normally provide.

1 — Website rebuild: from invisible shell to search-ready platform

The original site was a single-URL, client-side-rendered app: complete in the browser, but an empty HTML shell to crawlers — invisible to Google and unreadable to AI answer engines (GPTBot, ClaudeBot, PerplexityBot), which read raw HTML and never execute JavaScript. Rebuilt on Next.js App Router with true server-side rendering and static generation; a clean URL architecture mirroring the universe's structure (/worlds/warm/characters/[slug], /stories/[slug]); a three-world visual toggle with a dedicated entry sequence; schema markup, an XML sitemap, and semantic HTML throughout; Google Search Console verified with confirmed rendered content; and GA4 routed through GTM for future-proof tracking. Result: a fully server-rendered, crawlable platform — verified live in Search Console — architected for AI answer-engine visibility (AEO/GEO), not just classic SEO.

2 — Studio Command Center: a live production dashboard

Running a solo, multi-tool pipeline meant assets and status scattered across four platforms with no single source of truth. Built a persisted, filterable dashboard that queries connected creative tools live and renders every asset as one searchable gallery tagged by source and media type, an "AkshaAstra only" filter, plus an in-dashboard production tracker and a marketing launch checklist. Result: one operating view of the whole studio — the function a production coordinator or DAM system serves on a funded team, built for a team of one. Honestly documented: the asset library is genuinely live (re-queried on every open), while the production tracker and content planner remain local until a backend is connected — a clear next step, not a blind spot.

3 — Character-consistency pipeline: solving drift at production scale

The hardest unsolved problem in AI-generated serial storytelling is keeping a character, palette, camera language, and tone consistent across dozens of generations. Built a dedicated OpenArt project with a locked canonical-reference system — named, versioned locks per character, prop, and palette (Rudrakshi's warm teal/amber/ember-red world vs. Tamas Andhakara's cold indigo/obsidian/frost-blue world). The core technique: passing the canonical CDN reference URL directly as a visualReferences array entry to the generation API — not re-describing the character in text, and not relying on named handles, which resolve only in the browser UI and not the API. Combined with explicit continuation prompting, disabled auto-enhancement (which was silently overriding locked style rules), and a living ~29-page Universe Codex governing canon. Result: a reproducible system for visual continuity across an ongoing serialized universe — now the credibility backbone of a parallel consulting offer, "The Continuity Problem."

4 — Image manifest & search architecture

Generated assets are only useful in production if every component can reliably find the right image, and every public asset needs wiring into an indexable page. Built a structured image manifest (JSON) mapping every canonical slug to its hotlinked production CDN URL and thumbnail — sourced directly from the OpenArt production project, no local downloads — with a consistent access pattern across the codebase and a pre-deploy QA step rendering all mapped thumbnails against the entity list. Result: a single, disciplined data layer connecting the asset-generation pipeline to a stable, search-indexed site, so new characters and lore entries go from "generated in OpenArt" to "live and indexed" through one governed path.

How the systems connect

LayerSystemSolves
GenerationOpenArt canonical-reference pipelineCharacter/palette drift across generations
DataImage manifest (JSON)Reliable asset addressing, generation → production
PresentationNext.js SSR/SSG websiteSearch & AI-crawler invisibility
OversightStudio Command CenterFragmented, multi-tool production, no single source of truth

Why this matters

Most "AI-assisted" creative work is a single prompt producing a single asset. AkshaAstra is a governed, documented, repeatable production system — the discipline a studio's art department and dev team would provide — built and run solo across worldbuilding, image generation, web engineering, and search strategy simultaneously.