Aether Mind — Wrocław & Hyderabad
We build AI systems which write the books!
Aether Mind is the company behind PenMyBook.ai, a production system that turns a topic into a finished, long-form technical book — researched, written, and bound end to end by a multi-agent AI pipeline.
This page exists for one reader: the Azure OpenAI partnerships team reviewing our request for GPT‑5.5 model access. It lays out what we've built, who's building it, and why this specific workload needs this specific model.
PenMyBook.ai
A person gives PenMyBook.ai a topic, a few seed sources, and a chapter outline. From there, a pipeline of specialized agents takes over: one plans the book's structure, one researches each chapter against real sources, one writes the prose, and one reviews it — before the whole thing is bound into a finished, print- and Kindle-ready book.
It is live in production today, generating complete books for real users — not a demo, not a prototype pipeline run once for a screenshot.
Eighteen years of enterprise delivery, applied to AI
Aether Mind's founder, Appurv Saini, spent his career leading supply-chain and warehouse-execution programs for global manufacturers — the kind of work where a botched go-live stops a production line. That same delivery discipline — careful scoping, phased rollout, hypercare after launch — is what governs how we ship AI systems.
- Pharmaceutical manufacturer (US rollout)S/4HANA embedded EWM
- CanpackDecentralized EWM migration
- Arla FoodsLabor management & forecasting
- Ontex GlobalGlobal EWM template rollout
- McCormickWarehouse automation & MES integration
- EY Global Delivery ServicesDecentralized EWM, technical lead
- 3MEWM/TM/GTS, IT team manager
- Deloitte · HCL · DuPont · DellSAP SCM & supply chain delivery
The ask, plainly
Generating a full-length technical book isn't a single prompt. It's dozens of chapters, each reasoning over a large body of retrieved research, holding a consistent voice and continuity across tens of thousands of words, and coordinating multiple agents across planning, research, writing, and review. We currently run this pipeline in production on GPT‑5.4 via Azure OpenAI Service.
GPT‑5.5's longer effective context and steadier long-horizon reasoning are a direct fit for this workload, not a nice-to-have.
Higher quality, fewer retries
Better instruction-following across a 40+ chapter generation run means less rewriting and fewer flagged sections per book.
Lower latency at our scale
We operate a queue-based generation pipeline today; elevated quota reduces per-chapter wait time without changing that architecture.
Headroom to grow
Book and chapter volume can scale without degrading the quality bar we hold today.
Not a cold start
This is quota for a workload already running in production, with an existing Azure OpenAI integration.
Two-person leadership, complementary disciplines
CEO & Founder
Appurv Saini
18+ years leading enterprise SAP supply-chain and warehouse-execution programs for global manufacturers, including Arla Foods, 3M, DuPont, and McCormick. Applies that same delivery discipline — scoping, phased rollout, hypercare — to shipping a production AI system.
LinkedIn ↗CTO
Abhijeet Mishra
Senior Applied AI Engineer & Technical Architect. Ten years in software engineering, the last three focused on applied AI — building production agentic systems and LLM platforms, and training transformer models from scratch. Designs and operates the multi-service architecture behind PenMyBook.ai.
LinkedIn ↗Built and run on production infrastructure
PenMyBook.ai runs as a set of independently deployed services on AWS, behind managed load balancing and TLS, with a queue-based generation pipeline that isolates long-running AI work from the user-facing API. This is live infrastructure serving real traffic today, not a local demo spun up for this request.
Get in touch
For questions about this request or about Aether Mind, reach us directly.