The platform itself — a multi-tenant business system with per-organisation roles, an audit spine, and a domain-package architecture covering CRM, projects, people, payroll, billing, events and more.
Chitram.AI Private Limited
We build the systems Indian businesses actually run on
Chitram.AI Private Limited builds operating systems for work that has outgrown spreadsheets. Two products under one roof: Ojas, a multi-tenant business platform and the ventures on top of it, and Chitram.AI, a creative production system for studios — plus the client systems that taught us what both needed to be.
- 14 yrs
- Building software
- 2x
- Founder
- 20x
- Revenue growth enabled
- 95
- Domain packages
Full-stack from 2012, across consulting, consumer, fintech, media and now platforms.
Firebolt Labs in 2014, Chitram.AI today — with a CTO tenure in between.
At The Wedding Brigade, from the platform and team built there as CTO.
In the Ojas workspace. A subset is served by the live API today; the rest are models and services.
Countable figures only. Where a number needs a caveat to stay honest, the caveat is on it.
Why the company exists
Indian businesses are running more complex operations than ever, on software stacks assembled from tools designed elsewhere for someone else's assumptions. A CRM built for enterprise software sales. An HR suite that treats a contractor as an exception. An accounting product that has never heard of the festival calendar your revenue actually follows.
Each tool is defensible on its own. The cost lives in the gaps between them — the week spent reconciling what a project really earned, the artist double-booked across two productions, the compliance deadline nobody owned. We kept meeting the same problem in different industries, and it was always the same shape: the systems did not share a record, so people became the integration layer.
So we build platforms where the record is shared by construction. Ojas does it for business operations generally. Chitram.AI does it for creative production specifically, where the work being managed is also the work being made.
The platform
Ojas — the business platform underneath
Ojas is a multi-tenant business platform built as self-contained domain packages: organisations and roles, CRM and pipelines, projects and sprints, people, attendance and payroll, billing and payouts, events and bookings, compliance and audit. It is the company's largest engineering effort, and the reason Chitram.AI's operations layer is a port rather than a from-scratch build.
Several ventures run on it, each at a different stage. We label them honestly, because a platform that overstates its own maturity is not one you should trust with your payroll.
In development
Helps entrepreneurs and established businesses grow through events — expos, showcases and festivals — with the software to actually run them. Currently being rebuilt on the commerce foundation.
Live, partial
A curated credit marketplace with a named consultant. It is not a lender and does not underwrite: it compares options and connects you to people who do.
ojas.legal
Entity records, a compliance calendar and a document vault. Assisted filing, not a licensed practice — generated documents still need professional sign-off.
The studio product
Chitram.AI — and what else is in the workshop
Chitram.AI is the creative production operating system this site is about: vector, script, storyboard and culturally grounded generation, with a versioned brand registry and a review step before anything ships. Those modules are available today. The studio operations layer — client CRM, projects, talent, billing — is the Ojas platform being brought across, which is why we can describe it precisely and still tell you it is not here yet.
PujaVani
A traditional Panchanga calculator with a modern interface, computing Hindu Panchanga for any date from 5000 BCE to 5000 CE and any location.
Shabdashala
An early-stage content and commerce property. Still in development — we would rather name it here than describe a product that does not exist yet.
Client work
We ship these systems for other people too
Product work and client work feed each other. Building a real operations system for a business with real deadlines is how you find out which assumptions in your platform were wrong.
Party Cruisers Limited
LiveA complete enterprise management platform for one of India's established event companies: lead and campaign management, employee attendance, leave and expenses, applicant tracking, and content and master data — a single organisation-scoped service replacing what had been several.
Built and owned by Chitram.AI Private Limited as the developer, with the client owning their business data. We put that split in writing at the start of every engagement, because “who owns the code” is not a conversation to have later.
How we work
Four commitments we would like to be held to.
One record, not ten exports
A business does not experience its work and its books as separate problems. The shot, the artist who made it, the milestone it belongs to and the invoice it justifies are one chain of facts. Software that breaks the chain makes people the integration layer, every month, forever.
Say what ships
Every module on this site is marked available or roadmap, and every venture above carries its real status. Roadmaps move. We would rather tell you that in the first meeting than have you discover it during implementation.
Your data and your canon stay yours
We do not train public models on customer content, and everything is exportable in standard formats. When we build a system for a client, we are explicit in writing about who owns the code and who owns the data.
Built for how Indian businesses actually run
Contract-heavy crews, GST, multi-entity groups, festival calendars, regional languages. These are not localisation tasks bolted onto a Western product — they are the assumptions the software starts from.
Where the depth is
Seven things we have had to get genuinely good at
Not a services menu. These are the problems the work above forced us to solve properly, each with a shipped system behind it.
Multi-tenant platform architecture
Organisation scoping, per-org roles and an append-only audit spine designed as the kernel every domain package is written against — not retrofitted once a second customer appeared.
Ojas · Party Cruisers Limited
Domain-driven package design
Ninety-five self-contained domain packages with explicit dependency rules, so a capability can be added or unplugged without the rest of the system noticing.
Ojas workspace
Indian business operations
Contract-heavy crews, GST-valid invoicing, multi-entity groups and compliance calendars treated as first-class requirements rather than localisation applied to a Western product.
Ojas · Party Cruisers Limited
Generative AI in production, since 2020
Shipping on GPT-3 and GPT-4 for API design, code generation and behavioural analytics before the current wave — then constrained generation answering to a versioned registry, repeatable enough to schedule around.
Chitram.AI · Conektto · Riff Analytics
Real-time systems at public scale
Led the technology revamp of ESPNcricinfo via ESPN/Disney — live data, heavy concurrent traffic, and a failure mode where degradation is noticed by the public within seconds.
ESPNcricinfo
Cultural and computational precision
Sourced corpora with retrievable attribution, and astronomical computation accurate across a ten-thousand-year range — domains where being approximately right is being wrong.
Chitram.AI IKS engine · PujaVani
Interfaces that hold up
Design systems on semantic tokens, WCAG 2.1 AA verified by measuring rendered pages rather than by eye, in light and dark and across four brand themes.
This site · the Ojas front ends
Founder & CEO
Bala Subrahmanyam Varanasi
A second-time founder who has built for students, for weddings, for cricket at international scale, and now for the studios and businesses that run India's creative and commercial work.
Chitram.AI is founder-led and engineering-led. The platform architecture, the domain package design and the honesty rules this site is written under come from the same place — which is why you get a straight answer about what is built when you ask.
Connect on LinkedIn (opens in a new tab)The route here
2011 – 2014
Full-stack engineering and open sourceAgiliq · Crypsis Technologies
Built products for clients across several countries in Python, Django and Rails, from requirements through roadmap. Contributed to open source along the way, including scikit-learn, joblib and django-graphs, plus Android work shipped for Microsoft.
2014 – 2015
First venture: Firebolt LabsFounder
Built a food ordering application serving a student population of 25,000 within a five-kilometre radius, aimed squarely at food wastage and unwanted mess bills. Ran the company and its consulting work alongside it. First lesson in what still drives the work: the operations around an order are harder than the order.
2015 – 2017
Scale and delivery disciplineTinyOwl · Coverfox Insurance
Broke a monolithic food ordering platform into distributed services — inventory, location, notifications — then built car and bike insurance products and stood up the CI/CD pipeline behind them. Regulated products teach you what "correct" costs.
2017 – 2020
Chief Technology OfficerThe Wedding Brigade
Replaced a fragmented PHP estate with a single scalable architecture, then hired and led the in-house full-stack team that shipped on it. The web and mobile products from that period enabled 20x revenue growth. Three years inside a business where every transaction is a coordinated event with vendors, timelines and people.
2020 – 2023
Lead engineer, independentESPNcricinfo · Conektto · Riff Analytics
Led the technology revamp of ESPNcricinfo — via ESPN/Disney and Proximity Labs — for real-time data at international traffic levels, where an outage is a public event. In parallel, built production systems on GPT-3 and GPT-4 for API design and code generation at Conektto, and behavioural analytics at Riff Analytics. Shipping LLM products before the current wave, not after it.
2022 – present
Second venture: Chitram.AIFounder & CEO · incorporated June 2024
Began building in late 2022 and incorporated once there was something worth incorporating around. Generative AI inside real production workflows: plot, dialogue and character drafting, multilingual scripts so actors read in their own language, and script-complexity analysis for realistic schedules and resourcing. Ojas, Chitram.AI and the client systems all date from this run.
Where we are today
Chitram.AI's creative production modules are live. Its studio operations layer is in development. Ojas runs several ventures at varying maturity, and the client systems are in production use. That is the whole picture, including the parts that are not finished.
We are working with a small number of studios and businesses while the operations layer lands. If you run a production or a company and are willing to be specific about what is broken in your process, that is exactly the conversation we want.