How Partython grew.
A high-level look back at how the platform came together. We don't publish a release-by-release log yet — this is the story told in broad strokes, milestone by milestone.
Development in progress
Partython AI is actively built and shipped every week. A detailed, versioned release log is on our roadmap; until then, this page is the honest summary — turning points, not individual fixes.
Platform foundation
The multi-tenant core was built — shared infrastructure with each business's data strictly isolated at the database. Automated, key-less cloud deployment was set up so every change ships safely.
The public website
The marketing site went up — product pages, pricing, documentation, blog and careers — giving businesses a clear, self-serve way to understand and try the platform.
Channels go live
WhatsApp, Instagram and Messenger were connected through official, sign-in-based flows — no tokens to copy. Businesses could now put an AI agent on the apps their customers already use.
The Brain Engine
The knowledge engine matured — upload your catalogue, prices and policies, and the agent answers grounded in your own content. Tools and safety guardrails let it book, look things up and escalate when it should.
A developer platform
A public API and JavaScript, Python and React SDKs were released, alongside scoped API keys and an API portal — so teams and agencies can build directly on Partython.
Billing and trust
Self-serve subscriptions, plan entitlements and usage metering came together. Formal security work began in parallel, including a SOC 2 Type II audit that is currently in progress.
Built for India
Hindi language support, UPI and local payment options, and industry templates tuned for Indian businesses — meeting our first customers where they actually are.
Industry playbooks
A marketplace of industry-specific templates, per-vertical setup guides, and an internal quality benchmark so agents start strong and keep improving for each kind of business.
A note for whoever maintains this page
This timeline is written by hand, on purpose, at a deliberately high level. To refresh it later: review the project's git history, pick out the genuine turning points — new product surfaces and capabilities, not individual bug fixes — and express each as one generic milestone. Keep the granularity at the year-and-season level, leave out exact dates and internal detail, and fold related work into a single theme. The goal is a story a customer can read in a minute, not an engineering log.