Live product demoPrivate source

Tonalli AI — A multi-model AI workspace

A lab product that brings chat, tools, files, voice, messaging, artifacts, and scheduled prompts into a shared workspace.

The problem

AI-assisted work becomes fragmented when models, tools, documents, voice, messaging, and reminders live in separate products.

My role

Independent product builder — interaction design, model and mode routing, tool workflows, persistence, messaging reliability, and QA.

Contributions

01
Designed the model, mode, streaming-chat, and agent-tool experience.
02
Connected saved chats, agents, memory, files, voice, messaging, and artifacts.
03
Hardened scheduled delivery, Telegram fallback behavior, and visible delivery status.

Product capabilities

01
Streaming multi-model chat with search, URL, calculation, and sandboxed-code tools.
02
Voice, document Q&A, custom agents, and rendered artifacts.
03
Persisted chat context and scheduled delivery across web and Telegram.

Product decisions

Preserve a local-first fallback when account sync is not configured.
Persist schedule outcomes and retry Telegram as plain text when formatted delivery fails.

Proof gallery

Tonalli AI / 2026

Approved captures of public product surfaces. No customer, tenant, account, prompt, analytics, or provider configuration is shown.

Technology