Pro tip: Use ⌘K to quickly search anything in Kazi
Pro tip: Use ⌘K to quickly search anything in Kazi
Pro tip: Use ⌘K to quickly search anything in Kazi
KAZI puts 13 world-class AI models (Claude Opus 4.8, GPT-5.4, Gemini 3.1 Pro and more) inside one connected work ecosystem that already knows your projects, clients and invoices. Chat with your real data, run agents and a copilot that act on it, write invoices and proposals, generate content and video, transcribe meetings, and bring your own API key.
Free tier includedBring your own keyGrounded in your real data
One prompt to on-brand images, copy, voiceover and music, across 13 models.
Pick the model for each task, or let KAZI choose the best one for you and switch automatically if one is busy. A free tier handles everyday chat and text, or bring your own key to run any provider directly.
Picks the best model
KAZI sends each request to the best available model for the job, automatically.
Never gets stuck
If one model is unavailable, KAZI switches to another. No dead ends.
Your key or ours
Run on your own encrypted API keys, or the built-in platform models.
Four clusters of real capability (Create, Assist, Understand, Communicate), each grounded in the work you already do. Follow any card to its full feature page.
One studio turns a prompt into the deliverable (copy, images, music, and video), then saves it straight to the project you can bill for.
On-brand images, logos, social graphics and mockups from one prompt, plus copy, voiceover and music across all 13 models, saved to your projects.
ExploreLong- and short-form content in your brand voices, with humanize, fact-check, and tone and quality scoring built in.
ExploreGenerate and edit images in the studio, powered by fal.ai (Flux, Nano Banana) and Gemini.
ExploreScore a video or spin up a track, powered by fal.ai and the self-hosted ACE-Step engine.
ExploreReal-time green-screen background removal and auto-captions today, plus AI clip generation powered by fal.ai.
ExploreA copilot that reads your workspace, and agents that act on it, so the AI does the work, not just the talking.
Chat with your real data (projects, clients, invoices) and get instant answers grounded in your own workspace, not generic guesses.
ExploreMulti-step agents (planner, executor, reviewer) that act on your data and knowledge base. Entitlement-gated by plan.
ExploreDescribe an app and get a runnable multi-file project with live preview, ZIP export and one-click GitHub push, built on GPT-5.3-codex.
ExploreTurn calls, documents and raw numbers into structured, actionable knowledge, automatically.
Automatic transcription turns any call into a summary, action items and tasks, sent straight to your project board.
ExploreSnap a photo of a supplier bill and the AI reads it back as structured data (line items, totals and dates) ready to reconcile.
ExploreThe assistant draws on your real projects, clients and finances, so its answers reference your actual records instead of guessing.
ExploreReply faster, cross language barriers, and add a human voice, right inside the conversations you already have.
Context-aware reply suggestions in messaging, drawn from the client and project the thread belongs to.
ExploreTranslate chat and documents to work with clients across languages, an assistive layer inside your conversations.
ExploreNatural text-to-speech in six OpenAI voices for voiceovers, narration and audio replies.
ExploreKAZI’s AI is not a bolt-on chatbot. It lives inside the modules you run your business on and reads from the same source of truth, so it acts with real context, not guesses.
Draft a full invoice from a sentence. The AI reads the project and client, and even scans a supplier bill via OCR.
Generate a branded proposal with an AI win-probability estimate, straight from the deal you are working.
Spin up a project draft (tasks, milestones and briefs) from a chat prompt or a meeting transcript.
Smart replies and translation sit inside every client thread, aware of the work behind it.
AI insights, predictive analytics and revenue intelligence run over your real numbers, not sample data.
Green-screen backgrounds, captions and AI enhancements live inside the video studio your reviews attach to.
Your data works for you, answering your questions, not training someone else’s model.
Use your own OpenAI, Anthropic, Google or OpenRouter key (stored AES-encrypted), or run on the platform models. Your choice.
Your workspace data is used only to answer your own questions. It is not sold or used to train third-party AI models.
Pick the model for each task, or let KAZI choose the best one for you and switch automatically if one is busy. No lock-in to a single provider.
A free instruct-model tier covers everyday chat and text; paid plans lift limits and unlock heavier models and agents.
Straight answers about models, pricing, privacy and what you can actually do.
KAZI runs a 13-model engine you can pick from: Claude Opus 4.8 (1M-token context), Opus 4.6, Sonnet 4.6 and Haiku 4.5 from Anthropic; GPT-5.4, GPT-5.4-mini, GPT-5.3-codex and GPT-5.1-instant from OpenAI; Gemini 3.1 Pro (2M-token context), Gemini 3 Flash, and Gemini 2.5 Flash and Flash-Lite from Google; and Grok 4.1 Reasoning from xAI. KAZI picks the best available model for each task automatically, and switches to another if one is unavailable.
Yes. KAZI includes a free instruct-model tier for everyday chat and text generation, so you can use AI without adding an API key or a paid plan. Heavier usage, the most powerful models, media generation and AI agents are governed by fair-use, tier-gated limits on paid plans, or you can bring your own API key.
Yes. KAZI supports bring-your-own-key for OpenAI, Anthropic, Google and OpenRouter. Your key is stored AES-encrypted and used only to run your requests, so you can pay your provider directly and lift platform limits. If you prefer, you can skip BYOK entirely and use the built-in platform models.
Yes. Image generation is powered by fal.ai (Flux and Nano Banana) plus Gemini; music generation is powered by fal.ai and the self-hosted ACE-Step engine; and video generation is powered by fal.ai, alongside live real-time green-screen background removal and auto-captions in the video studio. Text-to-speech gives you six natural voices. Media generation runs through those providers, so output depends on your plan and provider availability.
Yes. From a chat prompt the AI can draft a complete invoice, reading the linked project and client, and generate branded proposals with an AI win-probability estimate. It can also scan an existing supplier bill to pull line items, totals and dates into structured data, and draft project plans from a conversation or meeting transcript.
No. Your workspace data is used only to answer your own questions about your projects, clients and finances. It is not sold or used to train third-party AI models. With bring-your-own-key your keys stay encrypted, and you control which model handles each task.
The AI Assistant securely looks at your real projects, clients, invoices and documents when you ask a question. That means answers are grounded in your actual records, so the AI can reference a specific client or project rather than giving generic responses.
Freelancers get the full 13-model engine plus the AI Assistant that knows their work, invoice and proposal drafting, invoice scanning, the AI Content and Create studios, AI Code Builder, meeting intelligence with automatic transcription, smart replies, translation and text-to-speech, all in one workspace. The free tier covers everyday chat and text; paid plans and bring-your-own-key unlock heavier models, media generation and agents.
One workspace that already knows your projects, clients and invoices, with a free tier and bring-your-own-key. See what AI can do when it has real context.