Private, local AI workspace

Your AI. Your machine. Served locally.

PizzaAI runs on your computer. Your files, prompts, and projects stay there. Nothing changes until you approve it.

No cloud inference. No per-prompt fees. No surprises.

Built for coding today. Automating everyday work next.

Join the beta See how it works
PizzaAI · local workspace
The PizzaAI home screen running on a local Ollama runtime, with a task composer set to Build mode
Runs locally through Ollama
Projects stay on your machine
Changes wait for approval
Open-weight models

Discuss, Plan, Build

Choose how far PizzaAI goes.

Every task starts in one of three modes, so you decide up front how much rope the model gets. None of them touch your project without approval.

Ask and explore. Reads your project, changes nothing.

Task composer
PizzaAI task composer in Discuss mode, reading the project without changing it

How it works

One workspace. Every step visible.

Ask for what you want, watch PizzaAI work through your project, and approve the result. Three steps, nothing hidden in between.

01 · START

Ask for the outcome.

Describe what you need in plain language. PizzaAI keeps the active project and workflow controls in one focused view.

Home
The PizzaAI home screen with a multi-line request typed into the composer in Build mode
02 · UNDERSTAND

See the work, not a black box.

Watch the active task, project sources, and agent activity as PizzaAI reads the context that matters.

Task workspace
PizzaAI task workspace showing active agents and project context
03 · REVIEW

Keep the final say.

Proposed edits stop at a review step. You can inspect what is changing before anything reaches your project.

Approval review
PizzaAI approval review with changed files and approval controls

Local by default

Your work stays on your machine.

PizzaAI runs models through Ollama on your own hardware. Your project context stays local. Your installed models stay yours. Normal local use never runs up a per-prompt inference bill.

Local runtimeHEALTHY
  • InferenceON DEVICE
  • Model providerOLLAMA
  • Project contextREMAINS LOCAL
  • Usage meterNO TOKEN BILL

Approval built in

See exactly what will change.

Every proposed edit and command stops for review first. Approval is a step in the workflow, not a setting you have to go find.

01See proposed files and edits
02Review commands and checks
03Apply approved work intentionally
Approval review
PizzaAI approval notice showing proposed edits and a review action

Coordinated work

Specialists when the task needs them.

PizzaAI routes work to agents focused on planning, building, reviewing, and testing. Describe the outcome and PizzaAI coordinates the rest.

PLAvailable

Plan

Breaks a request into a clear, workable approach.

BUAvailable

Build

Proposes focused changes against the active project.

REAvailable

Review

Inspects the work and surfaces issues before completion.

TEAvailable

Test

Runs approved checks and reports what passed or failed.

Current and coming next

Coding today. Automating everyday work next.

The current beta centers on software projects. The next phase extends the same private, approval-first approach to automating everyday work on your machine.

Available in the current beta

Current
  • Local project understanding
  • Repository search and reading
  • Discuss, Plan, and Build modes
  • Proposed file changes
  • Approval-gated commands
  • Task history and agent management

Coming next

Not yet available
  • Organizing and processing local files
  • Working with documents and spreadsheets
  • Generating recurring reports
  • Repository maintenance workflows
  • Scheduled or folder-triggered workflows

Coming-next capabilities are product direction, not features available in the current beta. Scope and timing may change.

Private beta

Join the private beta.

Early testers will help shape local setup, coding workflows, review controls, and the path toward automating everyday work.

Private beta, opening soon

Product updates and early-access invitations only. Read our Privacy Notice.

FAQ

Questions, answered plainly.

Does PizzaAI require an internet connection?

Core project workflows use models running locally through Ollama. Installation, updates, license checks, and website links may still require internet access.

Does PizzaAI upload my project files?

PizzaAI is designed to keep project context on your device for local workflows. External integrations, if added later, would be clearly identified.

Will I pay per prompt?

No. Local inference does not run through a cloud token meter. You bring the hardware and the models.

When can I try it?

PizzaAI is preparing for a small private beta. Join the beta list to receive access updates.