The agent that thinks ahead.

Your own AI agent that doesn't just answer, it gets work done. Open source, with memory and real tools. On your machine, with local models or an API key.

Install for macOS

25,000+ downloads·Open source, AGPL-3.0·macOS, Windows & Linux

Check supplier contracts

Read the supplier contracts in the Purchasing folder and tell me which notice periods end in the next 90 days.

Thinking Process3.1s

list_filesPurchasing/Contracts14 lines

read_fileLogistics_Framework.pdfDone

read_filePackaging_2025.pdfDone

Three notice periods end in the next 90 days:

Contract	Cancel by
Logistics framework agreement	11/15/2026
Packaging materials	11/30/2026
Printing	12/12/2026

Should I add each deadline to the calendar two weeks ahead?

Yes, all threeOnly the firstNo

tokens: ●○○○○○○○○○ 14% (18,240/128,000)

Ask anything...

Hand work over instead of operating a tool.

Less chatting, more done. Hand off the task, carry on, collect the result. You focus on what matters, your agent does the rest in the background.

Works around the clock

At set times or nonstop in the background, even while you're away. The result is ready when the run ends.

Long-term memory

The essentials from whole PDFs, even hundreds of pages long, and from your conversations. The knowledge stays, ready whenever a task needs it.

Works as a team

Large tasks, split into steps and run in parallel. Via an agent-to-agent protocol, also alongside Claude, Codex or OpenCode.

Research AgentrunningSearching web 3/5

RESEARCH REPORT

Market overview: recycled packaging

1,240 / 3,000 words · 12 sources

Suppliers and capacity

The analysis compares suppliers by capacity, price and delivery time. [1] Three suppliers cover most of the requested volume. [2]

Prices differ clearly by region. [4]

OUTLINE 2/5 done

Overview

Market size

Suppliers and capacity

Prices

5Outlook

SOURCES 12

[1]destatis.de

[2]eur-lex.europa.eu

[3]umweltbundesamt.de

Teaches itself new tools

It runs them, reads the results and adjusts until it can use them on its own, custom ones included.

Your machine, your server, your call

Local model, your own server, or one API key for every provider. You decide where the data lives and which model does the work. No per-seat subscription.

Local-first AI agents

What it means to run an agent on your own machine, and why local-first matters for privacy, cost and control.

Read the explainer

Local-first by design. You decide where VAF runs.

VAF runs on your machine, on your own server, or as a framework inside your code. Memory and keys live where the agent runs.

For power users

Desktop app

Install it, sign in, and the agent is in its own window.

Your data and keys stay on your machine
Local models, or one API key for everything
Schedules, memory and tools built in

Install the app

For teams & companies

Self-host it

Install it, share it, and VAF runs on your server at your own address.

One host serves the whole team over HTTPS
Separate accounts with isolated memory
No per-seat subscription

See how it works

For developers

Build on the framework

Import it as a Python library for your own automations, agents and software.

One pip install, provider-agnostic
Your own tools via BaseTool or MCP
Ship plugins under your own license

Explore the framework

The agent on the go.

More than a chat. Your agent, on the go. Web search, automations and a notification when something's done. History and memory are stored on your device, protected by your phone's encryption.

Voice messages and live calls
Automations with notifications
History and memory stored on the device

SoonFor iPhone on the App Store.

The main agent

You ask in plain language, and the main agent plans the work and delegates each part to the specialist sub-agents below.

Market report on recycled packaging

Research the market for recycled packaging and write me a report with sources.

Thinking Process2.0s

research_agentMarket overview: recycled packagingrunning… 42s

The research agent is looking for sources and writing the report. I’ll get back to you as soon as it is done.

tokens: ●○○○○○○○○○ 9% (11,820/128,000)

Ask anything...

Research AgentrunningSearching web 3/5

RESEARCH REPORT

Market overview: recycled packaging

1,240 / 3,000 words · 12 sources

Suppliers and capacity

The analysis compares suppliers by capacity, price and delivery time. [1] Three suppliers cover most of the requested volume. [2]

Prices differ clearly by region. [4]

OUTLINE 2/5 done

Overview

Market size

Suppliers and capacity

Prices

5Outlook

SOURCES 12

[1]destatis.de

[2]eur-lex.europa.eu

[3]umweltbundesamt.de

VAF │ amber482913 ready◉ mic │ veyllo · veyllo-chat

You · 14:02

Which issues in shop-backend are still open?

[ ] VAF · 14:02

· thinking

⏺ github_list_issues(state="open")1.4s

Seven are open, two of them marked as bugs:

#48 Checkout rounds discounts the wrong way

#51 Export fails on empty fields

The other five are feature requests.

▍enter send · ctrl+j newline · tab complete · up history

S Settings C Model L Voice T Theme H History ? Help /exit QuitContext ▰▰▱▱▱▱▱▱▱▱ 12% (15,360/128,000 Tok)

Your own agent, three lines in.

The desktop app runs on this framework, and you build on the same primitives. Install it from PyPI and build your own automations, agents and tools on top.

Provider-agnostic. The same code runs a local GGUF model or any API provider; swap with one config line.

Bring your own tools. Subclass BaseTool or wire in MCP servers, and the agent picks them up at startup.

Your license, your product. Plugins, tools and workflows built on the public interfaces may ship under your own terms.

Every model your agent needs, behind one key.

VAF needs models for text, vision and speech. Normally that means an account with every provider, every key wired in by hand, a separate invoice per provider and no combined view. The Veyllo API bundles that into one key and one bill.

One key instead of many. Every model VAF uses, behind a single API. No separate accounts or keys at each model provider.

No models to host yourself. No GPUs, no self-run model servers. Veyllo handles compute and model hosting.

Built for production. Keys, rate limits and usage metering from day one.

Get an API key Read the preview docs

terminal

$ curl https://api.veyllo.app/v1/chat/completions \
  -H "Authorization: Bearer $VEYLLO_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "veyllo-chat",
    "messages": [
      { "role": "user", "content": "Plan my week." }
    ]
  }'

# vision: add an image to any message
$ curl https://api.veyllo.app/v1/chat/completions \
  -d '{ "model": "veyllo-chat", "messages": [
    { "role": "user", "content": [
      { "type": "text", "text": "What is this?" },
      { "type": "image_url",
        "image_url": { "url": "data:image/jpeg;base64,…" } }
    ]}
  ]}'

Endpointapi.veyllo.app/v1

Compatible withOpenAI SDKs

Modelschat · vision · transcribe

Pricingpay as you go

Chat

€0.90 in · €1.90 out

per 1M tokens · final price, no VAT charged

Vision

€3.45 in · €20.70 out

per 1M tokens · final price, no VAT charged

Transcribe

€0.012 / min

billed per second · final price, no VAT charged

Billing

Pay as you go

prepaid credits · VAT invoices · full pricing →

Bring the agent to your desktop.

Two steps: clone the repository and run the installer. You pick the platform; VAF sets up the rest: container runtime, Python and Node.

macOSApple Silicon & Intel · install scriptInstall guide WindowsWindows 10/11 · install scriptInstall guide LinuxDebian, Ubuntu & Arch · install scriptInstall guide

Follow the build on GitHub.

Read the code, follow the commits, and raise issues that reach the developers directly.

25stars on GitHub

25,000+downloads and counting

Activedevelopment

View on GitHub

1 dot ≈ 3 agents deployed

Growth since the alpha · Jun 26, 2026

Stars25

Downloads25,065

Data was not previously recordedData Fetching

Jun 26Aug 28Oct 30

Published recently.

All posts

Oct 28, 2026Article

Ready for your agent?

The desktop app installs everything it needs; the Veyllo API supplies the models.

Install for macOS

Get an API key

The agent that thinks ahead.

Hand work over instead of operating a tool.

Works around the clock

Long-term memory

Works as a team

Teaches itself new tools

Your machine, your server, your call

Local-first AI agents

Local-first by design. You decide where VAF runs.

Desktop app

Self-host it

Build on the framework

The agent on the go.

The main agent

CLI mode

Your own agent, three lines in.

Every model your agent needs, behind one key.

Bring the agent to your desktop.

Follow the build on GitHub.

Published recently.

How VAF works in a tax advisory firm

How VAF works in a craft business

How VAF works in a law firm

Ready for your agent?