PLIA

What PLIA does

PLIA is Push & Lift's AI assistant. Ask by typing or speaking, and it answers.

• It knows about driving, setups, strategy and Push & Lift apps.
• It reads photos and CSV files you send.
• It can speak aloud and listen through your microphone, even during a race.
• It identifies the topic behind each tip so you can check your car's manual. It tells you when it does not know.

You can use PLIA for free. Choose its AI once. All three options are explained in 3. First use:

• Modelos locales (Local models): download a model with one button, for example PLIA 1 Mini or PLIA 1 Max, and run it on your own graphics card. Free, no key required.
• An API key from an AI provider. NVIDIA NIM, Google AI Studio and Groq offer free keys. Others, such as OpenAI or Anthropic, require payment.
• Your Claude Code or Codex account, if you already pay for one of those plans.

PLIA has several modes (PLIA, Setups, Coach, Spotter, Mi PC, Liveries, Vídeos, Directo, Coder and Fábrica). Each offers a different way of working. The normal mode is PLIA: your engineer, who helps you and distributes the work.

Opening it from the Suite

1. Open the Push & Lift screen from the bar logo.
2. In the left menu, group PIT, click PLIA.
3. PLIA opens in its own window, with its own Windows taskbar button.

Only one PLIA window opens. If already open, it does not open another.

PLIA uses the language selected in the Suite.

First launch takes a little longer

At first launch, PLIA prepares its engine. The waiting screen shows:

• «Preparando PLIA por primera vez: descomprimiendo su motor (… de …)» with the progress percentage: preparing PLIA and unpacking its engine.
• A progress bar and timer: «Llevo … segundos.» (elapsed seconds).
• «Sólo pasa la primera vez: después abre enseguida. No lo cierres.»: this only happens once; do not close it.

Do not close it. Once ready, the normal PLIA screen appears. Later launches are immediate.

If it does not open

If startup fails, the window shows PLIA no ha podido abrir (PLIA could not open), explains the problem and asks you to close and reopen PLIA. If it persists, send the error through the program's Discord so it can be fixed. The COPIAR EL ERROR (COPY ERROR) button copies the error for pasting.

PLIA requires an active Push & Lift licence. If none is found, Windows asks you to open Push & Lift, go to Ajustes > Tu cuenta (Settings > Your account), check your account, purchase and connection, then reopen the app.

PLIA buttons on the Suite bar

The Push & Lift bar has K, GPT, microphone, speaker and crossed-out microphone buttons. They let you talk to PLIA without opening its window, for example during a race. See the SUITE manual, The bar section. Inside PLIA the same controls are in the message box (see 4.5).

First use: give PLIA an AI

The welcome window

The first time, the Bienvenido a PLIA (Welcome to PLIA) box appears («Your race engineer inside Push & Lift»). It has three sections:

• Háblale (Talk to it): click K to start a conversation, or the microphone to dictate. The same controls are on the Push & Lift bar.
• Qué hace al arrancar (Startup behaviour): opens its own window with PLIA mode selected.
• Tu clave de IA (Your AI key): PLIA uses your chosen provider and saves the key on your computer.

Two buttons below: Conseguir una clave (Get a key) closes the box and opens Ajustes → API (Settings → API), and Empezar (Start) only closes it. It will not appear again.

PLIA cannot answer until you enter a key, start a local model or connect your Claude Code or Codex account. Sending a message shows «Este turno ha fallado» (turn failed), asking for a valid API key in Ajustes → API, and the button Abrir API (Open API).

Option 1: a model on your PC (free, no key)

1. Open Ajustes (Settings) (bottom left) and go to Modelos locales.
2. The top shows your graphics card and memory, for example «Tu gráfica: RTX 3080, 10 GB.»
3. Find your memory group or a smaller one. Compatible models are marked Te cabe (Fits your GPU).
4. Click Descargar y usar (… GB) (Download and use (… GB)) for your chosen model. «Descargando… %» and a progress bar appear.
5. When finished, PLIA automatically selects it in the chat box: «… seleccionado en la caja del chat.»

Details for this tab in 5.6 Local models.

Option 2: your API key

For free use, choose NVIDIA NIM, Google AI Studio (shown as google) or Groq (shown as groq).

1. Open Ajustes and go to API.
2. Choose a provider. At the top are GRATIS (FREE) and below DE PAGO (PAID).
3. Click Conseguir mi clave ↗ (Get my key ↗). The provider website opens: create an account and copy the key.
4. Paste the key into that provider's box («Pega aquí tu clave»). It saves automatically. The dot beside its name turns green and the box reads «Clave guardada · pega otra para cambiarla» (key saved; paste another to replace it).
5. Open the chat's model selector and choose a model from that provider.

The key is stored only on your computer. Details in 5.5 API.

Option 3: your Claude Code or Codex account

1. Install Claude Code or Codex on your PC and sign in. Each card's Cómo instalarlo y entrar (How to install and sign in) link opens its instructions.
2. Open Ajustes → CLI (Settings → CLI) and click Comprobar (Check). When ready, the card says Listo (Ready).
3. The chat model selector shows the groups Cuota Claude (Claude allowance) or Cuota ChatGPT (ChatGPT allowance). Choose one.

This uses your plan allowance without an API key. PLIA never sees your password. Details in 5.4 CLI.

Every screen explained

4.1 The main screen

Its areas are: on the left, the sidebar with your chats; at the top, the header of the open chat; in the centre, the conversation; at the bottom, the message box.

While PLIA thinks, the central logo breaks into particles with phrases such as «Rodando…», «Mirando la telemetría…» or «Hablando con el muro…» (running, checking telemetry, talking to the pit wall). Disable this animation in Settings → Appearance.

4.2 The sidebar

4.3 The chat header

4.4 The conversation

• Instrucciones (Instructions): the first row. Click to expand the mode's startup instructions.
• Inyección de contexto (Context injection): information PLIA adds automatically to your question (for example skill-catalog). Click to expand.
• Your messages appear on the right. Hover to see the time and Copiar (Copy).
• PLIA answers appear on the left. Failed turns show a red Este turno ha fallado (This turn failed) with the reason. A key error also asks for a valid API key in Ajustes → API and adds Abrir API, opening that tab.
• Small marks on the right of the conversation represent turns. Hover for «Ir al turno …» (go to turn); click to jump there.
• Below the message box is the chat's turn and step count, for example «2 turnos 2 pasos».

4.5 The message box

Type your question and press Enter or the orange arrow. You can also type / to see commands and @ to reference a file or session.

Voice notices appear below: «Escuchando…», «Transcribiendo…», «Pensando…», «Leyendo…» (listening, transcribing, thinking, reading) and errors.

Modo de acceso menu

Comandos menu

Seleccionar modelo menu

Models are grouped by provider (DeepSeek, groq, google, anthropic, OpenRouter…). Providers with active keys, installed local models and your accounts for Cuota Claude and Cuota ChatGPT appear first. Providers and models disabled in Ajustes → API (Settings → API) do not appear here.

Context window

4.6 New session

Clicking Nueva sesión shows the Push & Lift logo and «¿Qué hacemos hoy, piloto?» (what shall we do today, driver?). Two buttons sit above the box:

4.7 The Trayectoria tab

Shows PLIA's actions step by step: inputs, model answers and tools used. Helps explain delays or failures.

4.8 The right panel

4.9 The floating voice window

A separate small bar that stays above other windows. Lets you talk to PLIA without keeping its main window in front. Open it from Settings → Voice using Abrir ventana flotante (Open floating window), or automatically at startup by enabling «Abrir la ventana flotante sola» (open floating window automatically).

Move it by dragging any area without buttons.

Settings, option by option

Click Ajustes at the bottom left. Opens a window with left-hand tabs grouped into General; Preferencias (Preferences) (Aspecto and Voz: appearance and voice); Cerebro (Brain) (CLI, API and Modelos locales); Conexiones (Connections) (Paddock Connect and Tu canal); and Avanzado (Advanced) (Plugins and Ajustes de agente).

At the top right, Abrir archivo de configuración (Open configuration file) opens PLIA's settings file on your PC, and X («Cerrar») closes settings. Changes save automatically except where there is a Guardar (Save) or Aplicar (Apply).

5.1 General

5.2 Appearance

5.3 Voice

Browser recognition does not work inside the PLIA window. For dictation or K mode, choose Groq Whisper, con tu clave and enter your Groq key. Groq has a free tier. Without a key, written chat and reading answers still work.

5.4 CLI

Two cards: Claude Code and Codex. Each contains:

With Claude Code ready, select Haiku, Sonnet, Opus or Fable under Cuota Claude. With Codex, select a model under Cuota ChatGPT.

5.5 API

Paste your API key here. Each provider includes a link to obtain one. Some offer a limited free tier; others require credit. Providers appear in two groups: GRATIS and DE PAGO. Each provider has a card:

Free: Google AI Studio and Groq work immediately, and NVIDIA NIM provides free credits for DeepSeek V4 Flash and Pro. DeepSeek costs cents. Anthropic and OpenAI require credit.

5.6 Local models

They run on your own graphics card: free, no key, and conversations stay on your PC. Download once with a button; no extra installation.

The top shows your graphics card and memory. Below, Por memoria de tu gráfica (By graphics-card memory) lists tested models grouped by memory requirements. Choose your memory group or lower. Models that fit are marked Te cabe.

Selecting PLIA 1 Mini or PLIA 1 Max shows an Aviso (Notice): PLIA is experimental and models are being improved; report any failures. Click Continuar (Continue).

Local models handle conversation, sim-racing questions, your notebook, web searches and simple app commands. They make more mistakes than larger models: use an API model or CLI account for serious setup work, coding or videos.

Cualquier otro modelo (Any other model)

For another model, search ollama.com ↗ or Hugging Face ↗, paste its name into the box (example: hf.co/usuario/modelo:Q4_K_M) and click Descargar y usar (Download and use). Only use models from authors you know: the model operates PLIA tools on your PC. It needs approximately its file size plus 1 GB of GPU memory. If it cannot use tools, PLIA warns you and removes it.

5.7 Paddock Connect

Connect services you already use and ask PLIA where you lose time, who to race with or how your championship is going. Each service has a card:

Each card contains:

Local PC services (SIMRacingApps, SimHub) are tested automatically when opening the tab.

5.8 Your channel

Ask it to record, edit and upload: controls OBS, cuts the best moments, adds a title and thumbnail, and uploads to your channel. Uses Directo and Vídeos modes. Cards work like Paddock Connect:

PLIA always uploads privately to YouTube: you decide when to publish.

5.9 Plugins

Configure and inspect installed plugins. This advanced tab contains two subtabs.

Click each section to open it; each has Descartar (Discard) and Guardar.

Lista de plugins (Plugin list) shows each mode's components as Activado (Enabled) or Desactivado (Disabled). Includes a Buscar box, the group Plugins de la sesión (Session plugins) with a selector for the mode to inspect («Elige el preset de agente que quieres inspeccionar»), and the group Plugins globales (Global plugins).

5.10 Agent settings

All modes are here. A preset defines the plugins used by a session's agent: tools, prompt and capabilities.

• INTEGRADO (BUILT-IN): Coder, Modo PTC (PTC mode), Mínimo (Minimal) and Creador de modos (Mode creator). Modo PTC, Mínimo and Creador de modos appear only here, not in the new-session mode menu.
• PERSONALIZADO (CUSTOM): PLIA, Spotter, Setups, Coach, Liveries, Mi PC, Directo, Vídeos and Fábrica.

Typical tasks, step by step

6.1 Ask a question

1. Click Nueva sesión.
2. Check that the mode is PLIA and a model is selected at the bottom right.
3. Type the question in the box («Pregúntame de reglajes, pilotaje o estrategia…»: ask about setups, driving or strategy).
4. Press Enter or the orange arrow.

Continue the conversation in the same chat box. To return to an older chat, click it in the sidebar.

6.2 Send a photo or screenshot

1. Click the paperclip («Adjuntar») and select the image, or drag it into the box.
2. Write what you want to know about it and send.

Images can be PNG, JPG, WebP or GIF. The model must support images; otherwise it asks you to switch: «El modelo actual no admite imágenes; cambia a un modelo que las admita».

6.3 Send a CSV

1. Click the paperclip («Adjuntar») and select the CSV, or drag it into the box.
2. Wait for upload to finish («Subiendo…»).
3. Write what you want it to examine and send.

6.4 Talk to PLIA by voice

First, once only:

1. Open Ajustes → Voz (Settings → Voice).
2. In Reconocer tu voz choose Groq Whisper, con tu clave.
3. Click Conseguir mi clave ↗, create a Groq key and paste it into Clave de Groq.
4. Select your Micrófono, your Altavoz and the Voz que te contesta.

To dictate a question:

1. Click the message-box microphone («Hablar»). «Escuchando…» (listening) appears below.
2. Speak, then click again («Parar»). Text appears in the box.
3. Review and send.

For a hands-free conversation:

1. Click K. It turns blue.
2. Speak. PLIA answers aloud and listens again.
3. To interrupt its speech, say «para», «basta», «cállate», «silencio» or «stop». It continues listening.
4. Click K again to finish.

Click the speaker to hear the latest written answer. Click the crossed-out microphone to pause listening.

6.5 Talk to PLIA from your wheel

1. In Ajustes → Voz, click Abrir ventana flotante.
2. Click the gear in the floating window.
3. In Teclas, click the action button (for example Conversación) and press a key or wheel button.
4. Click Listo.

Keys work even while inside the simulator.

6.6 Change model

1. Click the model name at the bottom right of the box.
2. Click Modelo.
3. Select one from the list.

You can also type /model in the box.

6.7 Change mode

1. Click Nueva sesión.
2. Click the mode button (labelled PLIA) above the box.
3. Choose a mode and type your first message.

To make a mode the default, click it in Ajustes → Ajustes de agente (Settings → Agent settings).

6.8 Connect Garage 61 (or another service)

1. Open Ajustes → Paddock Connect (Settings → Paddock Connect).
2. On the card API de Garage 61, click Dónde se consigue and get your key.
3. Paste it into Clave de API and click outside the box. Status becomes Conectado.
4. If unsure, click Copiar guía para PLIA, paste into chat and PLIA guides you.

6.9 Find, rename or archive a chat

• Search: click the sidebar magnifier and type.
• Rename or archive: hover over the chat, click its actions button and choose Renombrar or Archivar sesión.
• Save chat to file: header three dots → Descargar el registro de la sesión.

6.10 Reduce PLIA's load during a race

In Ajustes → Aspecto (Settings → Appearance), set Efectos animados to OFF. They turn off automatically when iRacing or Le Mans Ultimate is open.

Common problems and solutions