Shinydocs Pro
Shinydocs Pro
Breadcrumbs

Shinydocs AI Interface Quick Start

This guide takes you from a fresh install to asking questions about your documents. It uses Ollama on the same Windows server as Shinydocs Pro. To run Ollama on a separate server, see the Ollama Quick-Start Guide.

1. Install Ollama

Install the latest drivers for your graphics card (GPU) first. Models run much faster on a GPU.

  1. Download and install Ollama from ollama.com/download.

  2. Open http://localhost:11434 in a browser. You should see Ollama is running.

2. Pull some models

You need a model that writes text and a model that makes embeddings. In a command prompt, run:

ollama pull nomic-embed-text:v1.5
ollama pull granite4.2:8b

Choose a text model that fits your GPU memory. See Sizing models based on your resources in the Ollama Quick-Start Guide, or browse ollama.com/search.

3. Install Shinydocs Pro and add a source

  1. Install Shinydocs Pro and upload your license. AI features need the Intelligence Engine add-on on your license. Trial licenses don't include it. See the Shinydocs Pro Control Center Guide.

  2. Add a source. For a first try, choose a small source (under 20 GB) and wait until its crawl and Reading file contents step finish.

4. Turn on the AI features

Open the feature flags page at https://<server>:9701/settings/flags and turn on:

  • Semantic search and AI Overview: search by meaning, the AI answer in Shinydocs Search, and the Generate embeddings tool.

  • AI Analysis Tool: only if you want to run your own prompts over documents. It shows the AI tab under Settings > Analysis and the AI analysis step on each source (see AI Enrichment).

5. Add Ollama as a provider

  1. Go to Settings > Intelligence and open the Providers tab.

  2. Under Add a provider, select Ollama.

  3. Enter a Display name. Leave Server URL empty for Ollama on this server, or enter its address, such as http://ai-server:11434.

  4. Select Save. The provider shows Reachable and lists the models it has.

cc-ai-08-form-ollama-26-5-light.png
cc-ai-02-providers-26-5-light.png

6. Choose a model for each task

  1. Open the Tasks tab.

  2. Select Change next to each job and pick a model:

    • Writes answers and Writes the AI Overview: your text model.

    • Finds passages and Turns content into embeddings: your embedding model.

    Models that can't do a job are shown but can't be picked.

  3. Select Save. A task shows Working when every job has a model that is switched on and its provider is reachable.

cc-ai-04-change-model-26-5-light.png
cc-ai-01-tasks-26-5-light.png

7. Embed your source

  1. Go to Sources and open your source.

  2. In Analysis, turn on Semantic search.

  3. Select Run now. When the step shows Up to date, the source is ready. The Tasks tab shows how many sources are ready (for example, 3 of 6 sources ready).

cc-ai-10-source-26-5-light.png

If you change the embedding model later, each source is embedded again the next time it runs. Until then, its results come from keyword matching only.

  1. Open Shinydocs Search from the search icon at the bottom left of the Control Center.

  2. Search for something in your own words. The AI answer appears above the results, with numbered citations to its sources.

  3. Open a result. In the AI assistant panel, type a question about the document and send it.

cc-ai-12-overview-26-5-light.png
cc-ai-13-assistant-26-5-light.png

AI responses may include mistakes. Check the cited documents before you rely on an answer.

Helpful tips

Ollama commands

  • ollama list: shows the models you have pulled.

  • ollama ps: shows the models loaded right now, and whether they are on the GPU or CPU.