Agent Skills/Features & Tools
Not all Agents are equal!
1. Agent Skills
All Sapience Agents are powerful, but they’ve all got different skills and features.
When you’re in the main chat interface, in the chat header you’ll something like this - click on any of these to see the skills/tools that a given Agent has.
Below is a screenshot of a particular agent, and you can read more below about all of the Agent features & skills currently in the app.
2. Details on Agent Skills & Features
2.0 All Tools Currently
Tool / Feature | Description |
Web Search | The Agent you’re using, will fire up the Web Research Agents (small Team-of-Agents) to run multiple queries in answering your question. |
Deep Research | Whether it can perform Deep Research. These are long jobs that take 10-30 minutes, but give amazing content and analysis when done. |
Coding Agent | Whether the Agent can talk to and use the Sapience Coding Agent to analyze data, do numerical analysis, generate visuals (static and interactive) and even build small web apps. |
Sapience Files | Actually three related tools: list files, search files, retrieve files. Basically gives the Agent access to your Sapience File System. |
File Upload | Whether or not you can attach a file to a Conversation with this Agent in Chat. Nearly all Agents have this feature. |
Agent Memory | Actually a cluster of tools, including: 1) search Memories; 2) list Memories; 3) edit Memories; 4) search prior Conversations. Part of our 4-Tier Memory architecture. |
Visual Mode | A specialized mode that uses the Sapience Coding Agent in the background to always produce static or interactive visuals. This will usually be on data agents, where the Agent creator has baked tabular data into the Agent’s core knowledge. |
AI Vision | Whether or not the Agent can read/analyze/understand a drawing, image, picture, whiteboard. |
Generate Image | Whether or not the Agent can create an image/visual. This is like having your own pro artist. |
Create Powerpoint | Whether or not the Agent can create/save a PowerPoint file. |
Scheduled Jobs | Whether or not the Agent can read/edit/create Scheduled Jobs on your behalf. This mainly applies to the Sapience Agent (the built-in default Agent of the Sapience Agent Operating System). |
2.1 Web Search:
This is what it sounds like! This is a tool available to the Agent to do web research for you. Note that this is really Agent X, using the Web Research Agent to do work for it. So you’re creating a small team-of-agents. Make sure you read the article about built-in Agents here:
2.2 Deep Research:
Agents that support Deep Research are capable of writing long, detailed, information rich reports for you. These reports can run to 20-30 pages. When you select Research mode in the chat interface, the Agent will ask you a series of clarifying questions. Its worth answering these properly, as it will steer the agent in its research run.
Some things to keep in mind:
- not all Agents can do Deep Research. Check the Agent Features card.
- A Deep Research run can take 15 minutes or more. But its worth the wait. The Agent will often make dozens of search queries to the web as well as to data that the Agent has access to. You can watch it work.
Pro-tip: have a regular conversation with the Agent in Default or QNA mode before you kick off the Research run. This is called “priming” the conversation. You will notice that if you have a conversation before doing Research, the clarifying questions it asks you will be far more on-point than if you run Research immediately in a chat.
2.3 Coding Agent:
If an Agent advertises this skill it means it can spin up a container (a small secure server in a sandbox) to write code and analyze data with. Sapience’s code-writing capabilities are very cool, and this is worth trying out on a data set that you want to analyze or visualize.
2.4 Search / Read Sapience Files:
Agents with this Feature can work with the Sapience object graph, so they can list all your files, perform full-text search across all your files, and generally leverage whatever documents and data you have uploaded to Sapience Files.
Pro-tip: if you know the file you want it to work with, tell it explicitly. Whilst it can search all your files, once you have hundreds or thousands of files, this will be slower than if you are explicit.
2.4 File Upload:
Whether you can attach a file (doc, image or tabular data) to a conversation. Most Agents will support this. See the help article about conversations for more info.
2.5 Agent Memory:
Sapience has a four-tier memory system. There’s a separate article on that linked below, but in general, if you tell Sapience to “remember this” then it can save facts and preferences (separately) in a global memory space that is accessible to all Gen3 Agents. You can see what memories have been stored (and delete outdated ones) in Menu > Settings & Customization > Advanced > View & Manage Agent Memories. See the dedicated article on Memory here:
2.6 Visual Mode:
Some Agents are turned to do data analysis out of the box. If an Agent has the Visual Feature advertised it will also have a Visual option in the Output Mode drop-down in Chat. You can learn more under Output Modes here:
2.7 AI Vision
These Agents can read images, whether its a mockup or a photo of a whiteboard. Read the article about Conversations to learn more:
2.8 Create Image / Visual
This is different to, but related to Visual mode. Visual Mode is used when an Agent has been built with tabular data in it (Excel, CSV, etc) or where you have added tabular data to a Conversation. The tool we are describing here is different, and is known as generate_image under the hood. The Agents that have this tool can generate any sort of image that you describe, with or without tabular data to inform it.
More info & screenshots:
Here, you can see that we ask the Sapience Agent, to “Generate a diagram showing that a Conversation can be made up of N Users, and Y Agents, and is built from Z Messages”. It thinks for a while, then decides it needs to create an entity relationship diagram (exactly what we wanted, but we didn’t have to specify that!).
Then you can see at the bottom of the chat, its produced a great diagram, depicting exactly what we asked.
Here’s the picture by itself (and remember that this is now always available for you in File System > Outputs:
Generate Image is NOT the same as Visual Mode. Let’s look at an example.
One thing that users get confused about is the difference between the generate image tool/feature, and Visual Mode. The easiest way of thinking about it is this: Visual Mode uses the Sapience Coding Agent to analyze data, and give you visualizations over that data. Its like a human with data-science and coding skills. The generate image tool on the other hand is like having a skilled artist at your disposal. It can/will create good business graphics, but you can also do this!
This next screenshot shows the prompt we used for this, click it to enlarge if its small on your screen:
2.9 Manage Scheduled Jobs
Agents with this tool can read/edit the Scheduled Jobs that you have setup inside Sapience. There is a whole separate article on that here: