Which Abilities Should I Turn On for My AI Agent?
In part two of my setup video I went through the abilities one at a time. I left all of them on except image generation, for that AI agent.
For the AI agent I built in part two of my setup video, I left all seven abilities on except image generation. That AI agent vets speaking opportunities, so I didn't need it to make a picture. Your own AI agent may call for a different setting, so go through the list with your own job in mind.
If you're looking at a list of switches on the abilities screen and don't know what each one gives the AI agent, this is the list I went through, with what I said about each. It comes from my setup video on YouTube, and one of the abilities, search and fetch, also came up in the live session on October 7, 2026.
What are the seven abilities?
I covered them in this order, with what I said about each. The setting column is what I chose for that one AI agent. Yours may differ.
| Ability | What I said it does | My setting for that AI agent | |---|---|---| | Web search | The AI agent goes out and looks at the web for real-time information. | On | | Web fetch | The AI agent reads and extracts content from a URL, within the parameters you share. | On | | Image generation | It creates images. I said it uses, I think, ChatGPT and Google's image tools up until this point. | Off | | Search past conversations | It reviews other conversations in that one AI agent only, not across different AI agents. | On | | Ask questions | The AI agent can ask you for context when you did not give enough. | On | | Tools discovery | It looks at the work and recommends a connector you have not added yet. | On | | Rules creation | It creates a set of rules for its own tool calls: what to do, which systems to use and in which order. | On |
Why did I leave the rest on?
I said I leave all of these on. Two of them came with a reason. For search past conversations:
From my setup video. I typically keep this one on because I want my agent to be smart.
That ties to self-updates, which I covered in part one: the AI agent can use the context of previous conversations to do things the way I like to see them done. See What Does Edit Own Instructions Mean for an AI Agent?. For ask questions, I said to keep it on and why:
From my setup video. It allows the agent to clarify, so it's not burning or wasting tokens, which is lovely.
For tools discovery:
From my setup video. I typically leave that one on, too.
For rules creation, I said it gets a bit technical and I'd leave the explanation there:
From my setup video. So, it'll create a set of rules on what to do and which systems to use and in which order.
Why did I turn image generation off?
Because that AI agent had no use for a picture:
From my setup video. I leave all of these on. The only one I typically turn off is image generation.
I turned it off because for something like looking for alignment with a speaking engagement, I'd not need an image created. If your AI agent builds slides or social posts, you may reach the opposite answer.
Where do search and fetch show up in the live session?
On October 7, 2026 I talked about search and fetch when I introduced my favorite connectors beyond email and calendar. The first was Exa:
From the recording, 27:30. Exa is similar to Google, but it's built for agents.
From the recording, 27:45. it is built natively for agents to be able to pull the information that humans are looking for.
That's the same job as the search and fetch abilities: the AI agent finds information on the web and reads it for you. For how to choose connectors, see Which Connectors Does an AI Agent Need?.
On my relationship to Gumloop, I said this:
From the recording, 33:22. I don't work for Gumloop. I found this tool over the summer through Lenny's podcast, and I started using it and have since made myself an ambassador for them. I've created a relationship with Gumloop. I am not an employee.
More on that answer: Do you work for Gumloop? (the answer from the session).
What does Gumloop's own documentation list today?
Gumloop's Agents documentation lists Web Search, Web Fetch, Image Generation, Chat Widgets, Search Past Tasks, Ask Question and Rules Creation as abilities, and says most are on by default. Checked on 2026-11-05 against Gumloop's Agents documentation. The labels differ from my video: the documentation says Search Past Tasks where I said search past conversations, it lists Chat Widgets, which I didn't cover, and it doesn't list tools discovery, which I did. Match each switch by what it does, and check the screen you're looking at.
What goes wrong
I didn't name a failure for the abilities. What I did name is the one setting I turn off: image generation, which I turn off when the AI agent has no use for a picture.
The captions don't describe the clicks for turning a switch off, so I'm not describing them either.
Where this comes from
This article grows from the session I taught live with Women Applying AI on October 7, 2026 and from my published setup video on YouTube. The quoted passages are my own words, taken verbatim from the recording and the video. The steps, tables and checklists around them were drafted by my team's AI agents from those words and my published pages, and checked against the recording before publishing. Platform facts that have a public source carry the dated line above, and screens change.
Build Your First AI Agent in One Hour: the full session with the clips, the prompts and the transcript.
Related reading: What Does Edit Own Instructions Mean for an AI Agent?, Which Connectors Does an AI Agent Need?, How Do I Keep an AI Agent's Credit Use Down?.
Frequently asked questions
Which abilities do I turn on?
For the AI agent in my video, all seven except image generation. I leave all of these on, and the only one I typically turn off is image generation.
What does search past conversations cover?
Other conversations in that one AI agent, not across different AI agents.
Why keep ask questions on?
Keep it on: it lets the AI agent clarify, so it isn't burning or wasting tokens.
Does the AI agent make images?
Only if image generation is on. In my video I said it uses, I think, ChatGPT and Google's image tools up until this point.
Build something that actually runs your workflow.
A focused, free 60-minute live session with Jacqueline. You build alongside her, on your own real task, and leave with an AI agent that is already running.