Transcription
Tools are changing fast. We're moving from dashboards and task lists to systems that operate on our behalf. Notion agents are part of that shift.
But most people are either using them for tiny tasks or building agent frameworks that sure do look impressive but don't actually move the business forward. I've built Notion systems for multiple companies over the past few years. And if I was starting today with Notion agents built into the system, I would build things very differently. So in this video, I'll show you how to actually use Notion agents properly and what that looks like in real business operations.
Hey there, I'm Alex Nolles from automationhelpers.com and we help businesses get set up and automated using industry-leading portals, apps, and integrations.
Now, before we dive into today's video, there's something very important that you must understand. Agents do not replace structure. Rather, they amplify it. If you have a messy system, an agent won't fix that. It'll just automate that mess. And if your delivery workflow is unclear, then the agent will just move that confusion around. And that is why the way that you use agents is more important than what they can technically do.
I've built a lot on Notion over the years. And if Notion agents had existed when I built those systems, I would have structured them differently. Not more complex, just more delegated. So, instead of walking you through a list of features, I want to show you how and when to approach Notion agents in a real business operating system. Things like client onboarding, delivery, project workflows, SOPs, and more. So, let's jump in.
First up, we want to answer the question, what are Notion agents? Now, in case you've been living under a rock, they're AI that act inside your Notion workspace, your business system. Not just answering questions, but they can create pages, update databases, complete multi-step workflows, analyze and execute on your behalf, write formulas, reorganize your pages, build databases. Now, technically that sounds impressive, but that isn't what makes Notion agents powerful. What makes them powerful is context. The context of the data that exists in your workspace. And this is where most people miss the opportunity. They see that an AI agent can build a page and they just use it to format docs. But the real opportunity here is in the operational leverage.
Okay, so let's jump in and take a look at how Notion agents can actually help within a real business workspace. So we're going to look at this firstly broadly understanding how to access Notion agents and what they can do. Then we're going to look at the specific use cases.
So we're in a workspace here. Now we can access Notion agent either through the Notion AI on the side panel or that circular icon in the bottom right which will open up this pop-up new AI chat. Now here we can see that we can summarize a page, translate it, analyze for insights, and we're currently working on the client tracker. We can remove this to jump into the broad spectrum of our workspace. However, before we actually jump into the how we want to look at how we can personalize your agent.
So, if you hover over, you'll see that personalize appear. Click this and then we can give a name. Let's call him Andy. And then we can give him a mustache, a rubber ducky friend, a chili, a cowboy hat. I'm going to go with the mustache. And then we'll select setup instructions. We can create our own, use some of the templates available, and we can jump into the larger template base, but we're going to create our own here. And this will open up this page. Let me get rid of this and bring this down. And this is the page where you define how your agent actually responds, acts, and interacts with both you and the data in your workspace. We've got things that we can update like agent identity, chat interaction, memories.
Now, perhaps you are working within a content creation business or maybe you are a marketing team. This is how you're going to best define how your Notion AI agent works. You'll see we've got the tip here. As a Notion user, you can edit this page to customize your agent's behavior, add more context, and even reference other pages in your workspace. Now, I'm sure plenty of you out there are familiar with the ability to create and configure your own ChatGPT. Like here we can see I've got a marketing assistant. It's the same process, the same setup. However, of course, Notion actually has access to the data that you get your work done in as opposed to creating a marketing assistant that is smart, approachable, that acts as a marketing partner that has particular conversation starters. With Notion, we can include the same information like agent identity, how it interacts, and the memories, the preferences, the things that I want it to remember. Create content like this. Approach an issue like that. Analyze like you are the COO. So, it's in here that you want to include the unique way that your agent should respond to your prompts. So, you will want to jump in and edit that so it suits your use case.
Now, we are going to be looking at how to use Notion agents broadly. So, I'm going to leave it as the default settings and we're not going to change anything there other than Andy having a mustache.
So, first let's quickly look at ways that we can query our agent because this is where everything starts with AI. It's how I you understand how to use AI. We prompt it, we query it, it gives us an answer or a response. So, the win here is that agents can search across your entire workspace, pull data from multiple databases, analyze relationships and service answers instantly, saving you from going through trying to find it, filtering your views, sorting what you're looking for. It removes blind spots and instead of manually jumping in, you can just simply ask, for instance, which client projects are behind schedule and above budget. But you can do this with almost any of the work management tools that are out there, ClickUp, Air Table. So there's nothing really special here. We are just asking the platform to bring that knowledge front and center. It already exists in there. We just want it right in front of us. So it's simple and nothing that competitors don't offer. But that doesn't mean that it doesn't help you. Instead, you don't need the perfect dashboard structure or multiple dashboards to bring your answers front and center. You can ask the question and then your agent, your system searches it for you across the entire workspace.
So, one of the more common ways to use agents is for page creation and page editing. Instead of starting from scratch or manually updating information, your agent can work on your behalf, handling content development, SOPs, targeted updates, meeting agendas. But the win here isn't speed. The win is context. So we could prompt it based on the marketing automation setup for Wombat App. One of those clients brainstorm create a kickoff document with goals, roles, timeline and risks. Then we shoot that off. And here we can see that it is crafting, pondering, searching. It's really looking at the context of your database, of your workspace, of your system to give you the best outcome, the best response.
Now, here we can see that it's currently creating pages. And once that has finished, we'll take a look at what we got. So, here we can see we have the kickoff document. So, we'll open this page. And this is currently just sitting in the side here. Here we've got Wombat.app marketing automation kickoff project snapshot primary contact. So it's bringing the data from our client goals, what success looks like, scope, deliverables, these are all things that are currently already within our project tracker. What it's done is it's brought it in front of us so that when we meet with the client or we meet with the team, all that information there. Here we can see implementation team is front and center making life a whole lot easier. Week 1 2 3 4 risks and mitigation open questions next steps confirm tools analyze. So from here we've already got a place to start. We could export that data jump into ChatGPT and get the document similar. However, we're saving ourselves so much more time because the context already lives within the same system as the agent.
Something else that we can think about is policies, procedures, knowledge bases, SOPs. As your team grows, your documentation becomes outdated faster, lightning fast, as your team grows, as your team scales. And unless you hire someone to actually manage your SOPs or manage the document, an operations manager, but they've already got so many tasks, well, your documents are going to fall behind. Plain and simple. So instead of someone manually searching and editing pages or reaching out to team members asking about the processes, we could simply say to like update the vacation policy with two extra personal days, a volunteer day and a birthday day. Agent's going to jump in, the agent's going to make those updates, and you have saved yourself time or one of your team members so that they can focus on the more important tasks at hand.
Now, of course, another thing to look at is that if you've got a meeting coming up, for instance, I have a meeting with Platypus Pool Services. So, I could prompt my AI agent, hey, I need you to create a meeting agenda for our weekly with Platypus Pool Services. It's going to jump into the project, the client, the tasks, understand where things are at, what the next steps are, and bring that into a clean meeting agenda. So, you save yourself time and you get prepared for your client meeting. The agent takes the context from the work getting done and prepares a meeting agenda so that you can focus on the work.
But creating and editing pages is pretty straightforward and it's something that other work management tools and tools like Gemini already offer. So let's look at other ways that we can use Notion agents.
Now within those pages you have information, meeting agendas, notes, brainstorming. So now comes the action, the jobs to get done, the work to do. That information now needs to become structured work. So instead of manually reading through everything and creating tasks one by one, we can again lean on Notion agents and ask it to extract the action items from this meeting agenda and add them to the correct client project. Then the agent will then ponder, start creating and the agent will move forward to generate create those tasks items that you and your team can then get done. Your projects get updated, your properties get filled and links get added and team members are assigned roles. You now have the time to focus on the work at hand and what moves the needle. From here, you're moving from loose information to structured execution within the same system. And that's where things become operational. But of course, if you want, before AI actually applies those changes, you can review and approve what the agent wants to get done. So you don't need to stress about your data structure, projects, or your tasks turning into chaos. There is a massive win here, and that is the removal of delay. Because there's a big difference between we talked about it or it's tracked properly. Agents remove that difference and reduce the delay.
So far, everything that we've taken a look at happens inside Notion. So, let's step out because Notion agents can also search the web, enrich your data, analyze PDFs, and pull in context from your connected tools.
So, let's say for instance, we're about to have a meeting with a client and we want to understand the issues of their specific industry. We could jump into a doc and prompt AI to search the web and prepare a proposal document that looks at everything issues, solutions, and competitors for their industry. What then is going to happen? The AI agent is going to start searching the web and finding similar competitors in their industry, looking at the issues they face and how solutions can be found. Before any sales call, you can ask the agent to research a company and summarize their business model, recent updates, and likely operational challenges. So, instead of opening multiple tabs and copying notes back into your CRM, the research and the update happens in one place. So, if a new lead comes in, you can have Notion Agent take a look online and generate a proposal document. But not only that, you could even have it update your client tracker or your CRM with that data it found.
Another way that we can use Notion agent is to analyze PDFs that we received from our clients or contacts. Let's say for instance we've received a project scope. Now this could also work for terms maybe report on the issues that a client is facing. We can upload that to Notion agent and then ask it to analyze but we can take it further and create tasks within the task tracker. Now, this wouldn't be the best process to follow, but we're going to take a look at what is really possible with Notion agents, not only within the context it has, but also things that we provide for it. Now, we could take it further and also bring in context from our other pages. Perhaps we know there is a similar client we've been working with that has similar issues or a similar project. We could reference that in our prompt, diving deeper and giving more context to the agent.
So here we can see it's gone ahead and it's turned that scope doc into trackable items in our tasks tracker. Let's open up this page Apex construction group. Let's open up what we've got here. The parent project discovery workshops reporting system architecture subtasks have also been created with that task. All of that just from the project PDF that we uploaded. That's a massive win here. You can see that not only are you working with the data inside your system, but you're giving it context from external sources.
Now, a massive win of Notion agents is that we can enrich the data that already exists in our workspace. Mammoth for sales teams, for services. So, we've got a CRM here and we want to jump in and let's just ask it to enrich our CRM data with websites, with other information. We're going to keep it simple to see what we get, but of course you can really drill down. Now, I will mention that if you do have a CRM or another database that includes a website URL, the results that you're going to get far exceed what you can by just simply having a company name. So, here we've got enrich our CRM data. Done. I added enrichment fields to your CRM. So, it's added it, but I don't see it. Let's scroll across here. We can see we've got website URL, LinkedIn company, industry. So, we can see that nothing's actually been brought through. I need you to search the web and update those fields as best you can. Let's see what we get back here. We can see we've got headquarter location, founded year, of course, the industry. It's searching for that data. It's organizing. Now, this is a feature that we see the likes of ClickUp and Air Table do really well through their field agent features. However, let's see how well Notion performs by just having the company name. We can see that it's found 99 results. It's organizing. And just like that, here we can see we've got notion.com. We've got the LinkedIn profile, software development, the headquarters, founded year. We are Trello, Reich. The only ones that we're missing that it wasn't able to find more information on was Basecamp, Freshworks, HubSpot, Pipe Drive. So here we can see a couple fields may still be blank where I could not confirm a single claim value. For example, KOD headquarters location updated. So if you have leads coming in, not only can you set up your onboarding process through an agent, you can actually enrich that data. So then the agent gets more context on the client. Big win here.
So far we have taken a look at querying, transforming info, editing, and creating pages as well as using external context. Let's look at how these can all come together. And they're combined through multi-step tasks. So rather than asking the agent for an output, let's ask for an outcome, a workflow.
Now, we're just going to open the agent. And let's say for instance that you have a heads of department meeting you want to run every week. You need to collect the right information from your business. So we're going to ask the agent to create a reporting page, which we'll use as a hub here in navigation. Then within it, we're going to ask it to create a report on the inner workings of our business, a meeting agenda for the heads of department, and in that meeting agenda, a database that will help us track the goals from which we have found from the report. So, we're going to shoot that off. And what the Notion agent will then do, we'll take a look at our projects, our tasks, our clients, our resources, find out the inner workings of our business, create a report from that and place it in the reporting hub and then create a meeting agenda which will then meet with the heads of department.
Okay. So what the Notion agent has gone ahead and done, it has created a ops report, weekly operations report, view of client tracker, heads of department, view of quarterly goals. So let's see where these changes have been applied. So, heads of department weekly meeting agenda. Here we can see we've got the report as well. If we take a look at that and it's jumped in. Let's take a look. It's actually visualized our tracker blockers decisions needed. Platypus Pool Services current status. It hasn't pulled that much information through, but it has done a great job of readying us with a heads of department meeting agenda. Now, I hope you understand that for this video, I've just used an example workspace. It isn't truly set up so a lot of information won't be pulled from it.
Somewhere where a multi-work step or flow would be a massive win is if you are a solo operator or small team that's onboarding clients. You know that there are always the same tasks or processes that you need to follow. You need to set up an initiation call. You need to get assets and information from the client. You need to prepare their projects. Set up their tasks. Onboarding phase 1 2 3 4. So rather than manually jumping in and doing that, we can lean on the agent and we can see here that we've got a lead Pawsome AI. Let's prompt it simply follow the onboarding process for Pawsome AI and generate the projects and tasks. Now you could have it so that when a particular client or lead has the status onboarding, then you prompt all of those currently in that onboarding status to get the same. Beyond this, you could also have an onboarding project template that Notion's agent could then pull information from so that you know that you are systemizing that onboarding process and not just simply asking it to follow the onboarding process for Pawsome AI and generate the projects, tasks cuz I don't have anything pre-existing. It's just going to look at what we've done for these other particular clients.
So here we can see we've actually got a Pawsome AI onboarding project and from there we can see that it's included tasks directly in our task tracker intake scope and plan access kickoff. So if you systemize and set up processes that will need to be followed each time a client is onboarded, you can have the agent do that for you and still retain your structure. And when onboarding starts cleanly, delivery starts cleanly. So that's a win-win.
Okay, so everything that we talked about so far works because your system is built on structure and that structure exists within databases. Notion agents can create databases, update properties, filter your views, and bulk update your records. That matters because businesses rarely look at shifting their structure. Restructuring is manual, migrating data is tedious, and updating properties across dozens or hundreds of records is risky, time-consuming, and frustrating. So business systems stay the same even though the businesses themselves change. But we now have Notion agents. So what are you waiting for?
We can ask Notion agents to create databases for us here. For instance, I'm asking you to create a simple CRM. What that will do, we'll take a look and we can see we've got the lead. We've got the dual value field, next action, next due date, source, and status. It is creating that. Perhaps you're not happy with the fields that are there. You need extra fields. You want to link this to other databases, your your contacts, your deals, your projects. Well, easy. Rather than jumping in and manually doing it, we could say, I need a relation field connected to our task tracker. Just to put it simply, what that's going to do is then look at the structure of your databases that exist, find a referenced database, our task tracker, and create a linked or relational field here or property. But of course, you want to ensure that you include as much context as possible so that your Notion agent can do the right job for you. With that linked property, we can then jump into our tasks and pull in what is related. Creating databases is useful. Editing them is where things get interesting.
Now, if you're just getting started with Notion or you're moving some of your data over to Notion, you can simply add a CSV file or PDF, which we looked at earlier, to then build your database from. So, build database and we'll shoot that off.
Another great way that we can use Notion agents is to calculate all ride formulas for us. So, you'll notice here that we've got a CRM with some familiar faces and we want to add a new field. So we get more clarity on what is going on. Going to open the agent. We'll start a new chat and we're going to just quickly paste with the magic of video editing. Create a new formula property called weighted deal value in this database. The formula should calculate potential deal value probability requirements. We've included more context to ensure the output or the outcome is exactly what we need. So we'll shoot that off and it's going to generate a new field, a new property here without us having to write the formula or selecting to create it ourself. Here we can see weighted deal value. It's starting to work. Let's minimize this. And here we go. Here's the weighted value of each of those clients. I didn't have to jump in. I didn't have to create a new formula field. I didn't have to spend that mental energy trying to ensure that the calculation that the formula is written properly itself. I can ask the agent to do that for me and then focus on the other tasks that I have at hand while it is getting that done.
Now, this could also help you. Let's say you've been running your Notion workspace for your business for a year and you know things are messy. They're chaotic. Well, rather than going through and analyzing doing a review of your workspace, you could lean on Notion agent to a first analyze and take a look, review and suggest improvements, and then implement those improvements on your behalf.
But what if you were just getting start with Notion? If I were to open a new Notion workspace in 2026, here's what I'd do. I would use agents not to build random pages, but to help me first build my structure. I'd start with core databases like clients, projects, tasks, and content. Set up clear relations between them and keep properties minimal and focused on decision-making. Then I would review what Notion creates, adjust it, and refine. Agents can build the structure, but you still define what matters. Next, I'd look at repetition. What things are me and my team handling again and again? Onboarding, meeting agendas, weekly reporting. Then I would use Notion agents to improve that system.
Using agents properly isn't about doing more. It's about deciding what not to do manually anymore. If something requires judgment, creativity, or strategy, you look after that. If it follows a clear pattern or process, you delegate that to an agent. The shift here is that we're moving away from managing tasks and moving towards managing systems that execute tasks on our behalf.
If you want to learn more or you're eager to set up your business for success, then don't hesitate to reach out to us at automationhelpers.com where a team of experts are offering a free 30-minute consultation. So, book yours today.