32 new integrations: Microsoft, Google Chat, Canva & more

Hey there, I'm Parker Chun, product manager here at Tasklet. Today we're

announcing 30 more native integrations, starting with Microsoft Suite, more Google integrations, Canva, and a ton others. Additionally, we've made big

others. Additionally, we've made big improvements to our popular integrations including Gmail, Notion, few others, and our direct API capability.

So, I'm going to walk through a few use cases to show some of the changes we made. All right. So for our first use

made. All right. So for our first use case, it's a check Outlook for emails with attachments of sales reports. Want

to download them, analyze the revenue, flag anything important, and then save this to uh let's do SharePoint.

SharePoint and then post a summary to Teams. Oh, so the agents off going and running.

Uh what's great about Tasklit is you don't need to preconfigure any uh integrations or connections. You just

tell it what you want. The agent will figure it out. It'll check what connections you've set up in past.

Go get what it needs.

And here it all is. So permissions are toolbased. So if you want to give your

toolbased. So if you want to give your agent ability to uh d say draft emails but not send emails, you can totally do that with tool level permissions. So you

can see outlook here. Uh we're only allowing searching, getting message, and downloading attachment. No, uh ability

downloading attachment. No, uh ability for us to send messages. Uh and the agent figures out what it needs and asks you for approval.

So now you can see we have our uh Outlook Microsoft tools uh and connections right here.

Well, the agent is going off and searching for emails. It made a mistake.

But what's great about these agents is they recover and figure out how to move on. So it seems like it's found the

on. So it seems like it's found the emails with attachments, three sales reports, right? Uh it's downloading its

right? Uh it's downloading its attachments. The agent has a each agent

attachments. The agent has a each agent has a file system which it can use to download files across integrations.

uh store your files that you upload via the chat um and do all kinds of manipulation uh in the file system uh as well as its sandbox

uh which the agent is doing now. So

we've downloaded the CSV files and what the agent is doing we can click in here is writing code to go and parse and process these CSVs.

So the agent has a full sandbox uh which you can use to write code, run scripts, um do all kinds of nice things.

Okay, so the analysis completed. We

found one store has seen a significant drop.

And now the agent's going to look for SharePoint.

Uh searching my sites. Uh I only have one site here, so we're going to use the shortwave site.

Now we are finding the right team channel and fails found the sales channel

and cool. So what the agent did was it

and cool. So what the agent did was it downloaded the CSVs for my email uh wrote some code to process them do some analysis combined that into an Excel

sheet and has uploaded that to SharePoint here. And now it's sending

SharePoint here. And now it's sending the message to teams. And cool, I get the results right here.

What's powerful about Task is we also have triggers which mean that the agent can run on its own. And I can actually have this whole thing be done for me

every week without having to come back here. uh run this every

here. uh run this every two Monday at 8 a.m.

So tasklet has a few different triggers.

It supports uh schedules, web hooks, uh RSS. So RSS is whenever a new item is

RSS. So RSS is whenever a new item is added to a feed, and you can use that for um podcasts, uh news, YouTube, Substack.

Um, right now we're just going to use a simple schedule trigger uh to run every week Monday at 8 a.m. Um, this is just going to run indefinitely, but you could

set it up to expire, have this run just for a month, or have it just run once.

And what the agent is doing here, it's going and creating a sub agent, which is a uh kind of containerized agent with specific instructions to kind of re

recreate the process that we just went through. And yeah, now this will just

through. And yeah, now this will just run for me every week and I'll get insights uh stored in SharePoint as well as in our team's chat. All right, this

next one is a is a real one for me. Uh

we're use BigQuery to find the top integrations from the past week.

I've attached a file here with a little bit of context on our BigQuery tables, but I'm mostly going to just let the agent figure it out.

Agent is going and figuring out that I have a existing BigQuery connection and I'm going to go and approve.

So the agent is going and looking at the events to kind of understand the structure before writing the actual query.

And here we go. Now we got some stats and some observations, some some things to take a look into.

So, I'm gonna I want to see how this compares to last week.

Okay. So now I'm gonna ask for Google slide deck and agent's going to go once again check what connections I have. Uh

I believe I've connected to Google Slides and another agent.

And here we go. Let's do this.

Once again, we're going to have this run automatically because I don't want to come in here and have to ask for this every time. Yes.

every time. Yes.

Let's do Friday at noon.

All right. Let's give Google chat permissions.

And here we go.

And cool. So once again the agent is kind of taking this whole process that we've went to writing this down into a set of instructions which it will use uh

every Friday to repeat this process for me.

And this one here can see the instructions it wrote uh and I can take a look over this to uh give it any feedback and tell it to correct. Uh now

we see we have our trigger here and this is going to run every Friday at noon.

All right, thanks Tlet.

Now I'm going to show you uh how to connect to X, formerly known as Twitter via direct API.

So as you all probably know, uh Axe recently made some changes to their API.

uh and pricing is now pay use and uh I think it's really fun to use in Tesla.

So when you ask an agent to connect to something, it'll first look for the native integrations that exist and if it can't find anything, it will go and

research that API and learn how to connect to that service. So with this capability, tasklet can really connect to any service and uh in the event that

a service doesn't have an API, tasklet also has computer use uh which it can use to navigate a computer.

But what the agent is doing right now, it's going and researching the X API.

It's finding the relevant endpoints and uh it's going to figure out the scopes, figure out all of this stuff that uh I

don't even need to think about.

Um and it's going to give me additional instructions on things I need to do.

Okay, so now I have this form. Uh let's

hop over to X developer platform.

And then let's go to apps. And what is taska telling me to do? All right. Well,

first I'll explain uh x has two ways to connect. You can either use a bear token

connect. You can either use a bear token which uh only allows read access, but we're going to use oath 2 because this

allows you to do a lot more. And

what we're going to first need to do is see we're need to go into X and create an app.

Let's just call this task demo.

And here going to hit close.

And here we're going to go view our app for ooth 2 in X. We are going to need to click set up.

And I want read and write. So I

configure this. Uh for X, we should do this. Um some of these things I know uh

this. Um some of these things I know uh because I've asked task and pass. So, if

you're if you don't know what to do when setting up an API connection, they just screenshot it and paste it back in. Um,

so tasklet saying that I need to copy this call back into here for website.

Just put tasklet and it's literally just that. So,

copy client ID, copy client secret. Uh,

don't share these with people.

Uh, task that's filled these things out for me. Read, write, users read offline

for me. Read, write, users read offline access.

Okay. Um,

let's do enable these two.

And remember how task went and researched uh everything earlier? Uh here's all the

everything earlier? Uh here's all the things it has uh written down um with a link to official documentation so I can look things up in

the future. This is editable too. So I

the future. This is editable too. So I

might want to put something in here like the X API is expensive. Make sure to not uh use my credits a lot. Um but I'm going to leave this blank. Uh this is a

useful thing for you to give more instructions. And let's uh connect.

instructions. And let's uh connect.

And okay, I'll authorize and what Tasklet's doing is going and calling the API here. And

here. And Cool.

Some cool things. Um, this one from Jake Lindsay. Let's uh repost that.

Lindsay. Let's uh repost that.

All right. Uh, and I think the agent ran into an issue the first time, but we succeeded in reposting. So yeah, from

here we could totally set this up to run automated. Um, the system I I find

automated. Um, the system I I find useful is to have an agent go and and find uh posts and topics that are interesting and then email them to me

with potential suggestions and wait for my approval. And with Tasklet, you can

my approval. And with Tasklet, you can just respond to emails it sends you uh and those will get sent as messages to the agent. So you could totally set up

the agent. So you could totally set up like a system like that. But I just want to show you how to set up a direct API connection um with uh the new the new

capabilities we've built.

And these use cases just scratch the surface on what Tasklet is capable of.

So check it out, enjoy the new integrations, and let me know if you have any feedback.