Cotera is the fastest way to build an AI agent.
Just type out instructions and background info, give your agent access to tools, and let it do the tasks that are mundane, complex, or distractions so that you and your team can focus on the highest leverage and most strategic work.
Busywork is boring. Outsource that to AI agents, and focus on the work that matters most.
We work with some of the best and most innovative teams in the game:
- Bilt Rewards
- Coterie
- Peloton
- Skims
- OpenSea
- and many more
Cotera has grown a ton in the last few months, and I'm looking to make a bunch of hires across Engineering, Operations, and Growth. You should be one of them :)
Note: A lot of these are for new grads, so don't shy away if you graduated this year! All of them require being 4 days a week in person in NYC or London.
AI agents are great at exploration, but they’re not always the right primitive for repeatable business processes.
This article is about the gap between agents and workflows: why agents feel magical for open-ended tasks, why workflows are better for trust and reliability, and why the best systems probably combine both.
I also dig into how workflow structure can reduce LLM cost by limiting what gets passed into agentic steps — less context bloat, fewer surprises, more reproducible outcomes.
Agents explore. Workflows deliver. The future probably needs both.
I've got 13 AI agents running my entire agency right now.
I used to spend hours wiring up workflows in traditional automation platforms. Connecting nodes, fixing broken triggers, and every single time something broke, I'd fix it manually... which would then break 20 other things in the workflow. And then I'd fix those, and three more would break. It was a full-time job just keeping the automation alive.
Now I just open Cotera, type what I want in plain English, and the agent goes and does it.
I don't build flowcharts anymore. I just type something like "monitor our product Slack for customer complaints and create Linear tickets sorted by urgency," and five minutes later that agent is running in the background while I focus on something else.
Here's what I actually love about it:
1️⃣ 𝗬𝗼𝘂 𝗱𝗲𝘀𝗰𝗿𝗶𝗯𝗲, 𝗶𝘁 𝗯𝘂𝗶𝗹𝗱𝘀 Setting up an agent feels like writing a message to a coworker. You tell it the goal, the tools it has access to, and how you want the output. That's the whole setup.
2️⃣ 𝗜𝘁 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝘁𝗵𝗶𝗻𝗸𝘀 Most automation tools move data from point A to point B. Cotera agents read the data, analyze it, make decisions, then take action. My deal health agent doesn't just pull deals from HubSpot. It looks at last activity, compares update history, figures out which ones are actually at risk, and writes me a summary.
3️⃣ 𝟱𝟬𝟬+ 𝘁𝗼𝗼𝗹𝘀, 𝘇𝗲𝗿𝗼 𝗔𝗣𝗜 𝗵𝗲𝗮𝗱𝗮𝗰𝗵𝗲𝘀 HubSpot, Slack, Gmail, Apollo, Notion, Stripe. They've got 100+ providers and 500+ tools built in. I've never had to touch an API doc.
4️⃣ 𝗔𝗴𝗲𝗻𝘁𝘀 𝗿𝘂𝗻 𝗶𝗻 𝘁𝗵𝗲 𝗯𝗮𝗰𝗸𝗴𝗿𝗼𝘂𝗻𝗱 I don't sit and watch them. They just work. While I'm building something new, my agents are handling customer complaints, flagging Stripe transactions, turning video transcripts into LinkedIn posts.
The whole experience genuinely feels different. With traditional automation, I'm mapping every single step. With Cotera, I set the direction and the agent figures out how to get there.
That's what actually changed how I run my business.
Try Cotera free (with bonus credits): https://lnkd.in/dE44r43a
Over to you: What's one task in your business you'd hand off to an agent first?
I've got 13 AI agents running my entire agency right now.
I used to spend hours wiring up workflows in traditional automation platforms. Connecting nodes, fixing broken triggers, and every single time something broke, I'd fix it manually... which would then break 20 other things in the workflow. And then I'd fix those, and three more would break. It was a full-time job just keeping the automation alive.
Now I just open Cotera, type what I want in plain English, and the agent goes and does it.
I don't build flowcharts anymore. I just type something like "monitor our product Slack for customer complaints and create Linear tickets sorted by urgency," and five minutes later that agent is running in the background while I focus on something else.
Here's what I actually love about it:
1️⃣ 𝗬𝗼𝘂 𝗱𝗲𝘀𝗰𝗿𝗶𝗯𝗲, 𝗶𝘁 𝗯𝘂𝗶𝗹𝗱𝘀 Setting up an agent feels like writing a message to a coworker. You tell it the goal, the tools it has access to, and how you want the output. That's the whole setup.
2️⃣ 𝗜𝘁 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝘁𝗵𝗶𝗻𝗸𝘀 Most automation tools move data from point A to point B. Cotera agents read the data, analyze it, make decisions, then take action. My deal health agent doesn't just pull deals from HubSpot. It looks at last activity, compares update history, figures out which ones are actually at risk, and writes me a summary.
3️⃣ 𝟱𝟬𝟬+ 𝘁𝗼𝗼𝗹𝘀, 𝘇𝗲𝗿𝗼 𝗔𝗣𝗜 𝗵𝗲𝗮𝗱𝗮𝗰𝗵𝗲𝘀 HubSpot, Slack, Gmail, Apollo, Notion, Stripe. They've got 100+ providers and 500+ tools built in. I've never had to touch an API doc.
4️⃣ 𝗔𝗴𝗲𝗻𝘁𝘀 𝗿𝘂𝗻 𝗶𝗻 𝘁𝗵𝗲 𝗯𝗮𝗰𝗸𝗴𝗿𝗼𝘂𝗻𝗱 I don't sit and watch them. They just work. While I'm building something new, my agents are handling customer complaints, flagging Stripe transactions, turning video transcripts into LinkedIn posts.
The whole experience genuinely feels different. With traditional automation, I'm mapping every single step. With Cotera, I set the direction and the agent figures out how to get there.
That's what actually changed how I run my business.
Try Cotera free (with bonus credits): https://lnkd.in/dE44r43a
Over to you: What's one task in your business you'd hand off to an agent first?
Do you want to learn how to give AI agents long term memories? Come to our webinar with Hyperspell (YC F25).
We're going to be using Hyperspell and Cotera to make an internal assistant that has memory - basically, it learns and remembers over time, so it gets better with age.
This lets an agent:
1. Better understand your ICP (Sales)
2. Understand how you talk to customers (Support)
3. Knows your brand voice (Marketing)
4. Understands code patterns (engineering)
Come one, come all.
https://lnkd.in/enQafw8i
Ibby Syed and I are going live next Tuesday to show how to build AI agents that remember all the context of your work.
We'll walk through a before-and-after: an agent that answers questions from a company's docs and community and gets smarter over time instead of starting from scratch every session.
If you're in CX, sales ops, or marketing ops, this one's for you.
📅 Tuesday, March 17 | 2:30pm ET 🔴 Live on LinkedIn
Drop a comment or register to get notified when we go live.
One of the top problems with AI right now: people were never taught how to use it effectively.
That starts with writing a good prompt.
If you're:
>> intimidated by the idea of using AI
>> not sure how to write a good prompt
>> not sure what makes a prompt "good"
>> hoping to improve the outputs from AI tools
Then we have the solution for you!
We made a prompting course to teach you how to write great prompts. It is fun, it is incredibly well done (s/o to Charis Y.), and there is absolutely no catch.
- it is free
- there is no signup required
- it isn't specific to Cotera, it is for prompting all LLMs
Check it out! https://cotera.co/learn
What am I thinking about when it comes to AI?
- How do we automatically categorize support tickets so we can route them to the right system or next step in Zendesk
- Can we use AI to generate data pipelines and transformations for clients? If so, how?
- Can I use call transcript data as a source of ideas for content?
- How do we best help clients to build AI agents and workflows (anything from summarizing weekly updates to send to their manager, to answering questions about people in a CRM)
But:
- I'm not an AI researcher
- I don't fine-tune models
- I don't have an ML background
I'm just a data person trying to create business value with AI.
__
And Ibby Syed is one of the smartest people I know when it comes to bridging the gap between data infrastructure and AI automation.
So we decided to grab some time on Monday to talk about what we're both seeing as similarities and differences between working with data pipelines and developing AI agents and workflows.
Want to learn about what it takes to make the shift from writing SQL pipelines to building AI agents or automating business workflows?
Want to hear if it's a complete shift from data pipelines to AI? Or if they are somehow related? (Hint: there are huge synergies)
__
Join us on Monday to follow along our journeys.
This won't be us preaching about a perfect solution, or an approach.
It'll be a casual conversation about what we've learned from our own internal initiatives and from talking to a bunch of data teams and business stakeholders about the new AI era.
Save your spot. It'll fill up fast!
(I don't think there are actual limits on who can attend, but you should still sign up!)
👩🚒 Call the fire department, because here's a hot take 👩🚒
Webinars are like...super boring
Here's my experience with webinars:
1. See an ad or get an email about a relevant-sounding webinar (sometimes genuinely interesting sounding, sometimes MEH)
2. Check my calendar -- if I'm free, I sign up
3. Completely forget about the webinar and double book myself
4. Tell myself its ok because I'll watch the replay
5. Open the replay only to realize I'm watching a Zoom call...tune out, eventually close the window early
Here's the thing, though...Cotera's webinars are run by Ibby Syed.
If you know Ibby at all, you know there is absolutely no way I can put him on a boring script or talk track. This guy is just too excited and excitable.
So our webinars have turned out (dare I say it) fun and engaging. They're highly interactive. (I actually asked if I could call them workshops instead, but ~2/3 of the people are still screen-off, and we didn't want to scare them away if they just WANT to listen/watch.)
And this next one is going to be really cool.
Ibby is going to show how you can use Cotera to handle phone calls.
Yes.
Live phone calls.
It is wild. I can't decide if I love it or hate it, but it is crazy to see. You need to check it out (signup link in the comments!)