Agentplace AI Agents - Create specialized AI agents for real tasks and workflows

@timte Thank you so much, really appreciate it! Excited to see where we can take it too :)

@timte means a lot. really excited to see where this goes too.

@madalina_barbu  Thank you so much!

@madalina_barbu Really appreciate it, thank you!

@abod_rehman Thank you for your support!

@abod_rehman That’s actually a great use case for it!

@abod_rehman I also like that it can answer (make a draft for me) based on my real voice and info. So i need just to approve it or add my bit.

@abod_rehman That’s a great use case email triage feels like one of the most natural jobs for an agent.

@evgeny_sorokin  @vlad_yanch  @maria_polesh Thank you, Maria!

Yep We’re already seeing agents used for research, content drafting, and packaging things for distribution. Marketing feels like one of the most natural areas for this.

Hi Maria@maria_polesh, I think research is the most exciting and powerful use case. AI can naturally process large amounts of data, and agents can search from multiple angles, refine their queries, review blog posts, and go deep into a project’s docs to build a full picture. Research is my personal favorite, but agents can improve every part of marketing, including content and distribution. We still think human judgment matters, so this should be a semi-autonomous agent.

@maria_polesh Yep research, content, and distribution workflows are already feeling like a very natural fit for agents.

@sayanta_ghosh Yeah we have a few. Lead router that catches leads from forms, chat and email and pushes them to the right place in HubSpot. Competitor researcher that gathers intel and puts together positioning briefs so you don't have to do it manually. What kind of GTM workflows are you thinking about?

@sayanta_ghosh  Thank you! Please share your thoughts once you try Agentplace 🙂

@sayanta_ghosh Oh man, I think there are many, many of them. The latest use case from the top of my mind is an agent for company analysis, gathering data, and making a landing page specifically for that business with their pain points.

@lak7  Honestly this is an area we're still working on. Right now each agent runs in an isolated VM and you have SSH access from the builder so you can check what's happening. For published agents we don't have built-in logging or audit yet. Since you have full access to the source code though, it's pretty easy to plug in something like Langfuse or Sentry or whatever monitoring you already use. So right now this is more on the user side, but built-in observability is on our list.

@lak7 That’s exactly why transparency matters so much to us. If agents are doing real work, people need a clear way to inspect actions, audit decisions, and catch failures early.

@piotreksedzik Both actually, it depends on how you set up your agent. By default agents can work with PDF, Word, Excel, CSV and images out of the box. But since each agent is a full app you can add any library (or MCP server) you need, so if you want something specific for medical docs you have full control over that. We also have integrations through Composio for things like Aryn or Affinda that handle unstructured document parsing, OCR tools like Google Cloud Vision or OCR Space for scanned docs, and CloudConvert if you need to deal with weird file formats. So the short answer is common formats work right away and for anything more specific you can plug in whatever fits your case.

@dan16 Thank you! Seemed like an obvious choice for us :) What's interesting is that agentic websites are really just one narrow use case. A website waits for someone to visit and then responds, but agents can do a lot more than that. Once we saw that, building only websites felt like using 10% of what agents can actually do.

@dan16 Thank you so much, that really means a lot. We learned a lot from how people were actually using it, so following that signal felt like the right move.

@dan16 that was a hard pivot tbh :)

@dan16 We tried to listen closely to how people were actually using it and follow that signal.

@a_6 I think “computers” will adapt to us. They’ll use whatever modality feels most natural, including adjusting within those modalities. In other words, if it’s easier for us to see a graph instead of code, AI will show us a graph.

To make this more obvious, just look at the GUI. It was a revolution when it replaced command-line interfaces.