Incredible
Deep Work AI Agents - powered by Agent MAX
Start new thread

Incredible - Deep Work AI Agents - powered by Agent MAX

by•
Create AI agents that work 24/7, cost 90% less, and don't hallucinate on data—powered by Agent MAX. Automate daily repetitive tasks with the world's most advanced AI agent engine built for autonomous work. Hundreds of integrations. 90% cheaper. Near-zero hallucinations. 100x memory.

Add a comment

Replies

Best
Maker
📌

Hey Product Hunt! 👋 Philip here from Incredible, a small (but dedicated) 11-person team from Sweden.

We've spent the last 4 years obsessed with one problem: why don't AI agents actually work?

... They hallucinate. They forget what they're doing halfway through. They loop forever. They fail silently. Everyone building with AI agents hits these walls eventually.

We hit them 14 times... across 14 different AI products... since 2021...

Each one taught us something about what breaks. Eventually we stopped patching at the product level and started building the foundational layer that handles all of it.

That became Agent MAX.

Introducing: Deep Work AI Agents — powered by Agent MAX

AI agents that can actually run on their own & are purpose built for autonomous tasks. Without breaking.

Here's what Agent MAX solves:

  • Near-zero hallucinations: Function calls use real data, not made-up parameters (maps in data)

  • 100x memory: Deep context compression so agents don't forget the goal mid-task

  • Anti-loop: Stuck detection (no more infinite retries or loops that burn money)

  • Self-healing: Automatic error recovery without restarts + very cheap

  • 90% cheaper: Context engineering + better representations for model to interact with data & actions

What our early users are doing:

  • 📊 Processing entire CRM databases — deduplication, enrichment, updates across thousands of records

  • 📧 Running email workflows that used to take 3 hours/day, now passive

  • 🔄 Syncing complex data across tools without babysitting

  • 📈 Generating reports from live data pulls across multiple systems

There's hundreds of integrations built in. Google Workspace, Notion, Salesforce, Slack, Linear, HubSpot... more coming.

We don't have billions in funding. We're just an 11-person team in Jönköping, Sweden who've been grinding in the basement on this for 4 years.

🚀 Try it:

🎁 Product Hunt exclusive: 90% off Pro for 3 months — first 100 users only. Code: PRODUCTHUNT90 (expires Friday 12th)

For developers: Incredible API gives you access to our Agentic AI models, and Agent MAX is in waitlisted early access, will soon be available for all as a research preview. - visit: docs.incredible.one

—

The entire Incredible team is here today — ask us anything! We'd love to hear your feedback.

Philip, Emil, David, Manas, Anika, Nikola, Fredric, Yesith, Abhinav, Suvi, Krrish

What repetitive work would you want agents to handle? Drop your use case! - we're building starter agents based on what people actually need.

💬 Join our Discord for direct support & ideas:

Tack! 🇸🇪

Excited to see what you create! Here's my most recent one:


Maker

 If anyone is interested in the Technology that is behind this product, check out Agent MAX that is coming soon to our API. It allows you to build powerful agents in just a few lines of code. Don't worry about tool-loops, context management, or optimization, we handle all of that for you. You as a developer just dump data, tools, goals, and the agent will make sure it happens. Sign up to our waitlist if you want to get access earlier than everyone else.

Learn more:
Waitlist: )

 Hallucinations and Agent forgets the thing are the 2 main problems I encountered on the other platforms. I was just curious to know, how you tackle these problems. Congrats on the launch.

Maker

 hey! thanks so much :)

Yeah those two are the big ones. For hallucinations, every output gets validated against your actual data before execution & the model maps in data into function calls = it doesn't produce tokens for it. If the info isn't there in previous data that it picked up, the agent won't make it up. There's lots more info here if you want to read about it:

For the forgetting problem, it's two things. We keep the agent anchored to the original goal throughout, with validation checkpoints along the way. And "Deep Memory" handles the context side, compressing context + different data representation, so the agents can work with 10-100x more data without losing key details.

These were the exact pain points we kept hitting too, we've definitely obsessed over them.

 Nice, going to check this out!

Maker

 let us know what you think would love to hear it! :)

How does Incredible differ from other solutions?
Maker

 

Great question!

vs ChatGPT / Claude: They're great for chatting, but they don't really do things. Incredible agents run in the background, connect to 200+ tools, and actually finish tasks without you sitting there. ChatGPT also forgets what it's doing after a few steps. Agent MAX holds 100x more context so it can get through long workflows without breaking.

vs Zapier / Make: Love Zapier but you gotta build everything yourself. Triggers, steps, conditions. (even with their new chat ai versions it's still hard to get it the full distance) With Incredible you just describe what you want and it builds the automation in like 30 seconds. And the agents can actually think and adapt, not just follow a rigid flow.

vs other AI agent platforms: Honestly the big ones are that it doesn't hallucinate (checks actions against real data) and it's about 90% cheaper to run. So you can actually afford to automate the smaller stuff too.

Happy to go deeper on any of these!

PS: Great read here on Agent MAX benefits:

Maker

 happy you're enjoying it Daniel! Let me know if you have feedback, would love to hear it :)

Agent MAX represents a significant move toward compliance-ready automation, its reductions in hallucination, structured task execution, and enhanced traceability push autonomous agents toward the level of reliability and auditability modern legal and operational frameworks demand.

Maker

 Yes! We're super excited about getting that out into people's hands very soon

great product-great team, congrats

Maker

 That's nice of you to say! A lot of time & energy went it to this launch.

We're trying to keep up with all the outreach we're getting right now

Amazing work by an Incredible team! 👌🏻

Congrats on this milestone. Keep up the great work!

Maker

 Thank you for the kind words!

Congratulations on the launch 🎉 🎉

Maker

 thanks a ton Shubham :) Entire team has been heads down working on this for soo soo long. Exciting times. Appreciate the support!

Did you try creating an Agent yet?

support from your follower on LinkedIn!

Maker

 haha awesome Mikitia, appreciate it a ton! Hope you enjoy the product. Let us know your feedback!

Many congratulations on the launch and the entire team!

Impressive approach. Most agent tools avoid the core reliability problems but Agent MAX seems to tackle them head on. One question. How does Agent MAX handle inconsistent or partial API responses during long runs.

Looking forward to trying it.

Maker

 thank you Mausam! Looking forward to hearing your feedback :)

Killer launch what does the pricing look like at an API / Production level? curious what baseline LLM this is built on (if multiple we'd love to know)

Maker

       

hey Daniel, awesome question -

Here's a short recap of the journey:

→ SWE-LLM (Sweden's largest open-source fine-tuned model at the time)
→ Agentic model family: small 1.0, small 2.0, big 1.0
→ Agent 4 — our most capable agentic AI model for "autonomous" work
→ And now: Agent MAX

All our agentic models are a mix of smaller open-source models fine-tuned on large scale (continued training) on a massive dataset of workflows, SOPs, unique data representation (we call it "symbolic data representation"), and our own function calling methodology. There's tons more of course, but this is the major components that lead to majority of benefits.

Pricing for the API is available directly in the API:

You can also find more information right here:

 Thanks for the detailed response - sounds comprehensive and like the model in an of itself could be part of your differentiator / USP.

😹 LOL
Kudos to this badass team from Sweden. Special shoutout to Fredric and David who’s been answering all my stupid questions while I’ve been playing around with incredible last month. Great work guys!
Maker

 Haha funny guy!! I heard Bazaar killing it!!

1234
Next