Tristan Pollock

Incredible Small 1.0 by Incredible - Agentic AI Models that can take 1000+ actions at once

RESEARCH PREVIEW β€” Agentic AI Models that can take 1000+ actions at once, and work with large datasets. 100+ app connectors. Try the research preview assistant now at www.incredible.one. Built in Sweden, by Incredible πŸ‡ΈπŸ‡ͺ

Add a comment

Replies

Best
Philip Alm

Hey Product Hunt! πŸ‘‹ Philip here from Incredible, a small 13 person team from Sweden.We've spent the last 3 years on a different path than most LLM/AI companies.

While others optimize for general intelligence, we've been obsessed with one thing: making AI that actually gets work done.

Our journey has been... unconventional. We built 14 different products over the past 3 years, tested with 1000+ users, served 250+ paying customers. Each iteration taught us something new about what holds AI back from doing real work.

Last year we released SWE-LLM (continued training, Sweden's largest language model) as an experiment in language understanding. That laid the groundwork for what we're sharing today.Introducing Incredible - Agentic AI models designed specifically for taking actions in your apps.

Here's what makes our approach different:

1. Hundreds of actions at once: While other models update 1 spreadsheet row at a time, ours can update 1000 rows simultaneously

2. Gigabyte-scale data processing: Handle 10,000+ CRM records in a single operation (vs the typical 50-100)

3. 200+ app connectors built-in: Google Workspace, Notion, Salesforce, Linear, Slack... more

4. No hallucinations on data: Our "live-code" architecture means exact accuracy in valid data operations

Some things our early users are doing:

- πŸ“Š Analyzed an entire CRM database, identified 847 duplicate leads, merged them, and updated all related records

- πŸ“ˆ Generated personalized sales reports across different regions with real-time data pulls

- πŸ”„ Synchronized complex project data across 5 different tools simultaneously

- πŸ“ Processed 1000 tickets of customer feedback and created actionable insights dashboards

We don't have billions in funding. This is still a research preview - we're learning and improving every day. We'd love for you to try it and tell us what real work you need AI to handle.

πŸš€ Try the research preview assistant: incredible.one

πŸ› οΈ For developers: Drop-in replacement for OpenAI/Anthropic APIs with MCP support

PS: Flagship model coming soon.

What work would you want AI to actually complete for you? Let us know in the comments - we're hoping to hear from Product Hunters sharing specific use cases!

Tack! (That's Swedish for thanks!) πŸ‡ΈπŸ‡ͺ

Sofiia Katalandze

@philip_alm_at_incredibleΒ Huge congrats on the launch! Can’t wait to see how Incredible reshapes workflows with so many integrations.

Philip Alm

@sofiia_katalandzeΒ thank you Sofiia, hugely appreciate the support!

@philip_alm_at_incredibleΒ How do you envision Incredible fitting alongside existing LLM APIs, as a complementary or a replacement in most workflows?

Philip Alm

@curiousheadΒ Honestly, we're just focused on making it able to do as much work as possible.

Other's really maximize for intelligence. I think there's a common thought with LLM developers that, if you want an LLM in your product to write code, you use Claude. For intelligence/creative writing, GPT-5. Incredible is for taking actions in apps & products.

There's a huge need for it. Every startup/scaleup who's currently burning through millions on inference could potentially use it to greatly improve their unit economics, while getting an even higher performance.

If you try AI inside of apps, like Notion's AI, gemini in spreadsheets, Slack AI etc, they're all struggling with large actions. We solve that. And we hope to provide it to a lot of companies currently trying to get their agents to work.

Ghulam Abbas

@philip_alm_at_incredibleΒ Super impressive work Philip & team πŸ‘ Excited to see AI focused on actually getting things done rather than just talk. Curious to try it out!

Philip Alm

@abbas143officialΒ thank you Ghulam! Looking forward to hearing what you try! :)

Tuneer Biswas

@philip_alm_at_incredibleΒ congrats on the launch !

Philip Alm

@cheerstΒ thank you a ton Tuneer! :)

Mcval Osborne
πŸ’‘ Bright idea

Hey @philip_alm_at_incredible this is an awesome platform, I've started messing around with it and the Google sheets to trello agentic example is killer.

Philip Alm

@mcval_osborneΒ haha love it! You should try the prompt "Make an industry report on AI agents in a google spreadsheets". It's insane what it puts together in just a few mins

Emil Wagman

@philip_alm_at_incredibleΒ  @mcval_osborneΒ Love the feedback! Please share any cool use-cases that you find πŸ˜„

Yagyansh Bhatia

Can we automate this?

Philip Alm

@whybee99Β there's actually a ton of potential there!

Right now we're seeing people save their prompts in notes and re-running them over and over haha.

We'll see about adding a schedule button asap, it seems to be highly requested! :)

Nika

Hey team, currently testing it and have a suggestion: It would be great if the tool could remember the prompt after signing up. :)

Philip Alm

@busmark_w_nikaΒ that's a great suggestion! What do you think about adding a prompt library type thing, where you can save + share prompts with each other?...

Or do you have any idea of how you would love it to work? :)

// P

Madalina B

Congrats! β™₯️😍 Looks great

Philip Alm

@madalina_barbuΒ haha thank you Madalina, glad you're enjoying it!

Shubham Pratap Singh

Congratulations on the launch πŸŽ‰ πŸŽ‰

Philip Alm

@shubham_pratapΒ thank you Shubham :) 100% appreciate the support. What have you been enjoying soo far? :)

Sumeet Thakkar

Congrats on the launch! πŸ‘ Incredible looks really powerful. While most people will think of productivity workflows, I see a big opportunity in employee offboarding automation, which is a huge compliance and security challenge.

πŸ‘‰ Without Incredible:

  • HR/IT manually revokes access across email, Slack, CRM, project management tools, and cloud storage

  • Transfers document ownership one app at a time

  • Sends manual notifications to stakeholders

  • Process takes hours, and things are often missed

πŸ‘‰ With Incredible:

  • Trigger an β€œemployee exit” flow once

  • Instantly revoke access across all connected SaaS tools

  • Automatically transfer files/documents to the manager

  • Notify HR, IT, and the team in Slack/Teams simultaneously

This kind of automation doesn’t just save hours β€” it also reduces risk by ensuring nothing slips through during offboarding. Excited to see how teams adopt Incredible for cases like this!

Mathias Karlsson

@sumeet_thakkarΒ Thank you! We're taking notes on everything here and will keep pushing forward to build the best platform we can. Thanks again for the support and for being part of the journey, it really means a lot! πŸ™

Andrei Tudor

This is super impressive, Philip! The focus on β€œAI that actually gets work done” really resonates, as most tools talk about intelligence, but real adoption happens when workflows are simplified at scale.

I had a similar learning while building Escape Velocity AI (different domain, strategy & business planning): users don’t want more prompts, they want outcomes they can trust. Seeing how you’re applying that principle at dataset/action scale is exciting, and I look forward to see Incredible's next phase.

Do you see your early adopters leaning more toward sales/CRM use cases, or are ops/finance teams picking it up too?

Philip Alm

@andreitudor14Β happy to hear Andrei! There's a huge amount of interest in excel/googlesheet reports from CRMs like Hubspot. Lots of CMOs, data analysts pulling out data and doing analysis. But we see a lot of promise in marketing/sales and ops too.

Looking forward to see what people build. Any ideas for what you could do with it? :)

Andrei Tudor

@philip_alm_at_incredibleΒ makes total sense, as spreadsheets and CRMs are where so much of the β€œgrunt work” still lives, so cutting that down is a huge win.

The first use case that comes to mind for me is tying strategy or planning outputs directly to live execution data. Imagine being able to map assumptions in a business plan against what’s actually happening in CRMs or ops dashboards and seeing the variance in real time.

Do you see more teams experimenting with this β€œstrategy meets execution” angle, or are they sticking closer to day-to-day reporting for now?

Philip Alm

@andreitudor14Β gonna try this one myself. One of the crazies ones we've seen soo far was "Figure out why some customers are performing better, and some worse. use a spreadsheet to break it down" connected to the CRM. Produced some very interesting insights.

Felix Wunderlich

Very cool, congrats on the launch! What's your approach to no hallucinations on data?

Philip Alm

@felix94123Β thank you Felix! :)

It's actually quite cool. We built an Architecture we call live-code. The models don't actually "see" the data, instead they see symbolic representations on it that we've trained them on. This means they never "predict" tokens like other LLMs do for data, instead they kind of map it directly to whatever action they're taking. E.g. instead of writing our JSON to take actions, it just maps in the information to whatever app or soo it wants to use = no hallucination on data, cause there's no LLM involved in that step.

We tried breaking down some more details here:
https://web.incredible.one/blog/announcing-incredible-small-research-preview

Felix Wunderlich

Intersting, thanks for sharing!

Wayne

A FRIENDLY AI-Agent to non-technical professionals. I just type in a taskβ€”like β€œSend weekly Friday report to team”—and suddenly I have an AI colleague doing it. AI also helps me build multi-step workflows with just a few clicks. If possible, I hope to set up a function in the dialog box that can give me officially recommended high-quality prompts to save time thinking about prompts.

Anna Willstedt

@wayne_appgrowingΒ Thanks for the comment Wayne, there's a lot more coming :).

You should 100% try out the Trello project planning demo. It shows the 100+ actions at once and how it can work with a lot of data :)

What's your favorite use-case soo far?

123
β€’β€’β€’
Next
Last