7

Toller - Toller retrieves info in your documents and shows sources | Product Hun...

 1 year ago
source link: https://www.producthunt.com/posts/toller
Go to the source link to view the article. You can view the picture content, updated content and better typesetting reading experience. If the link is broken, please click the button below to view the snapshot at that time.
neoserver,ios ssh client

Support is great. Feedback is even better.

"Hi everyone! Looking forward to having you test out our product! > You'll immediatly get a link to our Slack where you can test the tool by directly dm-ing Toller > We'll roll-out a "add to Slack" link so that you can use it in your own team

Do you have any questions for us? :)"

The makers of Toller
Sort by:
Kudos at shipping! I imagine this could be used in many different industries. Any industries or niches you are especially excited about to work with? Would you consider a deep-dive into some with subject matter experts?
@__tosh Thanks! Indeed, domains where large unstructured documentation is the core business is the key (e.g. legal, regulatory, consulting). Right now, Slack is just a good way to showcase the tool, but we already work with subject-matter experts and are excited to find new venture partners to tackle more specific problems. Then we can make it more usable for these experts in their day-to-day life, and have a more cognitive component that showcases how to connect the dots between the various sources.

Impressive Demonstration, i especially like the sources feature, as hallucinations in a business context would be the worst.

What happens if two teams in a larger company start loading up files with conflicting concepts/statistics? Or how would it separate market predictions from financial reports? (Even if there are more files for the former)

@felix_hau Thanks! From my own testing, it is capable of separating different opinions from different sources and laying them out in the answer. But prompting is important in this case. The window shows top sources, but we have all sources under the hood (just don't want to display giga-amounts of text). And of course, I haven't tried all use-cases! Looking forward to getting more feedback.
A powerful tool to make searching and synthesizing documents easier? Wow! I'm curious to know how well Toller works on large datasets - what are your experiences?

@samalyx Hi Samar! I think that's exactly a differentiator with current solutions: our system is quite lean, and large language models are just the "cherry on top" at the very end of the transformation pipeline, just to make it "useable" for humans.

For the free version since we literally just made it ;) we will give out 1000 pages and 25 users. The "sweet spot" is documents of ~200 pages or less (it takes only a few seconds to upload and analyze).

But we know that we can scale to heavier sets when we can buy ourselves a little bit more computational power! In the pipeline, we can also track websites for you and build automated updates, and also let you build your own app using our solution. But that's not in the scope of today's launch :)

@samalyx oh and right now it works best on unstructured text :)
Congratulations on your launch! Is it safe to send you confidential documents? How is data security ensured? Thank you!
@malkielfalcone Excellent question! Two aspects: 1/ Toller uses strong encryption to protect sensitive data, both in transit and at rest. We use a combination of symmetric and asymmetric encryption. Data is encrypted and we cannot read it :) 2/ If your organization needs it, it can be self-hosted ;)

report

Recommend

About Joyk


Aggregate valuable and interesting links.
Joyk means Joy of geeK