Hi All; Thanks for checking out HeadlinesAI! We haven't quite launched yet, we opened the registration floodgates briefly for an advertising test and got hunted - But thanks for noticing, Chris :) We're still improving the NLP - our early customers tell us that the product works really well for long text/articles in a few topic verticals, but as a few of you found there are still some challenges to solve for short queries and niche topics. Stay tuned in the next couple of months for some updates and a public launch!
We are building this product to replace a huge portion of article headline and content marketing A/B testing. I was previously a data scientist in the viral media industry (sorry for all the cat videos...) and friends have been asking me for the past 3 years, 'what headline should I use for my blog post?'. With HeadlinesAI, we hope to answer that question for everybody.
In the meantime before we launch, if you'd like a registration invite reach out to contact@headlines.ai and we'll give you a demo right now and a free trial when we launch :)
@flowsion thanks for trying the product Flowsion! We'll have some dramatic improvements coming online in the next month before we launch, ping us if you'd like another demo :)
Report
@patrick_martinchek Would love to! Let me know when it's up. flowsion@me.com. Reading through the comments, it looks like this is an alpha build. So, I will leave my feedback for when it's live! I'll update this as you send me the new demo later.
I work for a super niche industry so that could be why there was problems with accuracy. More than happy to work with your team to get more niche industries down right on HAI.
@flowsion Sounds great, will do - thanks for the support :)
Report
Needs better content similarity matching, as it is essentially the core of what your product does. I was hoping it can generate the 'best' headline automatically but this doesn't seem to be the case here.
@bluemonk482 Definitely, thanks for the feedback Bo. We originally wanted to build a product that would always deliver the single 'correct' and 'best' headline for a given query (perhaps even using NLG), but we found that even the best NLP wasn't good enough to get it right 100% of the time. So instead, we offer users the top similar and highest-performant headlines as suggested by the ML model, and then allow users to complete the last 5% of the process and filter out any egregious mistakes. It's also helpful to deliver a few options so users can select the headline inspiration that would resonate best with their audience. Thanks for the feedback!
Report
@patrick_martinchek You're quite right, it is still a big challenge to summarise large text corpus and generate a headline. Your approach makes more sense since your goal is to help with content marketing rather than generating the most accurate summary/headline. Best of Luck! :)
@chrismessina thanks Chris! This was impressive - you sniped the product during the very brief period that registration was open :) Thanks though, appreciate the hunt!
Report
Top 3 results for 'growth hacking'
Yeah, The NSA Was Probably Hacked
Google CEO Sundar Pichai Hacked By Zuckerberg’s Hackers
Hackers Are Spreading Malware Through Pokémon GO
It may need some tweaking :)
The interface on an iPhone needs work too.
@beuk Thanks for demo'ing the product Johan! Definitely still have some work to do with shorter queries before we officially launch:p I'm actually surprised that all of the top results directly had the word 'hack' in them, usually the matched keywords are buried in the text leaving a nice curiosity gap in the headline. For some of our early customers we've built customized models that are specific to their target audience, and this has greatly improved the performance of the short queries. But our large public models are still a bit rough for the short queries. Please reach out if you'd like another shot at your "growth hacking" query, we can put together a more focused demo for you :) contact@headlines.ai
Report
@patrick_martinchek Hi, Thanks for your reply. Just some more thoughts; I think it would be cool if the results have a drill down option by clicking suggested headlines?
Will you also make the user interface mobile friendly? I'm guessing a lot of bloggers write posts using their phones and yesterday the experience on my iPhone was not really good (yet?) ;)
What will your final pricing model look like?
@beuk sure thing, the multi-level drill-down option sounds interesting. We made some substantial breakthroughs on the NLP side re: semantic lookup efficiency, so I think we could actually build this feature to be consumer usable without causing too much frustration waiting for results .. grumble .. something about Google's insanely fast query times that all of us are now benchmarked against:P
Ah yes, mobile-friendly improvements are already under construction :)
Pricing model is a monthly subscription. We've seen a lot of social media and advertising tools explode in cost, but we really want to deliver this to everybody from startups to large corporations at an affordable cost to all. We're doing price sensitivity testing right now, so I don't want to post the current pricing but reach out to me directly if you'd like a preview of our current pricing tiers.
Report
Hmm wonder how this will do. Has anyone tried it yet?
From what I can tell, if the input is long and if it's well publicized it does ok - if you sort by similarity. I tried copy & pasting some articles like Snapchat IPO (did well), how stack overflow makes money (did poorly). But if you sort by performance, its a little weird: I copied the article where this product was discovered and the top suggestion is "A Complete Guide to Facebook’s Q2 2016 Earnings Call". Not sure if this would replace A/B testing but might be nice to see similar articles...
@seokho_seo Thanks Seokho for diving in. These results are also consistent with the feedback we're hearing from our beta customers. Stay tuned in the next couple of months, would be great to have you test drive the product again when we have an official launch!
Report
@patrick_martinchek Hey Patrick! Just read the top comment and that sounds awesome. Looking forward to future updates.
Product.Jobs
Product.Jobs
Product.Jobs
Product.Jobs
Product.Jobs
Raycast
Product.Jobs
Product.Jobs
Product.Jobs
Product.Jobs
Product.Jobs
Product.Jobs