We have trained Leo to detect and understand funding events. This means that you can now ask Leo to read your tech, business or industry specific feed and prioritize articles related to funding events – saving you a tremendous amount of time.
We have also trained Leo to detect and understand product launches.
This means that if you are part of a sales or sales enablement team, you can ask Leo to read TechCrunch or New York Times and notify you each time one of your customers or prospects launches a product. You can leverage that product launch event to create a warm approach and engage in a smart conversation.
Finally, you can also easily prioritize the fraction of articles referencing partnership announcements.
Composable with other skills
The business event skill can be composed with all the other Leo skills allowing you, for example, to easily prioritize articles referencing a product launch (business event skill) and related to #artificial intelligence (topic skill).
Trained across 24 industries
Different industries use different vocabulary to describe these business events so we trained Leo across 24 different industries.
You can use the Leo prompt or the “less like this” down arrow to correct Leo when the event detection is incorrect. This feedback is channeled to the Feedly ML team and to the datasets used to train Leo.
The Leo business event skill is available on Web and Mobile now for Feedly Teams users. Because we have been getting a lot of Leo requests from non-Team users, we will also be launching a $12/month Pro+ plan in October which will include Leo, a Twitter integration, and all other Pro features.
Curating content to share on Social Media and want to avoid mentions of your competitors?
You can train Leo to mute each of your competitors and automatically remove all the articles mentioning those competitors.
Want to avoid articles about specific celebrities, politicians, or executives?
Creating a Leo mute filter for a celebrity, politician or executive will automatically remove all the articles that mention that person from your feed.
Want to avoid a spoiler about Game of Thrones until you have finished reading all the books or tired of hearing about Pokemon Go or the latest Apple monitor?
You can train Leo to mute specific keywords and remove all mentions of those keywords from your feeds, temporarily or permanently.
Note: with Leo Mute Filters, you no longer need to use quotes around phrases with spaces. Leo will take care of converting the input into the right query.
Following a broad source like TechCrunch, Wired and Forbes but do not care about topics like gaming? Or following a keyword alert for a public company but do not care about financials or market reports?
With Leo mute filters, you can mute topics and increase the focus of your feeds. Leo ships with 1,000 pre-trained topics.
Do not like a specific author from one of the sources you follow?
With the author: operator, you can train Leo to look for specific authors and mute all the articles from that author in your feed. (Sorry Katherine, we actually love your work!)
Mute title patterns
Want to remove articles which have a specific keyword in their title?
With the title: prefix, you can train Leo to look for a mention of a keyword in the title of the article and mute the matches.
Finding some of the sources referenced in Google News Keyword Alerts irrelevant?
With the site: prefix, you can train Leo to mute specific sites from your keyword alerts.
Forever or temporarily
When you create a Leo mute filter, you can specify a duration.
Once you have trained Leo with a mute filter, you can easily remove, pause or resume that priority via the Train Leo page.
Like with all the other Leo skills, it was important for us that you always feel in control and can continuously refine your Leo as your needs evolve.
When reading articles, Leo will highlight the most salient entities mentioned in the content. This makes it easy to click on them and priorities or mute those entities.
You can also highlight any snippet of text and mute that phrase
Finally, when reading an article, you can click on the Less Like This button and easily mute one of the topics Leo has associated with the article
On mobile or on the web
The Leo mute filter skill is available both on the Web and on mobile (version 65+).
You can train Leo to mute topics and keywords on mobile.
From a feed
From an article
From less like this (long swipe from right to left)
Curious about trying Leo Mute Filters on some of your feeds? Join the Leo program
Pro users will be able to continue to use a more basic version of mute filters. The syntax of those mute filters have changed to the v2 syntax to allow more efficient processing on the back end.
Some of the v1 mute filters using advanced queries can not be migrated to v2 will remain active as legacy filters until user delete them.
Are there limits to the number of Leo mute filters a user or team can create?
One of the benefit of the Leo mute filters is that they can be processed more efficiently by our back-end. As a result, we are increasing the limit of Leo mute filters for Teams user from 100 total to 100 per feed.
Can non-Teams user access Leo?
We will be offering Leo to non-team users later this year via a Feedly Pro+ priced at $12/month. You can requestearly access to Pro+ here.
Can a mute filter target a specific source?
No. Mute filters can target a list of sources (what we call a feed) or all your feeds.
Over the last twelve months, we interacted with hundreds of cybersecurity teams. One of the common murmurs we are hearing is that it is increasingly harder to keep up with trends and threads in the security space.
In 2018, fifteen thousand vulnerabilities were discovered and the number of exploits doubled – resulting in about four new security articles getting published every second on the Web.
This is a problem we are very passionate about so we are excited to announce a new Leo Security Skill that allows you to prioritize within your feeds the articles that reference the most critical vulnerabilities.
It is a powerful way of focusing your attention on the 10% of vulnerabilities that matter the most – taking into consideration the CVSS score, the content of the article, the level of awareness of the CVE and the products/vectors your care about.
For example, here is a quick tour of how you can train Leo to prioritize the high severity threats related to Microsoft products.
Discover the Best Cybersecurity Sources
The first step, if you do not follow vulnerability sources yet, is to click on Add Content and search for #security or #vulnerability. You will see a list of about one thousand security publications, blogs, and subject matter experts you can easily add to your Feedly. Create a Vulnerabilities feed and add ten to fifteen sources.
Because Feedly is an open platform, you can add any source you want to follow that publishes an RSS feed.
Train Your Leo
The second step is to train Leo to prioritize the most critical vulnerabilities in your feed. Most security teams care about the top 10% of the vulnerabilities that have a CVSS score greater than 8 and/or have an exploit.
The Leo Security Skill allows him to either lookup or predict the CVSS score of a vulnerability mentioned in an article. So when a new article is published in your feed, Leo will first try to lookup the CVSS and exploit information from the Web. If there is no CVE or CVSS, it will try to predict the severity of the vulnerability based on the content and terminology used in the article.
Training Leo to prioritize high severity vulnerabilities around products . you care about is simple.
In the priority modeler, add a first layer of type Security Threat and select the High threshold.
Then add a second Topic layer and pick the list of products you would like Leo to track. Leo will combine both layers and look for high severity vulnerabilities mentioning the products you care about.
Read, Share, and Shine
Leo will continuously read your Vulnerabilities feed and when an article matches the high severity and mentions the products you care about, Leo will annotate that article and move it to your priority queue.
When you open your Vulnerabilities feed, you will first see the shortlist of articles Leo has prioritized. If Leo has found the CVSS information for the mentioned vulnerability, you will see it as part of the metadata of the article.
Prioritized article have a green marker with the name of the priority. If you click on that marker, you will be presented with a short explanation of why Leo prioritized this articles and the controls for you to refine Leo’s training.
This aspect around control and transparency is really important to us. It is what we call collaborative intelligence.
If you see an article or vulnerability that is particularly important, you can save that article into a Feedly board and configure that board to push the content to an email newsletter, a Slack channel or a Microsoft Teams channel. Boards are a powerful way to keep important articles for reference and easily share with your teammates.
Continuously Learning and Getting Smarter
One of the powers of Leo is that he is constantly collaborating with you and learning from you. If you see an article that is highly relevant, you can save it to a board and then use the content of that board to re-enforce Leo’s learning via a Like-board skill.
If Leo was wrong about detecting a vulnerability, assigning a severity to it, or detected a product you are interested in, you can at any point of time click on the down arrow icon (also called Less Like This icon) and provide feedback to Leo.
That feedback is process daily and used to continuously improve the various machine learning models used to power Leo.
Join the Leo Beta
The Leo cybersecurity skill was created over the last 12 months in close collaboration with two of the largest and most advanced security teams in Silicon Valley.
We are excited to hear what the Leo beta community thinks about this new skill! If you are part of the security team and would like to test drive Leo Cyber Security, please join the beta program.
We pushed Leo 0.5 to a limited beta in early March and collected lots of interesting feedback. The team is listening and crunching through all that feedback and adapting Leo to improve UI/UX as well as the relevance of the underlying machine learning models.
Here is a summary of the changes we are pushing out today as part of Leo 0.6 Beta
One of the feedback we collected was that the difference between mentions and topics was not clear. So in 0.6, we merged these two concepts into a single one we call Smart Topics. Just search what you want to prioritize and Leo will start analyzing the content of your feeds and prioritize the articles which are a match.
Level of Aboutness
Sometimes you are interested in a company, product, or topic and you want to see every article mentioning that topic. Sometimes, for more popular topics, you are only interested in reading an article if the article is truly about that topic or company.
Leo 0.6 exposes a “level of aboutness” knob that gives you more control over the model so that you can cut out low salience matches.
For example, if you are interested in NLP or BERT, you can train Leo to only prioritize research articles that are prominently about those topics (as opposed to articles which only briefly touch on those topics).
This is a particularly powerful feature when combined with Google News Keyword alerts.
Some Leo 0.5 beta customers mentioned that it was critical for them to be able to define priorities that span across multiple feeds. For example, you might be doing research about Stablecoin and want to prioritize that topic across both your Tech feed, your Business feed, or all your personal or team feeds.
In Leo 0.6, the priority designer allows you to pick “All Team Feeds” or “All Personal Feeds” as the scope of the priority.
This change reduces the total number of priorities you need to create and manage when researching topic and trends across multiple of your feeds.
Some users mentioned that they would like to be able to navigate their content by priority. If you are interested in a specific topic like Docker, it makes sense to be able to quickly see if there are new Docker related articles in your Feedly and easily access those articles.
In Leo 0.6, we added a new Priorities section to the left navigation bar that surfaces all your priorities and gives you quick access to all the article Leo has flagged as important.
We added two settings in the Leo settings to let you personalize this feature. You can decide if you want to see priorities in your left navigation. If you want to see all the priorities or all the global ones (default). If you want to see all the priorities or only the ones which hav unread articles.
Your interests and priorities are continuously evolving. Often, you discover a new company, product, or topic while reading an article and you want to be able to teach Leo about it.
In Leo 0.6, the most prominent topics mentioned in an article are highlighted so that you can quickly prioritize them (or mute them)
As part of Leo’s Cyber Security skill, you will also see highlights of CVE entities. More to come soon.
Like for the Quick Access feature, there is a Leo setting that allow you to turn off Inlined Entities if that is your preference.
Like Board Improvements
The ML team is spending time understanding how you are engaging with your priority feeds (which articles are saved to a board, which articles are being Less Like This’ed) and tuning the underlying ML models to improve accuracy. You should expect to see the quality of your priority feeds improve over the next 8 weeks.
A lot of Feedly Pro and Feedly Teams customer rely on power search to find specific articles in their feeds and boards. In Leo 0.6, we are expanding power search and let you search with your priority feeds.
For teams using Leo to discover and track trends, opportunities, and trends across industries, the combination of Leo priorities and Power search is a powerful way to quick find the most crucial information
We want to thank all the beta customers who have been working very closely with us over the last few weeks (and sometimes months). We are very grateful for your time and precious feedback. This open collaboration is not only powerful and efficient but it is also very fun. We look forward to the next 3 months!
Sometimes you want to follow high volume publications like The Verge, NY Times, or VentureBeat because you trust them but you are only interested in narrower topics, trends, or mentions.
Reducing noise and information overload is a problem we care passionately about. We have been working over the last 12 months on a new feature called Leo. You can think of Leo as your non-black-box research assistant – an easy-to-control AI tool which helps you reduce noise in your feeds and never miss important articles.
Here is a quick overview of the Leo 0.5 Beta feature set.
New Priority Tab
If you are part of the Leo 0.5 Beta Program, each of your feeds has now 2 tabs.
The All Tab includes all the articles published by the sources you follow.
The new Priority Tab includes the subset of articles flag by Leo as important – based on the priorities you defined for your Leo.
Three Core Prioritization Skills: Mentions, Topics, and Like Board
Leo 0.5 ships with three core skills: mentions, topics, and like-board. Each of these skills allow you to prioritize articles differently.
The mentions skill allows you to prioritize articles based on mentions of people, company or keywords which are important to you.
For example, you can ask Leo to prioritize all the articles that mention “JP Morgan”
The topic skill allow you to prioritize articles which are about a specific topic you are interested.
For example, you can ask Leo to analyze your tech feed and prioritize articles which are about artificial intelligence, quantum computing, or gaming.
Leo ships with one thousand pre-trained topics. If the topic you are interested in is part of that list, the topic skill is a powerful tool to let you focus your feed on what really matters to you.
Sometimes, the topic you are interested in a very niche. This is where the Like Board skill is very useful and powerful.
For example, if you are in the Sports industry, you might be interested in the emerging Smart Venue trend. Leo does not know out of the box about Smart Venue but if you can create a board and save 30-50 articles about Smart Venue, you can use the Like Board skill to teach your Leo a new personalized topic and ask Leo to prioritize future articles which are similar to the ones you save in that board.
Once you have defined the priorities of your Leo, he will continuously read your feed and flag articles which are aligned with those priorities.
The Like Board is particularly powerful because the more articles you save to that board, the more accurate Leo’s recommendation will become.
Finally, you can easily define more sophisticated priorities by combining multiple skills/layers.
Feedback Loop Via Less Like This
When Leo makes a back prioritization, you have the control to provide him feedback via the Less Like This button.
There are 5 different classes of feedback you can offer to your Leo:
The “Not About” feedback allows you to teach Leo that it matched the wrong keyword or topic. For example, you were interested in ICO (Initial Coin Offering) and Leo detected ICO (Internet Commissioner Office).
The “duplicated article” feedback allow you to flag articles which are on topic but you have already read about via a different source
The “I’m not interested in” feedback allow you to flag class of articles you are not interested about. For example, you might not be interested in market research type articles. If you can flag 10-20 articles as I am not interested in market research, Leo is going to learn and start prioritizing fewer market research articles.
Sometimes (specially for keyword alerts), you might get articles from sources you do not care about. The ‘mute domain’ feedback allows you to train your Leo to mute articles from those domains.
Finally, sometimes, the reason is more complex. The ‘Something else’ feedback offers you an easy way out.
Control and Transparency
A very important aspect of the Leo promise is that it is a fun, non-black-box AI you fully control and can easily collaborate with.
Transparent because each time Leo makes a prioritization, he will explain why the article was prioritized and give you the opportunity to refine that prioritization.
Control because you explicitly define all the priorities of your Leo and you can at anytime go in the Train Leo section and remove or refine a priority. No black box. No lag.
Goodbye Information Overload
Leo 0.1 Alpha customers saw 40-70% noise reduction on their feeds. More targeted feeds mean that you can save time while reducing the risk of missing important articles, or being the last to know about an important risk or market opportunity.
We look forward to seeing how your will be training your Leo!
It is frustrating to be skimming through your feeds and run into duplicate articles.
This happens for example when you have overlapping keyword alerts where two different keywords exist in the same article. It also happens when some sources publish the same articles into different RSS feeds. Finally, it happens a lot when a company issues a press release and other sources publish that press release with some minor changes.
Giving you the tools and control to tune your feeds is something we care passionately about. Today, we are excited to announce the beta release of a new Leo skill called Deduplication.
What is Deduplication?
This skill helps Leo detect that multiple articles are near exact duplicates of each other and cut that noise from your feeds. On the Web version of Feedly, you will see a small notification at the bottom right of your screen each time Leo removes duplicate from your feeds.
Which language does Deduplication work on?
The Leo deduplication skill works across all languages?
Which Feedly Plan does this skill require?
Because processing duplicates at scale is expensive, this skill will be initially rolled out as part of the Feedly Teams plan.
If you are part of Feedly Teams, there is a preference knob in the Leo settings page to disable this skill.
Beyond near exact duplicates
The deduplication skill is focusing on near exact duplicates. These are articles which have 85% or more overlap. We are working on a different skill called Business Events for articles which are reporting on the same event but with different content. In the case of business events, the content will be grouped instead of being removed.
Are you passionate about the Web, reading, and NLP? Are you curious about how machine learning can help process, filter, and prioritize information more effectively?
The goal of this tutorial is to show you how to leverage the content of your Feedly feeds and boards to run machine learning and NLP experiments using the Feedly API and Colaboratory.
To make things concrete, we are going to create a simple KNN classifier that takes as input the content of The Verge and Engadget, and a board with 50 positive AI articles. Using these inputs, the classifier reliably learns to classify AI and non-AI articles from those sources.
The Colaboratory Cloud Notebook we created has all the building blocks you need to create and run this AI experiment. All in a browser. All this within 20 minutes.
Once you are done with this first experiment, you will also have an example you can easily adapt to your own feeds, boards, and machine learning models!
People love RSS because it lets you to aggregate in one place all the topics and information you care about. No more zigzagging.
People also love RSS because of its control and transparency. It is not a feed that tries to manipulate and monetize your attention. It is a feed that feeds your mind and makes you smarter.
But sometimes RSS feeds can become noisy.
For example, you might follow broad sources like Forbes but only care about narrow topics like FinTech or climate change. Or you might have a keyword alert for General Electric but only care about product launches and partnerships. Or you might have a cybersecurity feed but want to focus your attention on the most critical software vulnerabilities first.
That noise can be overwhelming and make you waste time.
This is a problem we are very passionate about and have been focusing on over the last 18 months.
Today, we are excited to announce Leo, your AI research assistant.
We have been teaching Leo how to read so that he can help you declutter your feeds and dig deeper into the topics and trends you care about without losing control.
Leo continuously reads your feeds and short-lists the most relevant articles in the priority tab.
For example, you might have a broad business feed connected to HBR, Bloomberg, NY Times, etc. with thousands of new articles per month.
You can train Leo to read those 1,000+ articles and prioritize the 30 or so related to specific companies (Slack), topics (#leadership) or trends (Facebook and #crypto) you care about.
Leo is not a black box recommendation engine. Instead, Leo ships with a set of skills that gives you control over defining what information is important to you:
The Topic skill lets you prioritize mentions of topics, keywords, people, companies, products, etc.
The Like Board skill learns by example based on articles you’ve saved to boards and prioritizes similar articles
The Business Event skill lets you track product launches, funding events, partnerships, etc.
The Security Threat skill lets you prioritize articles related to critical software vulnerabilities and specific vendors.
The Mute Filter skill let you remove articles mentioning specific keywords and topics.
The Deduplication skill removes duplicate articles from feeds and keyword alerts
You can easily assemble these skills into Leo models and preview in real-time what kind of information Leo will prioritize.
Control and transparency are core Leo design principles.
All the articles prioritized by Leo have a green priority marker. Clicking on that marker offers an explanation of why the article was prioritized and the opportunity to refine, pause or remove that priority.
Leo learns from both positive and negative feedback:
When a recommendation is useful, you can save it to a board to send Leo a positive signal.
When a recommendation is not useful, you can use the “Less-Like-This” down arrow button to correct Leo.
Leo learns from your feedback and gets continuously smarter!
Leo is generally available to all Feedly Teams users and in early access to Pro+ users. If you have any questions or feedback regarding Leo, you are welcome to join the Feedly Lab Slack and connect with the dev team.
Connecting people to the best sources for the topics that matter to them has been core to our mission since the very start of Feedly.
But discovery is a hard problem. The web is organic, a reflection of the global community’s changing needs and priorities. There are millions of sources across thousands of topics and we all have a different appetite when it comes to feeding our minds.
About twelve months ago, we created a machine learning team to see if the latest progress in deep learning and natural language processing could help us crack this nut.
Today, we are excited to give you a preview of the result of that work with the release of the new discovery experience in the Feedly Lab app (Experience 06).
Two thousand topics
The first discovery challenge is to create a taxonomy of topics.
You can think of Feedly as a rich graph of people, topics, and sources. To build the right taxonomy, we started with the raw data on all of Feedly’s sources. We had to create a model to clean, enrich, and organize that data into a hierarchy of topics. Learn more about the data science behind this.
The result is a rich, interconnected network of two thousand English topics. And it’s mapped well with how people expect to explore and read on the Web.
On the discovery homepage, we showcase thirty topics based on popular industries, trends, skills, or passions. You can access all of the topics in Feedly via the search box.
The fifty most interesting sources
The second discovery challenge is to find the fifty most interesting sources someone researching any topic might want to follow.
Ranking sources is hard because not all sources are equal. In tech as an example, you have mainstream publications like The Verge or TechCrunch, expert voices like Ben Thompson, and lots of B-list noisy sources which don’t add much value.
In addition, for niche topics like virtual reality, some sources are specific to VR while others cover a range of related topics.
To solve this challenge, we created a model which looks at sources through three different lenses:
relevance (how focused is the source on the given topic)
engagement (a proxy for quality and attention)
The outcome is new search result cards. You can explore the fifty most interesting sources for a given topic and sort them using the lens that is most important to you.
One of the benefits of the new topic model is that the 2,000 topics are organized in a hierarchy. This makes it easy for you to zoom in or out and explore many different neighborhoods of the Web.
For example, from the cybersecurity topic, you can jump to a list of related topics that let you dig deeper into malware, forensics, or privacy.
One more thing…
We have done a lot of research over the last four years to understand how people discover new sources. One insight we learned is that people often co-read certain sources. For example, if you are interested in art, design, and pop culture and you follow Fubiz, there is a high chance that you also follow Designboom.
With that in mind, we spent some time creating a model that learns what sources are often co-read. The idea is that a user could enter a source that they love and discover another source they could pair it with.
As a user, you can access this feature by searching in the discover page for a source you love to read. The result will be a list of sources which are often co-read with that source.
I would like to thank Paul, Michelle, Mathieu, and Aymeric for the great research work they did to take this project from zero to one. People who have tried to tackle discovery know that it is a very hard challenge and the results of this project have been very impressive.
We would also like to thank the community for participating in the Battle of the Sources experiment. Your input was key in helping us learn how to model the source ranking. We are going to continue to invest in discovery and we look forward to continuing to collaborate with you.
We would also like to thank Dan Newman, Daron Brewood, Enrico, Joey, Lior, Paul Adams, Ryan Murphy, and Joseph Thornley from the Lab for reviewing an earlier version of this article.