Q: How does SAM work? Does it actually use AI?

A: SAM uses AI and social media data to detect anomalies and world events in real-time, informing users of events happening around the world, when they happen. SAM has and continues to build our own proprietary AI stack (with NLP, ML and more). You can read our blog announcement which addresses some of the AI challenges we tackle.

Q: How does your beta program work?

A: Simply request early access and your details will be added to the beta access queue in the order they were received (unless you are a current SAM client - those requests are prioritised). SAM will be releasing access in stages over the next several weeks and will only be accepting early beta request until May 1st, 2018.

Q: How does SAM work with clients/beta users? Do you share your roadmap?

A: SAM works with large and small clients around the world. We strive to understand your unique requirements whether you’re a city, airline, or newsroom. Our sales, data and operations teams are available for consultation specific to your use case and can offer more details about how roadmap aligns with your current and future business objectives.

Q: Can I flag mistakes the machine makes?

A: Yes! Our beta users are helping us shape this product by flagging mistakes. This community of contributors helps our team address issues quickly and improves the SAM platform at a quicker pace.

Q: What is the rate of error?

A: Currently we’re nearing 90% accuracy on properly classifying social posts as correct event type (fire, bomb scare, explosion, etc). Each event type gets stronger as it gathers more data over time. New event types may be weaker at first, but will improve as we continue to train the category.

Q: Why do I see Tweets on Tweetdeck first sometimes?

A: Tweetdeck is unverified and will allow you to see single, unconfirmed reports of individual incidents. SAM AI tries to verify or debunk an incident—which might require the system to look for additional corroborating posts before you see it in the SAM dashboard.

Q: How does SAM know the location of an incident? Why is it more precise sometimes than others?

A: Our AI extracts key location descriptions from multiple posts to try and triangulate the location of a particular incident. Sometimes when breaking events occur, limited information is available, making it hard to get an exact location at the time of alerting.

Q: How does the Most Coverage filter calculate event coverage?

A: SAM calculates the total number of social media posts talking about a specific incident, while taking into account the number followers that the user has. The formal scores is based on their social reach, informing us how much coverage a particular incident is getting.

Q: What types of events does SAM currently detect?

A: SAM is currently detecting a wide range of events from power outages to terror attacks. When you log into the beta you’ll have full control over what events you want to see (or not see). This list is always growing and our beta users are helping us to define what events they want to be alerted to.

Q: Is anything happening with my current SAM account?

A: All current SAM users will not be affected by this new update. We anticipate a lot of current SAM users will want to get access to the new beta—all early access requests from current customers will be prioritised in the queue!