Anthro.ai is a collection of anthropologists, mathematicians, data-scientists and market specialists building ethical, machine-based systems and maps to navigate complex human landscapes.
Ask us a question about an audience you want to understand, or a policy landscape you want to map, or an issue you want to explore. We give you answers backed by insights about why something is happening, what’s driving behaviour, and we attempt to figure out what you can do about it.
We do this at scale, and at blinding speed.
Frequently Asked Questions
What is the mission of Anthro.ai?Our mission is to understand the world better and help people navigate complexity better. The accelerated pace of change being caused by new technologies and complex political circumstances demands data-fueled, ethically-driven analyses of the world around us. We believe empathy and understanding makes the world a better place. As a result, we use cutting-edge, purpose-built technology to work with diverse actors across the scale of social, political, and economic transformation. Instead of deploying these tools towards garnering power or wealth, we analyse powerful data in order to develop localized insights into issues of pressing public concern.
What is Anthro.ai’s USP?We answer really hard questions. Ask us a question about an audience you want to understand, or a policy landscape you want to map, or an issue you want to explore. We give you answers — why something is happening and what you can do about it. We tailor the granularity of our response - do you need extensive ethnographic field notes or a quick answer to a vital question? We can do this at scale and at blinding speed.
What data sources do you use?We construct signals composed of a carefully curated set of underlying data sources. Our data sources can include all forms of publicly available data sets — from published news (in nearly any language), to demographic and census data, to historical data to social media posts. We can access paid databases, build custom access points for data on the Internet and run wide-reaching active-collection measures (surveys, for example). We can focus on specific periods of time, and / or, specific regions, across multiple languages. We meet local data-collection, analysis, transfer and retention regulatory requirements in all cases and usually go a step further by anonymising, avoiding scraping, and meeting restrictions when using APIs.
What is your methodology?We build custom data signals. Then, our machines read these the custom data signals to unearth patterns, clusters, and potential co-relations. Our analysts look at these to not just understand the issue, but extract deep insights about the whys and the wherefores. We use custom insights delivery frameworks — for example, a Hopes-Fears-Aspirations-Resonance (HFAR) framework, to try and understand what drives attitudes, choice and behaviour.
What are the examples of your previous work?This is the short version of the report we wrote for Europe Delivers — a platform for collaborative action led by Xynteo. It is a great example of the kind of work we do — answering truly complex questions about abstract concepts like the social contract between citizens and the state.
Can I use your Election Insights?Yes, you can, as long as you give us credit. Drop us a line at up.elections [at] Anthro (dot) ai and we will set you up.
Why are you doing this?For the scale and size it offers. When we decided to stress test our tools and methodology, we realised that nothing in the world comes close to the Indian elections. And Uttar Pradesh is THE bell-weather state. There is a wide consensus that the road to Delhi goes through UP. When we decide to test capability, it was clear that the litmus test for our tech and our methods lay in deploying and testing in UP.
What data are you analysing? And what is the scale of the data you are mining for insights?We started setting up custom data signals in September of last year. This cut across news, blog posts, tweets, public instant-messaging groups, photos being posted across the state, past electoral data, census data, etc. We have analysed nearly 150,000 booth level records across multiple vectors. We are analysing between 300,000 and 500,000 geo-located tweets at the constituency level, and sampling another quarter of a million tweets every day to analyse softer vectors like personality, attitudes etc. We ingest and analyse upwards of 500 news articles a day from a curated list of sources. We are constantly tracking shifts in share and opinion about key political figures and issues. All of this is being plotted across different types of maps — many of which we hope to publish over the coming weeks.
Who is funding you?We are self funded.
Are you using off-the shelf technology or in-house developed tech?We are building our tech on the shoulders of giants. We owe a lot to academicians who are exploring the frontiers of human behaviour, machine learning, and political science. We use commercially available APIs from Amazon, Google, Microsoft, and IBM. We use a host of data-sources (many paid for) including Webhose, EventRegistry and Wikipedia. We have built custom data signals, analysers, pattern engines, visualisations, and little widgets. We have tested different methods to analyse the data and have arrived at a model that is proving to be fairly successful in predicting what the narrative will be, what resonates, and how we see vote shares shifting. All of this at blinding fast speed — we can update projections in a matter of hours (not days). Of course, the ultimate test for our predictions will be on May 23rd when the election results come out.
What kind of models/algorithms do you use?We use a variety of models for carrying out our analysis — t-SNE, Principal Component Analysis, Google NLP, IBM Watson Personality Insights, Vader Sentiment Analysis. It is important to note that Anthro.ai is more focused on finding the right tool for the job and the analysis produced by these models rather than the models themselves. We see ourselves as bakers rather than oven makers. Having said that we do have both mathematicians and programmers in-house who understand the models we implement and how a particular model is implemented.
Are you a for profit or non-profit organization?We are a for-profit organization. However, our clients and prospective clients belong to both for-profit and non-profit sectors.
Do you have a website that I can check out?This is it, we will continue to build it out.
Can you tell me a little bit about your team.We have three founders: Narendra Nag (Naren), Raheel Khursheed and Tirthraj Nath Singh (TR). The election project is being led by Narendra while Raheel and TR focus on a few paid projects we have on. We encourage you to check out our profiles on Linkedin. In addition, we also work closely with leading experts in the field that we are exploring in order to provide our clients with the deepest and most comprehensive insights and understanding possible.
Are you focused on India or do you look at projects anywhere in the world?We are an international organization with a global footprint and as such expect projects to be distributed all over the world.
Can you do a survey/focus group for a project?Yes we have full capability to carry out such research if a project calls for it. Typically, our research process negates the need for in-person surveys but if we find that added insights can be found by carrying out surveys then we will conduct them, either directly or by partnering with reputable research organizations.
Do you comply with GDPR and other privacy laws across the world for any data that you may analyse during the course of a project?Yes we have set up our systems to be fully compliant with local laws and regulations. Data and privacy laws are a critical part of our project design and we work closely with our clients to ensure that not only are we in compliance with local regulations but also with our clients’ standards.
Where are you located?Currently incorporated in Singapore — but we are in the process of changing the nature of our corporation to make it easier for us to operate in different geographies (and potentially raise money). Two of our founders were based out of Europe till very recently, and one has been splitting his time between the US and India.
Can you help my organization?We would love to explore a commercial partnership with you. Please reach out to us at Humans [at] Anthro (dot) ai.