Sourav Howlader
Chairman, ProflakesOver 15 years of technical experience, including Marlabs and Samsung R&D. He shapes the technology strategy of the companies he leads.
Full profile of Sourav Howlader (opens in a new tab)We build artificial intelligence for the problems people meet every day, in the way they live, work and connect. Proflakes has done this work since 2014, with one rule we keep: AI should be developed ethically, and serve the people it affects. GEOFLAKES is that work applied to search, so that a business is described accurately when people ask an AI assistant about it.
We build AI that reads the internet carefully and judges it fairly.
So that when an AI tells you about a person or a business, it is right, and you can check where it got that from.
As featured in
YourStoryMoneycontrolANIIIFLThe Economic TimesThe TelegraphYahoo NewsProflakes started in 2014 with one aim: use artificial intelligence to solve problems people meet every day.
Most of our work since has been teaching machines to read what people put online, and to judge it fairly. Products such as ContentAnalyzer, NextID and ScanScam came out of it. Building them taught us how these systems read a page, what they trust, and where they get it wrong.
That is the experience GEOFLAKES is built on. People now ask AI assistants which business to choose, and those assistants answer from what they can read on a website. GEOFLAKES shows a business what they read, and what to change so they describe it accurately.
Over 15 years of technical experience, including Marlabs and Samsung R&D. He shapes the technology strategy of the companies he leads.
Full profile of Sourav Howlader (opens in a new tab)
An entrepreneur and business leader with a background in IT services and digital communications. Earlier in his career he worked with IBM, Unisys, Reuters and ITC Infotech.
Full profile of Suman Howlader (opens in a new tab)Four commitments that decide how we build GEOFLAKES, and what we will not do.
We would rather show you what an engine actually answered than predict what it might. Where we could not measure something, we leave it out of the score.
When we improve a file you already wrote, we edit it and tell you what changed. We never swap your text for ours or put our name in it.
If a site refused our requests, or a plan leaves something out, the report says so in plain words. A gap you cannot see is worse than one you can.
Our crawler gives its name in every request, follows robots.txt and slows down when asked. It never tries to get past a bot check or a sign-in.
An audit shows what AI answer engines can reach on your site, which pages they would quote, and what to fix first.