Fixing AI Made Easy: RagaAI Comes Out of Stealth with an Automated Testing Solution

As THE request For AI keep on going has to grow, A new category of tools East portion development And deployment join THE scene. Case In indicate: RagaAI, A Based in California to start up offer A platform has test And fix AI, Today emerged Since stealth with $4.7 million seed funding Since pi Businesses. Global investors Anorak Companies, TenOneTen Companies, Arka Companies, mana Businesses And Exfinite Adventure The partners Also participated In THE round.

Based by ancient Nvidia executive Gaurav Agarwal, RagaAI will to use THE capital has advance research And to strenghten It is automatic essay platform has establish A robust frame For on And reliable AI.

"Guide by OUR heart values, We are engaged has push THE borders of automatic AI issue detection, automatic root cause analysis And fixation THE problems, stay has THE foreground of avant-garde methods ", Agarwal said In A statement. He note THE business East Already portion Fortune 500 companies has address problems such as bias, precision And hallucinations In different to use case.

Building And deployment AI In production East not A to walk In THE park. Teams to have has gather data, form THE models And SO be vigilant about how they work In production has see if they are deliver what is this expected — Or turning disabled track In unexplored territories. A little gap here Or there And THE entire effort come to crash down, leading has high costs And lack opportunities.

Agarwal saw This issue first hand When functioning with Nvidia And Indian mobility business Oh there. He decided has tackle he with A automatic essay platform that could detect AI problems, diagnose them And fix them on THE fly. This directed him has build RagaAI. But, here is THE interesting part: THE platform do not check For A little dozen problems. He door out as a lot as 300 tests, covering all kinds of problems that can lead A AI model has fail, RIGHT Since data And model problems has operational shortcomings.

Once THE platform identifies A issue, he help users triage THE issue has It is root cause. This can be as varied as bias In THE training data, poor labeling, data derivative, poor hyperparameter optimization while training Or A lack of model robustness has contradictory attacks. SO, as THE last stage, he provides exploitable recommendations has fix THE issue, as portion teams withdraw poorly labeled data points In A Click on Or suggesting reconversion THE model has fix problems with data And concept derivative.

HAS It is heart lies RagaDNA foundation models that generate high quality integration – representations of data In A compressed And significant format. Most tests on THE platform to use these integrations as A base For issue detection, diagnostic And remediation.

“RagaAI DNA represented vertical specific fundamental models which are custom qualified For essay purposes. This allow RagaAI has automatically add intelligence has THE essay workflow as definition THE Operational Design Domain (ODD), identify edge case Or has model carried out poorly Or correlative he with missing Or bad quality training data," Jigar Gupta, THE head of product has RagaAI, writing In A Blog post.

While THE essay platform has just spear publicly, RagaAI complaints that several Fortune 500 companies are Already using THE technology, including AI first companies such as LightMetrics And Saturday, of course. In A case of Implementation, A e-commerce business was able has identify hallucinations And reduce Errors In It is chatbot. In another, A automobile business was able has improve THE precision of It is model aiming has detection Vehicles In low light scenarios.