Scaler Cofounder Anshuman Singh Unveils AI Training Data Venture Arcfront

Tech upskilling platform Scaler’s cofounder Anshuman Singh has unveiled Arcfront, an AI research and training data venture that turns records of how professionals solve problems into material for training and testing advanced AI models.
Singh is building the venture with Naman Bhalla, Scaler’s former head of AI products and an ex-Google software engineer. In a post on X, Singh said Arcfront also has “deep support” from Scaler cofounder Abhimanyu Saxena.
Singh claimed that four of the five frontier AI labs became Arcfront’s customers within less than a year of operations. However, he did not identify the labs.
Arcfront operates under Evaratus Inc., according to its website’s legal disclosures. Company recruitment material identifies Evaratus as formerly Scaler AI Labs, Scaler’s frontier AI research and engineering division.
Scaler’s privacy policy also lists Scaler AI Labs Inc. among its group companies. However, the public disclosures reviewed do not establish Evaratus’ current shareholding or whether the business has been separated from the Scaler group.
Inc42 has reached out to Arcfront’s founders for clarity on its ownership structure and relationship with Scaler. The story will be updated based on their response.
Singh’s LinkedIn profile lists November 2025 as the end of his tenure at Scaler and the start of his work on Arcfront. Bhalla’s profile indicates that he began working on the venture around the same time. However, it remains unclear whether Singh retains any operational responsibilities at Scaler or how the transition relates to Scaler AI Labs’ subsequent identity as Evaratus.
Turning Expertise Into Data
Arcfront’s proposition centres on the gap between AI models’ ability to solve structured problems and their performance in workplace situations requiring context, judgement, and decisions involving competing trade-offs.
Singh argued that models capable of solving difficult coding and mathematical problems can still struggle when placed in professional roles such as finance management.
Arcfront aims to address this by working with organisations to reconstruct how experts reach decisions, including the options they reject, revisions they make, and eventual outcomes.
The startup calls this approach ‘Decision Arc Data’. It converts these records into training data, reinforcement learning environments, and evaluations for frontier AI labs.
Reinforcement learning environments allow models to practise tasks and receive feedback on their results. Its work covers software engineering, site reliability engineering, cybersecurity, and collaborative workplace tasks. It also builds sector-specific evaluations using workflows from healthcare, life sciences, legal services, and finance to identify where models fail.
Separately, angel investor and former Google, Facebook, and DoorDash executive Gokul Rajaram identified Arcfront as the company behind his earlier claim about a startup reaching a $1 Bn annualised revenue run rate faster than any other company.
In the earlier post, Rajaram said the unnamed company had no revenue in November 2025 and generated $80 Mn in September 2026, putting it on course to reach a $1 Bn annualised run rate within 11 months of starting operations.
The AI Data Opportunity
Arcfront joins a growing group of startups building training datasets, testing tools, and simulated work environments for AI developers.
For instance, Deccan AI raised $25 Mn in March to expand its post-training data offerings, research, and enterprise infrastructure. Its products include reinforcement learning environments that recreate enterprise workflows and failure scenarios, allowing AI agents to practise before deployment in business operations.
A related opportunity is emerging in physical AI, where startups collect recordings of human activity to train robots. Startups including Human Archive are building data pipelines using footage of workers performing physical tasks.
Earlier this week, Humyn Labs raised $47 Mn, comprising $26 Mn in equity and $21 Mn in venture debt, to expand its data collection and robot deployment operations.
The opportunity has also drawn founders from India’s consumer internet ecosystem. Pronto founder Anjali Sardana recently launched San Francisco-based Provision Labs to collect data for training and testing physical AI systems.
The post Scaler Cofounder Anshuman Singh Unveils AI Training Data Venture Arcfront appeared first on Inc42 Media.
This content is published by Inc42 and has not been created or edited by Hello Headlines. View original




