Refuel
Refuel.ai, founded in 2021 and headquartered in San Francisco, built a platform that uses large language models to automate data labeling and enrichment for enterprise machine learning workflows.
Profile
Refuel.ai provides a platform that uses large language models to automate data labeling and enrichment for enterprise machine learning applications.
Refuel.ai, founded in 2021 and headquartered in San Francisco, built a platform that uses large language models to automate data labeling and enrichment for enterprise machine learning workflows. The company claimed to reduce engineering time per project by three months and increase labeling accuracy by 30%, citing a case study with client TeachFX. In May 2025, Refuel was acquired by Together AI, a competitor in the AI infrastructure space, and now operates as a subsidiary.
The acquisition price was not disclosed. Prior to the acquisition, Refuel had raised venture capital, though the total amount and specific rounds are not publicly detailed. The company listed over a dozen logos on its website as customers, including Birdeye, Affinity Solutions, Chattermill, Harmonic, Fuel Cycle, Percent, Path Stream, Enigma, TeachFX, RudderStack, AIMon, Middesk, Sense, Beni, Boltwise, GoodFit, and Volza.
However, no revenue figures or headcount numbers beyond PitchBook's estimate of five employees were available. The acquisition by Together AI, which itself raised significant funding in 2024 and 2025, effectively ended Refuel's independent operations. The company's product was designed to handle unstructured data tasks such as classification, extraction, and enrichment, positioning it in the crowded AI data preparation market alongside players like Scale AI and Labelbox. The acquisition likely reflects Together AI's strategy to expand its tooling and customer base, though Refuel's small team and niche focus may limit its long-term impact within the larger organization.
Who buys this
- Enterprise data teams needing to label unstructured data for ML models
- Companies building AI applications that require high-quality training data
- Organizations looking to automate data enrichment for business intelligence
- SaaS companies integrating AI features into their products
Publicly disclosed clients
- Birdeye
- Affinity Solutions
- Chattermill
- Harmonic
- Fuel Cycle
- Percent
- Path Stream
- Enigma
- TeachFX
- RudderStack
- AIMon
- Middesk
- Sense
- Beni
- Boltwise
- GoodFit
- Volza
Strengths and what to watch
Strengths
- Claims significant time savings (3 months per project) and accuracy improvements (30% increase) for data labeling tasks, as evidenced by a case study with TeachFX.
- Acquired by Together AI in May 2025, providing access to a larger infrastructure platform and customer base, potentially accelerating product adoption.
- Listed a broad set of customer logos across industries including SaaS, market research, and education, indicating some market validation.
Watch for
- Acquired by Together AI in May 2025; the company is now a subsidiary, and its independent product roadmap and team may be absorbed or deprioritized.
- Very small team (estimated 5 employees per PitchBook) and no disclosed revenue, suggesting limited scale and resources prior to acquisition.
- No publicly available funding details beyond being 'formerly VC-backed,' raising questions about the company's financial trajectory and the terms of the acquisition.
Key Information
- Industry
- Data Generation & Labelling
- Founded
- 2021
- Headquarters
- San Francisco
Frequently Asked Questions
What is Refuel.ai and what does it do?
Refuel.ai is a platform that uses large language models to automate data labeling and enrichment for enterprise machine learning applications. Founded in 2021 and based in San Francisco, it helps teams handle unstructured data tasks like classification, extraction, and enrichment.
How does Refuel.ai automate data labeling?
Refuel.ai leverages large language models to label and enrich unstructured data automatically, reducing manual effort. The company claims this approach cuts engineering time by three months per project and boosts labeling accuracy by 30%, based on a case study with client TeachFX.
Who acquired Refuel.ai and when?
Refuel.ai was acquired by Together AI, a competitor in the AI infrastructure space, in May 2025. The acquisition price was not disclosed. Refuel now operates as a subsidiary of Together AI, ending its independent operations and likely integrating into a larger platform.
What are the main benefits of using Refuel.ai for data labeling?
Refuel.ai claims significant time savings, reducing engineering time per project by three months, and a 30% increase in labeling accuracy. These benefits come from using large language models to automate tasks, as demonstrated in a case study with TeachFX.
Who are Refuel.ai's typical customers and notable clients?
Refuel.ai serves enterprise data teams, companies building AI applications, and organizations automating data enrichment. Notable clients include Birdeye, Affinity Solutions, Chattermill, Harmonic, TeachFX, RudderStack, and Sense, among over a dozen listed logos on its website.
Is Refuel.ai still operating independently after the acquisition?
No, Refuel.ai was acquired by Together AI in May 2025 and now operates as a subsidiary. Its independent product roadmap and small team of about five employees may be absorbed or deprioritized within the larger organization, limiting its standalone future.
Sources
- www.refuel.ai — Company description, customer logos, and product claims including 3 months saved and 30% accuracy increase.
- www.refuel.ai — Announcement of acquisition by Together AI in May 2025.
- pitchbook.com — Company founding year (2021), employee count (5), status as acquired/merged, and former VC-backed status.
- techcrunch.com — Context on AI funding landscape in 2026, though Refuel is not listed; provides market backdrop.