Datasaur announced the launch of Forge, an AI Native Service that embeds the company’s engineers within regulated enterprises to design, build, and operate AI systems that run entirely inside the customer’s own cloud or on‑premises environment. The offering targets financial services, healthcare, insurance, legal and government organizations that must keep data on‑premise for compliance and data‑sovereignty reasons.
What Happened
Datasaur introduced Forge as a new service line that places Datasaur engineers inside client organizations to deliver AI solutions that never leave the customer’s environment. The service is positioned as “privacy‑first by construction and model‑agnostic by design,” allowing clients to select any frontier or open‑weight model and run it on Datasaur’s proprietary agent harness. The announcement follows recent enterprise services launches by OpenAI and Anthropic, which each focus on deploying a single proprietary model for customers.
Product and Platform Context
Forge is built around three explicit commitments:
- In‑environment deployment – All inference, fine‑tuning, and evaluation occur inside the customer’s cloud or on‑premises infrastructure; no data is sent to Datasaur or to external model vendors.
- Model agnosticism – Clients may run frontier APIs, open‑weight models such as Google’s Gemma, OpenAI OSS, DeepSeek, or custom fine‑tuned small language models (SLMs) hosted on their own GPUs. The model is treated as a swappable input, while Datasaur’s orchestration layer remains the durable asset.
- Customer ownership – At contract end, embeddings, fine‑tuned models, evaluation benchmarks and the training data transfer to the client. Only Datasaur’s internal data engine, which generates those artifacts, stays proprietary.
Datasaur frames its agents as infrastructure rather than end‑user tools, arguing that infrastructure adoption is centrally controlled and applies to every relevant record by default. The service will be staffed by Datasaur’s existing solutions‑engineering and AI‑delivery teams. Engagements start with a paid scoping phase and aim for production deployment within 30 to 60 days.
Why It Matters for Enterprise Buyers
Regulated sectors often cannot transmit sensitive records to third‑party model endpoints, a limitation that traditional SaaS AI offerings do not address. Forge’s in‑environment approach satisfies data‑residency mandates and mitigates concerns that frontier model vendors could inadvertently train on proprietary data.
The model‑agnostic stance also protects buyers from vendor lock‑in. As the AI frontier evolves, enterprises can swap models without re‑architecting the surrounding orchestration layer. Ownership of the final AI assets—embeddings, fine‑tuned models, and benchmarks—provides a clear exit path and aligns with procurement policies that favor capital‑expenditure models over recurring usage fees.
Datasaur cites existing deployments with a leading global systemically important bank (GSIB), multiple federal agencies, Am Law 100 firms, and other Fortune 500 organizations. Reported use cases include PII redaction across hundreds of millions of records, legacy‑system automation for federal compliance deadlines, and legal‑document review under strict data‑residency requirements.
Market Signal
Forge’s launch signals a differentiated approach to enterprise AI services: rather than selling inference‑as‑a‑service tied to a single model, Datasaur offers a “buy‑side” model where the AI system is owned and operated by the client. Founder and CEO Ivan Lee contrasted this with the “rent‑side” economics of OpenAI’s and Anthropic’s consulting arms, which depend on ongoing inference fees. By positioning itself on the buy side, Datasaur aims to attract regulated enterprises that prefer capital‑based ownership and full control over AI workloads.
Key Takeaways
- Forge embeds Datasaur engineers inside regulated enterprises to deliver AI that runs entirely within the customer’s own cloud or on‑premises environment.
- The service is model‑agnostic, supporting frontier APIs, open‑weight models such as Google’s Gemma, OpenAI OSS, DeepSeek, or custom fine‑tuned SLMs hosted on the client’s GPUs.
- Deployments are completed in 30‑60 days after a paid scoping phase, and all AI assets (embeddings, fine‑tuned models, benchmarks, training data) transfer to the client at contract end.
TechInsyte's Take
Forge addresses a concrete compliance gap for highly regulated firms that cannot outsource inference. Its model‑agnostic, ownership‑focused design may appeal to CIOs seeking to avoid vendor lock‑in and recurring inference fees. However, the service’s success will depend on the maturity of Datasaur’s orchestration platform and the ability to scale engineering resources across multiple large clients. Buyers should monitor early deployment outcomes and assess whether the promised 30‑ to 60‑day rollout aligns with internal change‑management timelines.
Source: accessnewswire
