After Six Years Spell Wires AI Into MLOps

For machine learning teams at companies like Instacart and Snap, the most expensive part of the job isn't the model architecture. It's the time spent wrangling cloud infrastructure, tracking thousands of failed experiments, and ensuring a colleague can reproduce your results six months later. Spell, a New York-based MLOps platform founded in 2018, has built its business on automating that drudgery. Its core bet is that data scientists should spend their cycles on science, not sysadmin work, a proposition that has attracted over $60 million from investors like Khosla Ventures and Insight Partners [TechCrunch, Mar 2021].

The wedge of reproducibility

Spell's initial product wedge was straightforward: simplify GPU-accelerated machine learning workflows. In a landscape dominated by AWS SageMaker's sprawling console, Spell offered a managed service that handled provisioning, scaling, and collaboration from a single interface [TechCrunch, Jul 2018]. The platform's early differentiator was a focus on reproducibility, automatically versioning code, data, and environment settings for every training run. This addressed a chronic pain point in research and development, where months of work could become unrepeatable due to undocumented dependencies or configuration drift. The company's Google-alumni founders, CEO Vishal Kapadia and CTO Suryan Sriram, leveraged their distributed systems expertise to build this layer of abstraction [Crunchbase].

A pivot toward automation

While infrastructure management remains a core offering, Spell's strategic direction has shifted toward higher-level automation. The launch of SpellML 2.0 in March 2025 introduced what the company calls "agentic workflows" and fine-tuning APIs [VentureBeat, Mar 2025]. In practice, this means the platform can now orchestrate multi-step ML pipelines,like data preprocessing, model training, evaluation, and deployment,with less manual intervention. This move aligns with a broader industry trend where MLOps tools are evolving from pure infrastructure platforms into intelligent co-pilots for the ML lifecycle. The company has also cultivated hardware partnerships, including one with NVIDIA for GPU optimization and another with Graphcore for next-generation AI infrastructure, suggesting a focus on performance at scale [TechCrunch, Mar 2021][HPCwire, Feb 2022].

Traction and the talent signal

Spell's public customer roster includes tech-forward names like Instacart, Snap, and augmented reality pioneer Niantic [Spell]. These are organizations with substantial, ongoing ML workloads where reproducibility and team collaboration are non-negotiable. The company's last funding round was a $40 million Series B in March 2021, led by Insight Partners at a post-money valuation estimated at $250 million [The Information, Apr 2021]. While there have been no subsequent funding announcements, active hiring provides a signal of continued operational momentum. As of April 2026, the careers page lists openings for an ML Infrastructure Engineer, a Product Manager for MLOps, and a Sales Engineer, indicating investment in both core platform development and go-to-market expansion [Spell, April 2026].

Round	Date	Amount	Lead Investor
Seed	July 2018	$4 million	Khosla Ventures
Series A	October 2019	$16 million	Gradient Ventures
Series B	March 2021	$40 million	Insight Partners
Table: Spell's disclosed funding history. Total raised stands at approximately $60 million [TechCrunch].

The crowded field ahead

The strategic risks for Spell are not subtle. The MLOps platform space is intensely competitive, with well-capitalized incumbents and a swarm of startups. The company must contend with:

• The cloud hyperscalers. AWS SageMaker is the default for many teams already embedded in the AWS ecosystem, offering deep integration but often criticized for complexity.
• Specialized rivals. Companies like Weights & Biases have built strong reputations on experiment tracking and visualization, directly overlapping with Spell's collaboration features.
• The open-source stack. Composable tools like MLflow, Kubeflow, and Ray allow large enterprises to build custom, albeit labor-intensive, MLOps platforms.

Spell's rebuttal rests on integration and ease of use,presenting a cohesive, opinionated platform that reduces cognitive load. Its recent push into agentic workflows is an attempt to move up the value chain, competing on intelligence rather than just infrastructure. However, the absence of public revenue or customer growth metrics since its 2021 Series B makes it difficult to assess the commercial success of this bet. The company must demonstrate that its automation features drive tangible efficiency gains compelling enough to justify switching costs in a budget-conscious environment.

For the data science teams Spell serves, the current standard of care is often a fragmented patchwork. It might involve manually scripting jobs on cloud VMs, struggling with shared notebook environments that lack version control, or gluing together disparate open-source tools. This leads to wasted GPU hours, irreproducible results, and significant friction in team collaboration. Spell's entire proposition is to be the integrated system that replaces that chaos. The next twelve months will be critical in showing whether its automation-focused SpellML 2.0 can convert its reputable early adopters into a broader, sustainable enterprise business. The hiring of a Sales Engineer suggests the go-to-market motion is now a top priority.

Sources

1. [TechCrunch, Jul 2018] Spell raises $4M seed from Khosla to take on cloud ML training giants | https://techcrunch.com/2018/07/18/spell-seed-khosla/
2. [TechCrunch, Oct 2019] Spell raises $16M Series A from Google's Gradient to build ML dev platform | https://techcrunch.com/2019/10/16/spell-mlops-series-a/
3. [TechCrunch, Mar 2021] Spell nabs $40M Series B led by Insight Partners to grow ML ops platform | https://techcrunch.com/2021/03/10/spell-series-b/
4. [The Information, Apr 2021] ML startup Spell raises $40M at $250M valuation | https://www.theinformation.com/articles/ml-startup-spell-raises-40m-at-250m-valuation
5. [VentureBeat, Mar 2025] Spell launches ML 2.0 with agentic features | https://venturebeat.com/ai/spell-launches-ml-2-0-agentic-features-2025/
6. [Crunchbase] Spell organization profile | https://www.crunchbase.com/organization/spell
7. [Spell] Spell customers page | https://spell.ml/customers
8. [Spell, April 2026] Spell careers page | https://spell.ml/careers
9. [HPCwire, Feb 2022] Spell partners with Graphcore to deliver next generation AI infrastructure | https://www.hpcwire.com/aiwire/2022/02/08/spell-partners-with-graphcore-to-deliver-next-generation-ai-infrastructure/