Vertical AI Startups Move Into Engineering and Scientific R&D
Vertical AI startups for engineering and scientific R&D are companies that build domain-specific artificial intelligence systems to automate narrow but complex technical workflows, such as physics simulation, formulation design, and laboratory decision-making, rather than offering general-purpose chatbots or productivity tools. In the past year, investors have backed several such companies that focus on AI for engineering and R&D automation software, signaling a clear shift toward specialized tools that plug into real-world technical processes. Three early examples stand out: Inherent, a new AI lab developing a system for AI-native science; Mafer AI, which is building an AI operating system for formulation research tools; and NP Company, which is developing physics simulation AI for industrial engineering teams. Together they have attracted more than $60 million in early funding, suggesting growing belief that deep, vertical AI can change how technical work gets done.
Inherent Wants to Rewrite the Scientific Method with Faraday
Inherent has emerged from stealth with a $50m seed round co-led by Index Ventures and Radical Ventures, aiming to “write the playbook for AI-native science”. The lab is building Faraday, an AI system designed to let humans and self-improving AI collaborate on hard scientific problems, rather than bolting AI onto existing workflows. Several co-founders previously worked at DeepMind, Reka AI, Microsoft, and in AI policy, giving the team a mix of research and governance experience. According to Index’s Danny Rimmer, Faraday is “a system designed to help humans and self-improving AI work together on genuine scientific discovery — not AI plugged into the same methods we've used for 400 years, but a reimagining of the scientific method from first principles”. While Inherent has not yet detailed concrete products, its focus on AI for engineering and discovery puts it squarely in the vertical AI startups camp.
Mafer AI Targets Formulation R&D Bottlenecks with MaferOS
Mafer AI has raised a €2 million pre-seed round to build MaferOS, an AI-native operating system for R&D teams in formulation industries such as specialty chemicals, food and beverages, cosmetics, personal care, and fragrances. The company argues that decades of experimental data, failed formulas, lab analyses, and regulatory files form a “silent asset” that most formulation R&D teams cannot use at scale. MaferOS combines proprietary models trained on each client’s historical data with structured data layers and specialised AI agents to automate tasks like lab analysis, data structuring, regulatory checks, and formula recommendation. The platform is designed as full-stack R&D automation software, echoing enterprise models seen in other sectors, including forward-deployed engineers embedded with customers. This approach positions Mafer AI as a provider of highly targeted formulation research tools that aim to shorten innovation cycles, cut manual regulatory work, and reduce knowledge loss when senior experts leave.
NP Company Builds Physics-Based Transformers for Industrial Simulation
NP Company (NP Co.) is developing physics simulation AI using transformer architectures pre-trained on industrial physics data. The company has secured a €6 million pre-seed round led by Partech, with participation from the Peugeot family office and angel backers including Mistral AI’s co-founders. NP Co. aims to replace traditional simulation software that can take days or weeks to run a single iteration with models that can deliver results in seconds while preserving fidelity. The startup reports speed-ups of up to 1,000 times on industrial benchmarks and is pushing toward larger assembly simulations. Co-founder Emmanuel Menier argues that “the next major breakthrough for AI will come from engineering applications rather than conversational systems,” highlighting their focus on AI for engineering workflows in aerospace, defence, energy, electronics, data centres, and automotive. Pre-trained foundational models promise quicker deployment than earlier AI simulators that needed extensive customer-specific training.
Investor Confidence Shifts to Domain-Specific AI for Technical Work
Taken together, Inherent, Mafer AI, and NP Company show how investors are moving toward vertical AI startups that tackle specialised technical problems. Inherent’s Faraday targets AI-native science; MaferOS focuses on formulation R&D automation software; and NP Co. develops physics simulation AI that can slot into existing engineering toolchains. The combined funding of $50m plus €8 million signals clear appetite for AI that accelerates deep, domain-specific workflows rather than generic productivity. This aligns with a broader belief that the next wave of AI value will come from systems tightly integrated with scientific data, lab equipment, and engineering models. For R&D leaders and engineering teams, the message is that domain-focused AI for engineering and formulation research tools is moving from concept to funded reality, promising faster experimentation, shorter design loops, and closer links between historical know-how and new product development.
