Solvera Raises $6.3M to Build an AI Agent Engineering Firm

An Oslo startup is building an engineering firm whose workforce is substantially AI agents — and clients pay for finished designs, not hours or seats.

Solvera Raises $6.3M to Build the Agent-Powered Engineering Firm

Oslo-based Solvera announced a $6.3 million pre-seed round on October 15, led by Seedcamp, to scale what it calls an AI-native engineering firm for the construction industry. The company, founded in July 2026 by Arif Mirza, Pål Rasmussen and Thorvald Thorsnes, is not selling software to engineers. It is selling finished engineering work — mechanical, electrical and plumbing (MEP) designs, delivered end to end — with AI agents doing the bulk of the production work and experienced human engineers holding responsibility for judgment and sign-off.

The round was joined by Construct Venture, Inception Fund, Momentum, Obos Ventures, PSV Tech, Sondo and The Nordic Web Ventures, alongside angel investors including Spacemaker founders Anders Kvåle and Håvard Haukeland and former OBOS chief executive Daniel Kjørberg Siraj, according to Tech.eu. The Spacemaker connection is a notable pedigree signal: Spacemaker's AI-assisted building design platform was acquired by Autodesk, and its founders backing Solvera suggests they see the AI-native firm as the natural next step from AI-assisted tools.

Agents as the Firm, Not the Tool

The central idea behind Solvera is an inversion of the usual enterprise AI pitch. Instead of licensing an agent to an engineering firm's employees, Solvera makes the agent the firm's operating system. Its purpose-built platform routes requirements and constraints through AI agents that draft, coordinate and revise designs, while human engineers focus on the high-impact decisions, according to Seedcamp's announcement. The model breaks the two revenue structures the industry has run on for a century: software seats and billable hours. Solvera clients pay for outcomes — coordinated, construction-ready designs — rather than licenses or consulting time.

Each project feeds back into the company's operating system, capturing engineering judgment so it can be reused across future work, reported by EU-Startups. That compounding loop is the bet that turns an engineering services business into something with software-like margins: the agent gets better at the firm's craft with every engagement, rather than every project starting from zero. For the agentic-economy beat this publication has been tracking — from agent identity to agent budgets — Solvera is a new specimen: agent labor sold as professional services, with the human role recast as judgment layer rather than production layer.

Solvera is starting with MEP engineering, a discipline where the complexity is coordination-heavy and much of the work remains manual and fragmented, according to Tech.eu. The company says it takes projects from initial requirements through to coordinated, construction-ready designs. Early commercial work is in Norway, where Solvera is delivering projects with contractor Betonmast, and the company plans to expand its Oslo engineering and technology team and open a London office, as TechNews180 reported.

The Productivity Case Is Measured in Orders of Magnitude

Mirza argues that two decades of better engineering software barely moved the industry's fundamental productivity, and that capturing AI's full potential requires rethinking how the firm operates and how it makes money — a structure he says is nearly impossible to change inside organizations built around billable hours, as reported by Seedcamp. The gap between potential and practice is striking: Mirza estimates that 85% of tasks in architecture and engineering could be completed at least twice as fast, yet AI is currently used in fewer than 5% of them, according to the Seedcamp announcement.

The traction data so far is early but directionally supportive of the thesis. Solvera exceeded $1 million in contracted revenue during its first three months of operations and grew to a team of nine, according to TechNews180. That revenue figure represents signed contracts rather than cash collected, but for a firm founded in July 2026 it suggests the outcome-based pricing pitch is landing with buyers. CTO Pål Rasmussen frames the value proposition around engineers' time: the most valuable asset in building engineering has always been engineering judgment, yet much of an engineer's day is consumed by sorting information, manual data entry and coordination, and Solvera's goal is to let agents handle alternatives-testing, coordination and production work while humans focus on the decisions that matter, as he told EU-Startups.

What to Watch Next

The immediate question is whether Solvera can scale from Norwegian MEP projects into the broader building-engineering market without the agent-to-engineer ratio breaking. The company's model only works if the agents genuinely carry production weight — if human engineers end up re-checking every calculation, the firm is just a traditional consultancy wearing AI branding. The London expansion will be the first test of whether the operating system travels: captured judgment from Norwegian commercial builds must transfer to different building codes and construction cultures.

The longer-horizon signal is about labor structure. If outcome-priced agent labor can displace billable hours in engineering — one of the most credentialed, judgment-heavy professions — the pattern will not stay in construction. The same inversion is already appearing in adjacent agent-services stories: agents that manage corporate spending on a corporate card, agents that secure other agents, agents that carry context across workflows. Solvera is simply the clearest version of the formula yet: a firm where the agents are the staff, the humans are the editors, and the invoice says "finished design" instead of "42 hours." The investors writing the checks — Seedcamp plus a deep bench of construction-tech specialists — are betting that this is how professional services gets rebuilt, one disciplined agent at a time.

Sources

Reporting in this article is drawn from Tech.eu, Seedcamp, EU-Startups, TechNews180 and EU Startups News. For more on the agent economy this publication covers, see this publication's reports on Ramp's $1.85B raise to put AI agent spending on a corporate card and Opsin's $21M round for AI agent security.