Philipp Wehn spent over a decade watching engineers at Siemens and Kearney wrestle with fragmented software. ERP, PLM, MES, and a dozen legacy tools sat in separate windows, each requiring manual clicks and cross-referencing. The automation tools on offer were brittle, built for clean data environments, not the bespoke, human-in-the-loop workflows of heavy industries like rail, energy, and construction. In 2024, he and co-founder David Huang started Nexxa.ai to build a different kind of assistant. Their bet: specialized AI agents that work by watching the screen, not by ripping out the stack.
Nexxa.ai has since raised $13.4 million to prove it. A $4.4 million pre-seed round led by Andreessen Horowitz in July 2025 was followed by a $9 million seed round led by Construct Capital in January 2026 [PRNewswire, July 2025][Yahoo Finance, Jan 2026]. The capital backs a wedge into one of the most stubborn markets for software adoption.
The computer-use wedge
Nexxa.ai’s product is a multi-agent system designed to sit on top of existing industrial software. The agents operate via what the company calls “computer-use AI,” interacting with the same screens and interfaces as a human engineer [Nexxa.AI]. The goal is to automate complex, end-to-end workflows,like processing a change order across ERP and PLM systems, or scheduling maintenance tasks in a MES,without requiring deep API integrations or system replacement [Nexxa.AI].
This approach targets a specific pain point. Traditional robotic process automation (RPA) and integration platforms often fail in heavy industries where workflows are highly customized, decisions require expert judgment, and software ecosystems are a patchwork of modern and legacy tools [Nexxa.AI]. Nexxa.ai’s agents are built to learn from expert users, collaborating on tasks rather than replacing them, a concept the company terms “agent-augmented engineering” [Nexxa.AI]. The company claims no pre-training on customer data is required to start, aiming for faster time-to-value [Nexxa.AI].
Founders with field and code credentials
The founding team pairs domain depth with technical pedigree. CEO Philipp Wehn brings over ten years of experience in rail, metals, automotive, and construction from his time at Siemens and consultancy Kearney [a16z Speedrun, retrieved 2026]. He has publicly discussed industrial AI applications, including on a Siemens-focused podcast about Real-Time Location Systems [Cognaize | Hybrid Minds Podcast]. CTO David Huang was previously a security engineer at Facebook, bringing experience from scaling complex systems at a major tech firm [TechCrunch, 2018]. The team now numbers between 11 and 50 employees, according to LinkedIn [LinkedIn].
Early traction and the metrics question
Nexxa.ai points to strong early signals. Company materials presented in the a16z Speedrun accelerator claim the startup went from “0 to 980K in 9 months” and has sustained a growth rate of “3x every quarter” [a16z Speedrun, retrieved 2026]. They also report a “100% activation rate on all projects” [a16z Speedrun, retrieved 2026]. Siemens and Matikon are named as early customers in the company’s public summary.
These metrics, while compelling, are company-provided and lack independent verification. They represent the kind of early momentum that attracts venture checks but will face closer scrutiny as the company scales beyond its initial deployments. The real test will be renewals and expansion within these large, slow-moving enterprises.
Where the wheels could come off
The market Nexxa.ai is chasing is famous for long sales cycles, entrenched procurement processes, and a deep aversion to operational risk. The company’s differentiation rests on a few key assertions that will be tested in the next twelve months.
- Integration depth. The promise of “no deep system replacement” is a major selling point, but it relies on the AI’s ability to reliably navigate and manipulate a vast array of legacy UIs. Any failure in this “computer-use” layer could erode trust quickly.
- Defensible moat. The specialized knowledge of industrial workflows is a clear advantage. However, competitors like Ameba AI and Makersite are also targeting manufacturing and heavy industry with AI solutions. Nexxa.ai’s lead may depend on how quickly it can embed its agents into critical, daily processes.
- Economic proof. The cited growth to $980K (estimated) in nine months is impressive for the sector. The next milestone is proving that initial projects convert into larger, enterprise-wide deployments with significantly higher annual contract values.
The company’s answer, for now, is the early customer logos and the activation rate claim. The $9 million seed round from Construct Capital suggests investors see a path through these hurdles.
The next twelve months
Nexxa.ai is hiring for customer-facing roles, including a Customer Solutions Architect and an Account Executive & Business Development Lead, indicating a push to scale its go-to-market motion. The focus will be on converting pilot projects into expanded deployments within its current verticals,rail, energy, construction, and manufacturing.
The funding provides a substantial runway. The $13.4 million total capital raised came from a notable syndicate: Andreessen Horowitz leading the pre-seed, joined by Augment Ventures, Propeller Ventures, Plug and Play, Beat Ventures, SBI, Untapped Ventures, Cross Atlantic Angels, and angels like Rick Berry and Pat McCarthy [PRNewswire, July 2025]. Construct Capital then led the $9 million seed round [Yahoo Finance, Jan 2026].
| Round | Date | Amount | Lead Investor |
|---|---|---|---|
| Pre-seed | July 2025 | $4.4M | Andreessen Horowitz |
| Seed | January 2026 | $9M | Construct Capital |
For investors like a16z’s Speedrun program and Construct Capital, the bet is that Wehn’s field experience and Huang’s technical build can unlock a productivity leap in industries that have largely missed the last decade of software innovation. The question for 2026 is whether those industrial engineers, now working alongside AI agents, will agree.
Sources
- [PRNewswire, July 2025] Nexxa.ai Raises $4.4M to Bring Specialized AI to Heavy Industries | https://www.prnewswire.com/news-releases/nexxaai-raises-4-4m-to-bring-specialized-ai-to-heavy-industries-302508814.html
- [Yahoo Finance, Jan 2026] Rebuilding America and the World's Industrial Backbone With AI | https://finance.yahoo.com/news/rebuilding-america-worlds-industrial-backbone-170000874.html
- [Nexxa.AI] Nexxa.AI, https://nexxa.ai/
- [a16z Speedrun, retrieved 2026] a16z Speedrun materials | Source materials from accelerator program
- [Cognaize | Hybrid Minds Podcast] Cognaize | Hybrid Minds Podcast | Philipp Wehn, https://www.cognaize.com/podcast/philipp-wehn
- [TechCrunch, 2018] Facebook tool warns developers of phishing attacks dangling lookalike domains | https://techcrunch.com/2018/05/02/facebook-tool-warns-developers-of-phishing-attacks-dangling-lookalike-domains/
- [LinkedIn] LinkedIn company size data | https://www.linkedin.com/company/nexxa-ai/
- [iHeart] Automation That Doesn’t Break, Philipp Wehn, Nexxa AI | Ep 17 - SaaS Founder Stories w/ David Brennan | https://www.iheart.com/podcast/269-saas-founder-stories-w-dav-259658595/episode/automation-that-doesnt-break-philipp-259658755/