The Manual Assembly Instruction Bottleneck

Creating clear assembly guidance for furniture is still often done the way factories built jet engines a decade ago. The bottleneck is not the design, it is the translation of CAD into something a human can follow on the floor.

Fil Aronshtein, cofounder and CEO of Dirac, described the typical workflow he saw in manufacturing: “Some dude gets a CAD file over email, takes hundreds of screenshots manually, figures out what order to do the assembly, then takes all those screenshots and throws them together into a 100-page PowerPoint.” He called it super manual, and super tedious.

That manual loop shows up in furniture assembly businesses too. Design files arrive, someone screenshots views, decides sequence by eye, and builds a booklet or video set by hand. Errors creep in, updates require redoing the whole deck, and speed stalls when SKUs multiply.

Aronshtein framed the pain with the IKEA analogy everyone recognizes: “You know that paper instruction book that tells you how to build something? ... It turns out everything around us also needs assembly instructions.” The need for clear, step-by-step guidance scales from flat-pack furniture to complex machinery.

Dirac was founded in 2023 as a small startup and raised roughly $11 million from investors including Founders Fund and Coatue Management. Its BuildOS product aims to use AI to generate assembly instructions for manufacturers, replacing the screenshot-to-PowerPoint process with automated sequencing.

The problem is not unique to aerospace or defense. The same friction costs furniture assembly businesses time, rework, and support calls when instructions are ambiguous or outdated.

Manual instruction creation typically involves:

• Receiving CAD files via email and manually extracting views
• Taking hundreds of screenshots to document each step
• Deciding assembly order by human judgment
• Compiling everything into lengthy PowerPoint or PDF booklets

When a design changes, the entire manual package must be rebuilt. That is why vendor selection for AI support must focus on customization, compliance, integration with existing tools, and long-term ownership rather than a generic template.

For SMBs evaluating AI partners, the lesson from manufacturing is clear: the value is in automating a tedious, high-touch workflow without creating new lock-in. AIQ Labs approaches this through systems businesses own outright, with payment models that let you start monthly, build in milestones, or own it outright. If instruction generation is just one of many workflows eating your week, an AI operating system can connect it to the rest of your operation instead of leaving it as another disconnected tool.

The manual assembly instruction bottleneck persists because the process was never designed for speed or scale.

How AI Is Automating Assembly Instructions

Manual assembly instructions are a hidden bottleneck that still relies on screenshots and PowerPoint. Fil Aronshtein, Dirac cofounder and CEO, described the typical workflow as “Some dude gets a CAD file over email, takes hundreds of screenshots manually, figures out what order to do the assembly, then takes all those screenshots and throws them together into a 100-page PowerPoint.” A Business Insider report on the Siemens-Dirac partnership highlights how super manual and super tedious that process remains.

Dirac is tackling that problem with BuildOS, a software tool that uses AI to generate assembly instructions for manufacturers. Instead of manual sequencing, the system ingests CAD files and determines order of assembly automatically. The company, founded in 2023, raised roughly $11 million from investors including Founders Fund and Coatue Management according to 2025 coverage of its Siemens partnership.

The shift matters for furniture assembly operations because the core pain is the same as industrial manufacturing. Aronshtein noted, “You know that paper instruction book that tells you how to build something? ... It turns out everything around us also needs assembly instructions.” The same article draws the IKEA analogy to explain why clear sequencing is critical from consumer furniture to complex machinery.

BuildOS was being integrated with Teamcenter, Siemens’ engineering management software, through a formal partnership. Tomas Klausing, a director of technology partnerships at Siemens, said the integration would allow Dirac to work more closely with Siemens’ customer base in the same Business Insider report. The target industries cited included aerospace, defense, automotive, agriculture, construction, machinery, and maritime.

For furniture assembly businesses evaluating AI partners, the Dirac example illustrates what to examine:

• How CAD input is handled and whether sequencing is automated rather than manually curated
• Integration with existing engineering and production tools, as shown by the Teamcenter integration
• Transparency on AI approach, since Dirac declined to disclose whether it uses a foundation model or trains its own

Automated sequencing reduces human error in step order and cuts the time spent building instructions from scratch. When instructions are generated from the source CAD, updates propagate automatically instead of requiring new screenshots and a new PowerPoint.

This is where vendor selection criteria matter. If instruction generation is the single workflow costing you the most, a fixed-price AI workflow build rebuilds just that piece, in phases, with clear milestones. The Dirac case shows why true ownership and integration are not optional, especially when assembly accuracy directly impacts cost and safety.

Competitive context is also relevant. Tulip, a manufacturing software startup, raised $100 million in 2021 and builds software for machinery with a platform that aims to track factory production and equipment assembly per the same coverage. The contrast highlights different product focuses within the same automation wave.

Choosing an AI partner for assembly instructions means looking beyond the demo to how the system fits your tools, data, and ownership model.

Lessons From Industrial AI Partnerships

When evaluating AI partners for furniture assembly, industrial case studies reveal that successful integration hinges on more than just technical capability, it requires deep alignment with existing engineering workflows and economic realities. The Siemens-Teamcenter integration with Dirac’s BuildOS platform exemplifies this principle: by embedding AI-generated assembly instructions directly into Siemens’ widely used engineering management system, the partnership enables seamless data flow from CAD models to shop-floor guidance without disrupting established processes. This approach addresses the core pain point described by Dirac’s cofounder, where creating instructions involves a "super manual, and super tedious" workflow of manually screenshotting CAD files and assembling them into PowerPoint decks, a method that scales poorly across complex products.

Dirac’s work across aerospace, defense, automotive, and maritime sectors further underscores the importance of partnership maturity and economic viability in vendor selection. As a small startup founded in 2023 that raised roughly $11 million from investors including Founders Fund and Coatue Management, Dirac demonstrates how focused AI solutions can gain traction in high-stakes manufacturing environments despite limited scale. Investors like Thomas Laffont of Coatue explicitly tied their support to the need for cost-consistent reindustrialization, framing AI as a pathway to make automation work within tight economic constraints, a consideration directly relevant to furniture assembly businesses evaluating ROI and long-term sustainability. These industrial examples highlight that the best AI partners don’t just offer novel technology; they prove they can integrate into complex toolchains, respect operational constraints, and deliver value without requiring rip-and-replace overhauls of existing systems. For furniture assembly operations, this means prioritizing vendors who demonstrate compatibility with common design tools (like CAD platforms), offer transparent economic models, and show evidence of solving analogous instruction-generation challenges in regulated or high-complexity settings, even if those examples come from submarines rather than sofas. The goal isn’t to replicate industrial scale, but to adopt the same disciplined approach to partnership evaluation: one that values integration depth, economic alignment, and proven applicability over flashy demos or unsubstantiated claims of AI prowess. AIQ Labs brings this same rigor to SMBs, with 200+ multi-agent systems deployed and payment models, start monthly, build in milestones, or own it outright, designed around ownership and long-term operational fit rather than vendor-driven hype.