An architecture firm's most valuable asset is the work it has already done. The problem is that this asset, a sprawling archive of drawings, details, and layouts, is often trapped in a labyrinth of folders and forgotten project files. ARKI Digital is building a search engine to find it.
The Toronto-based startup, founded in 2023, is developing what it calls an AI-powered visual search engine and system of record for architecture and engineering (AE) firms. The core proposition is straightforward: ingest a firm's past project data and turn it into a searchable, reusable knowledge base. The goal is to let an architect query for, say, a specific type of bathroom detail from a past project and get the relevant drawing in seconds, rather than spending hours manually digging through old files [getarki.com/about].
The bet on institutional memory
ARKI's bet is that the bottleneck in design speed isn't a lack of creativity, but a lack of access to institutional knowledge. The company positions its platform as a "second brain" for firms, aiming to eliminate redundant design work by making past solutions instantly retrievable [Practice Disrupted, Oct 2025]. The platform is built to organize, search, and version design details, with a particular focus on BIM managers working in tools like Autodesk Revit [getarki.com/bim-managers].
This focus on a specific, painful workflow for a defined user gives ARKI a clear wedge into a conservative industry. The founders report the product is being developed with "real-time input from firms," suggesting an iterative, feedback-driven approach to product development [getarki.com/post/business-of-architecture-podcast, May 2024].
The founding team's dual expertise
The company's founding team pairs deep domain expertise with technical product-building experience. Co-founder and CEO Natalia Bakaeva is a registered architect in Ontario who ran her own multidisciplinary practice before starting ARKI [Practice Disrupted, Oct 2025]. Her co-founder, Mehdi Karamnejad, brings a software engineering background from previous roles, including at Salesforce [Mehdi Karamnejad on LinkedIn].
This combination is critical. Bakaeva's credibility within the architecture community helps open doors and shape the product around real practitioner pain points, while Karamnejad's experience informs the platform's technical architecture. The company participated in the Antler W23 accelerator cohort, which provided its initial backing [LinkedIn].
| Role | Name | Background |
|---|---|---|
| Co-founder & CEO | Natalia Bakaeva | Registered architect; former practice owner [Practice Disrupted, Oct 2025]. |
| Co-founder | Mehdi Karamnejad | Software engineer; former Salesforce employee [Mehdi Karamnejad on LinkedIn]. |
Early traction with design firms
While ARKI has not disclosed funding details or revenue, it has publicly named over a dozen architecture firms as clients. This early list includes MJMA Architects, Sweeny&Co Architects, RAW Design, and HGA, among others [getarki.com/about]. For a tool that requires ingesting and indexing a firm's sensitive project data, each new client represents a significant vote of confidence in ARKI's security and value proposition.
The technical breakdown of ARKI's offering hinges on two core components: the ingestion and structuring of heterogeneous design data (from PDFs and images to native Revit files), and the AI models that power semantic and visual search across this corpus. The promise of a 50% acceleration in project design hinges entirely on the accuracy and speed of these retrieval systems [getarki.com/about].
Scaling this system presents the primary technical risk. Architectural data is complex, proprietary, and often messy. Training models that can reliably understand the intent behind a rough sketch or a detailed section drawing across thousands of unique projects is a formidable machine learning challenge. Furthermore, as the corpus grows for each firm, maintaining low-latency search performance while ensuring data isolation between clients will test the platform's underlying infrastructure.
The competitive landscape
ARKI does not operate in a vacuum. Its value is in aggregating and searching a firm's own data, which positions it differently from large design suites like Autodesk's Civil 3D or Navisworks, which are creation tools first. A more direct comparison might be to platforms like Cupix, which focus on digital twins and reality capture. ARKI's differentiation rests on being a dedicated system of record for design knowledge, not just 3D models.
The path forward will be defined by depth of integration and proof of the efficiency gains. The next twelve months will show whether ARKI can move from being a useful search tool to becoming an indispensable layer in the architectural workflow. Success means architects won't just use it; they will design with the assumption that any past solution is a query away.
Sources
- [getarki.com, retrieved 2026] About | ARKI DIGITAL | https://www.getarki.com/about
- [Practice Disrupted, Oct 2025] Architecture's Memory Problem | The Profession That Keeps Forgetting | https://podcasts.apple.com/us/podcast/practice-disrupted-by-practice-of-architecture/id1518820081?l=ru
- [getarki.com, retrieved 2026] For BIM Managers | https://www.getarki.com/bim-managers
- [getarki.com, May 2024] Business of Architecture Podcast Recap | https://www.getarki.com/post/business-of-architecture-podcast
- [Mehdi Karamnejad on LinkedIn, retrieved 2026] Mehdi Karamnejad Profile | https://www.linkedin.com/in/mehdikaramnejad/
- [LinkedIn, retrieved 2026] ARKI (Antler W23) Company Profile | https://ca.linkedin.com/company/getarki