For a civil engineering firm, the most valuable data isn't on a spreadsheet. It's in the 500-page proposal that won the last bridge contract, the annotated budget from a similar highway project, and the 10-year-old RFP response that somehow secured the municipal deal. This is the archive Tyce wants to turn into a living library, one AI-generated document at a time.
Founded in 2025, the Toronto-based startup is building an AI-native document editor specifically for the built world. The pitch is straightforward: feed the platform your company's historical documents, and it will help you search, generate, and edit new ones, pulling context from your own past work. It's a bet that the next wave of productivity in construction, engineering, and architecture won't come from generic chatbots, but from tools that speak the industry's specific, document-heavy language.
The wedge: an AI editor for the built world
Tyce's differentiation is its focus. While generic AI tools can draft a contract, they don't inherently understand the difference between a construction bid, a civil engineering proposal, and an architectural specification. Tyce is built to index those documents, understand their components, and use them as a knowledge base. The platform promises to auto-fill forms, generate new documents from templates, and make edits that reference multiple past projects [Tyce, retrieved 2024]. Its AI estimating engine, for instance, is designed to benchmark new project estimates against weighted historical data [Tyce, retrieved 2024]. The goal is to turn years of project history into what founder Lylia Djait-Paulien calls an "AI-powered knowledge system" that can write proposals and documentation [Lylia Djait-Paulien on LinkedIn, retrieved 2026].
The team and the accelerator stamp
The company is led by solo founder Lylia Djait-Paulien, a former product lead at developer tools company LayerCI [Perplexity Sonar Pro Brief, retrieved 2024]. Her public speaking, including at the Carbon Capture Canada conference in 2022, points to an interest in the intersection of technology, data, and heavy industry [Carbon Capture Canada, retrieved 2026]. Co-founder Keyu Long is listed as CTO [Founders Network, retrieved 2026]. While the company has not publicly disclosed any funding rounds, its participation in the Formwork Labs accelerator's 2026 cohort places it firmly in the pre-seed, venture-building stage [Formwork Labs, retrieved 2026]. This provides an early stamp of validation and likely some initial capital, but leaves the scale of its runway and commercial traction as open questions.
| Role | Name | Background |
|---|---|---|
| Founder & CEO | Lylia Djait-Paulien | Former Product Lead at LayerCI (Encore.dev); speaker on climate data. |
| Co-founder & CTO | Keyu Long | Technical leadership role; background not detailed in public sources. |
Where the blueprint meets reality
Tyce's ambition is clear, but the path is crowded. The company is entering a space with established players who have already planted their flags. Its identified competitors, Document Crunch and Egnyte, are not hypotheticals. Document Crunch, for example, has built a similar AI-powered contract review platform for construction and has partnerships with industry giants like Trimble. For a young company like Tyce, the risks are structural:
- The data onboarding burden. Convincing a construction firm to upload a decade of sensitive project documents is a major ask. The value is only realized after this heavy lift.
- The generic AI ceiling. Large language models are improving at document tasks universally. Tyce must prove its industry-specific tuning and workflows create a defensible moat that a ChatGPT Enterprise plugin cannot easily replicate.
- The sales motion. Selling a system-of-intelligence platform requires a consultative, high-touch enterprise sale. This is a different game than selling point productivity software.
The rebuttal, presumably, is that generic tools fail precisely because they don't understand the nuance. A proposal isn't just text; it's a structured document with cost breakdowns, project timelines, compliance clauses, and firm qualifications that must be perfectly aligned. Tyce is betting that depth beats breadth.
For a firm with 50 project managers, each spending an estimated 15 hours a week on document drafting and review, the potential time savings are tangible. If Tyce can cut that time in half, it's reclaiming nearly 20,000 billable hours a year. The incumbent it must beat isn't just another software company, it's the status quo: the sprawling network of shared drives, outdated templates, and institutional knowledge that walks out the door with every retiree. The bet is that in an industry measured in concrete and steel, the real foundation for the next decade will be built on documents.
Sources
- [Tyce, retrieved 2024] Company website and product claims | https://www.tyce.ai/
- [Perplexity Sonar Pro Brief, retrieved 2024] Product and team details | Sourced from web-grounded research
- [Lylia Djait-Paulien on LinkedIn, retrieved 2026] Founder statement on AI knowledge system | https://www.linkedin.com/in/lyliadjaitpaulien/
- [Carbon Capture Canada, retrieved 2026] Conference speaking engagement | https://carboncapturecanada.ca/
- [Founders Network, retrieved 2026] Keyu Long role listing | https://foundersnetwork.com/
- [Formwork Labs, retrieved 2026] Accelerator cohort announcement | https://www.formworklabs.com/