Our AI principles
Our approach to AI comes from our founding mission: prevent conflicts on construction projects before they get expensive.
We believe an approach to AI should be both bold and responsible. In construction, where trust, precision and accountability matter, we build for transparency, security and human control while delivering capabilities that improve project outcomes.
A note for teams using AI for the first time
AI can be powerful and unfamiliar, especially in construction, where a missed notice period or a misunderstood requirement carries real consequences.
Our approach is deliberately bold in what we build and disciplined in how we protect your business, your data and your decisions.
Practical innovation that improves project results
We build AI features that address real problems on projects, not novelty.
What we focus on:
• Automated document review to identify contractual risks and key obligations
• Smart search that understands the relationships between construction documents
• Risk analysis that flags issues before they become disputes
• Deadline tracking that helps your team stay ahead of contractual dates
What this looks like in Telamont
• Automated document review: the AI analyzes your contracts and identifies clauses, risks and obligations in minutes
• Natural language search: ask questions about your documents the way you would ask a colleague
• Proactive risk analysis: alerts on potential issues before they become problems
Privacy, security and data control by design
Construction documents are sensitive. Your contracts, pricing, correspondence and claim strategy are not training data.
Your data is not training data
We do not use your documents or data to train AI models. Our AI providers do not train on them either.
Where your documents are stored
Your documents are stored in Canada, including disaster-recovery copies.
Encrypted in transit and at rest
Your data is encrypted in transit (TLS 1.2 or higher) and at rest.
Separation between organizations
Each organization's data is isolated from every other organization's data.
Retention
We keep your documents for as long as your account is active. After your agreement ends, we delete them within 30 days, as set out in our Data Processing Agreement.
Secure access
Least-privilege permissions and strong authentication.
What this looks like in Telamont
• Cloud infrastructure in Canada, managed by us
• Administrative controls for users, permissions and access
• Audit-ready access and operational practices
AI under human control, transparent and accountable
AI should never quietly change obligations, invent facts or push a team into a risky action.
Grounded in your documents
Answers cite the passages of your own documents they rely on.
Project documents take precedence
When contracts conflict, we read the order-of-precedence clauses in your contracts. We do not apply static assumptions.
No invented facts
If the answer is not in your documents, the AI says it found nothing rather than guessing.
Clear review points
Notices, claim language and high-stakes outputs require human review. The AI drafts and recommends. A person decides and sends. Telamont sends no notice on its own.
Explainability rather than magic
We show why something is flagged: which clause, which specification section, which record.
What this looks like in Telamont
• Citations and traceability back to the underlying documents
• Safeguards built for construction workflows and construction risk
• Feedback loops so your team can correct and improve results
Our AI principles in action
Our principles guide how we build and operate Telamont.
Lifecycle governance
Design, test, controlled deployment, monitoring, iteration.
Due diligence on suppliers and components
We assess the security, privacy and reliability of the suppliers and components we rely on before we use them.
Adversarial testing
We test AI outputs against curated test sets covering fabricated quotes, wrong document versions and missed obligations.
Provenance and traceability
Document versioning and source attribution, so anything can be checked quickly.
Operational accountability
We maintain documented processes for security, reliability and incident response.
Frequently asked questions
Is my data used to train AI models?
No. We do not use your documents or data to train AI models, and our AI providers do not train on them.
How does the AI produce its answers?
It works only from the documents in your project. Each answer is based on specific passages from your contracts, with precise references to the source clauses and documents. To produce an answer, the relevant content is sent to our AI providers for processing. They are listed on our Trust Center.
Can the AI make decisions on its own?
No. AI assists and recommends. Important decisions, such as sending a contractual notice, assessing a claim or pursuing legal remedies, require explicit human approval. You stay in control.
Where is my data hosted?
Your documents are stored in Canada, including disaster-recovery copies. Data is encrypted in transit and at rest. To produce answers, content is sent to our AI providers. Those providers are listed on our Trust Center.
