§ Group · 03 · Cloud & AI Partners

The platforms our software runs on.

Three hyperscalers, two frontier model providers, one data-and-ML platform, and two iPaaS layers — the eight foundations Droz designs and operates on for our customers' enterprise software, applied AI, and automation work. We are cloud-agnostic by design; the engagement chooses the stack, not the firm.

Group
03 of 03
Members
8
Practice
Software & applied AI
Other groups
Associations · Technologies
§ The eight platforms

Cloud, AI & automation partners.

Cloud, AI, and automation platforms underpinning Droz's software and intelligence delivery. Click any card to jump to its detail block below.

§ Member · 01

Microsoft Azure

Microsoft's hyperscale cloud — including Azure OpenAI Service, AI Foundry (Azure AI Foundry), Azure Machine Learning, and the broader IaaS / PaaS portfolio. The hyperscaler of choice for many Canadian public-sector and regulated workloads thanks to Canada Central / Canada East residency and the Government of Canada (GoC) cloud framework agreements.

How Droz applies it

Droz builds and operates production workloads on Azure for customers with Microsoft-aligned IT estates, GoC residency requirements, or existing Azure OpenAI Service deployments. AI Foundry is used for managed model orchestration where customers want a vendor-supported MLOps surface rather than a roll-your-own stack.

Where it shows up


§ Member · 02

Amazon AWS

Amazon Web Services — including Amazon Bedrock (managed access to frontier and open models), Amazon SageMaker (managed ML training and deployment), and the broader compute / storage / data portfolio. AWS retains the largest North American market share and is the default for many private-sector enterprise estates.

How Droz applies it

Droz builds on AWS for customers with existing AWS estates or those choosing Bedrock for multi-model access (Claude, Llama, Mistral, Cohere Command) under a single managed surface. SageMaker is used for custom training where the engagement justifies the heavier ML lifecycle.

Where it shows up


§ Member · 03

Google Cloud

Google Cloud Platform — including Vertex AI (managed model and ML platform), Gemini (Google's frontier model family), BigQuery for analytics, and the broader data-and-AI portfolio.

How Droz applies it

Droz works on GCP for customers standardized on Google's data stack (BigQuery, Looker) or those preferring Vertex AI's model-and-MLOps surface. We often pair Vertex AI with Anthropic Claude through Vertex's third-party model access for customers who want Claude on GCP.

Where it shows up


§ Member · 04

Anthropic · Claude

Anthropic builds Claude, a frontier large-language-model family designed around safety, reliability, and long-context reasoning. Claude is available directly from Anthropic, via Amazon Bedrock, via Google Vertex AI, and (where supported) via private deployments under enterprise agreements.

How Droz applies it

Claude is Droz's default frontier model for production agentic systems — used inside Droz Legal, Samson AI, RX Change, the Droz Predictive copilots, and most custom AI engagements where reasoning depth, long context, and instruction-following matter more than raw cost. Droz selects the deployment surface (direct API, Bedrock, Vertex) based on the customer's data-residency and procurement posture.

Where it shows up


§ Member · 05

Cohere · Canadian LLMs

Cohere is a Toronto-headquartered frontier-model company offering the Command family of large language models and Embed (state-of-the-art embedding models for retrieval). Cohere offers private deployment and Canadian residency options, including AWS Canada and Azure Canada regions.

How Droz applies it

For Canadian customers with strict data-residency requirements — particularly GoC and provincially regulated workloads — Cohere is a preferred default. Droz often pairs Cohere Embed for retrieval with either Cohere Command or Claude for generation, depending on the reasoning load.

Where it shows up


§ Member · 06

Databricks

Databricks is a unified data, analytics, and ML platform — built around Apache Spark and Delta Lake, with the Unity Catalog governance layer, Mosaic AI for model training and serving, and the lakehouse architecture. Available natively on Azure, AWS, and GCP.

How Droz applies it

Databricks is Droz's default lakehouse for customers consolidating fragmented data estates. The reliability data platform is often built on Databricks where the customer already has the entitlement; otherwise we choose the data layer per cloud.

Where it shows up


§ Member · 07

Zapier · Make

Two of the leading low-code workflow automation platforms. Zapier and Make (formerly Integromat) connect SaaS endpoints with visual builders, schedule jobs, and orchestrate light-weight integrations across business systems.

How Droz applies it

Zapier and Make are Droz's default tooling for business-process automation engagements where the integration sits between SaaS endpoints and does not warrant a dedicated iPaaS or custom code. Most BPA engagements start in Zapier or Make and graduate to n8n / Tray.ai or to custom code only when complexity demands it.

Where it shows up


§ Member · 08

n8n · Tray.ai

Two iPaaS (integration platform as a service) layers used where workflow automation has outgrown Zapier / Make. n8n is open-source, self-hostable, and code-extensible — a Droz favorite for customers who want sovereignty over their integration layer. Tray.ai is a managed enterprise iPaaS with stronger native AI orchestration.

How Droz applies it

Droz selects n8n for engagements where the customer wants ownership of the integration runtime — particularly common in GoC and regulated industries that prefer self-hosted infrastructure. Tray.ai is selected where the customer values managed AI agent orchestration and is comfortable with a SaaS integration layer.

Where it shows up

§ How we choose
Selection logic.

Droz is cloud-agnostic by design. Four short rules govern how we pick the stack for any given engagement.

Rule · 01

Residency first.

Where data must stay in Canada, Azure Canada Central / Canada East, AWS Canada Central, or Cohere's Canadian deployment lead the shortlist.

Rule · 02

Customer estate.

We meet customers on the hyperscaler they already operate. Switching clouds is a project, not a side-effect.

Rule · 03

Model selection.

Claude is default for reasoning-heavy production agents. Cohere is default where Canadian residency matters. Other model families are used where the engagement justifies it.

Rule · 04

Automation graduation.

Automation work starts in Zapier or Make and graduates to n8n, Tray.ai, or custom code only when complexity demands it.

§ Disclosure
Droz builds on these platforms; inclusion here is not a formal partnership claim except where a specific partner tier is stated separately.