Systems & models
Systems, models and platforms
WithPraxis works around the client's existing systems and chooses models, tools and platforms around the problem.
We do not lead with a model or a vendor. The workflow leads.
Model and platform agnostic AI
We do not start with a preferred AI model. We start with the business problem, workflow, data, systems and level of control needed. From there, we choose the right model, tool or platform for the job. Model and vendor names are used only as selection context — never as the lead.
Microsoft 365 and Copilot
We work with Microsoft 365, Teams, SharePoint, Outlook and Excel where they are the right fit. We assess Copilot the same way we assess any other tool: against the workflow, the permissions and content involved, and the level of control needed.
OpenAI, Claude and Gemini
We use frontier models from OpenAI, Anthropic (Claude) and Google (Gemini) where their capability and pricing fit the workflow. We do not commit to a single vendor and we are not paid to recommend one.
Azure AI and managed services
We use Azure AI services where they fit the customer's existing stack, data residency or commercial position.
Open-source and private LLMs
Where workflow sensitivity, control or cost call for it, we use open-source or privately hosted models. This is often the right answer for internal knowledge, sensitive data or high-volume routine work.
CRM, ERP and spreadsheets
Most of the work that matters runs across CRM, ERP and spreadsheets. We integrate around them rather than replace them.
Ecommerce, PIM and product data
For commerce teams we work with ecommerce platforms, PIM and the product data feeds that flow into them.
Documents, knowledge bases and reporting sources
Documents, internal knowledge bases, BI tools and reporting sources are first-class inputs. We expect AI workflows to reach them safely and reliably.