Michael Barrett

Michael was the CEO of the first biotech firm to implement Snowflake, back in 2017. Most of the life sciences industry is still relying on system integrators like ZS Associates and Accenture to retrofit AI and LLM capabilities onto legacy platforms, and the result is a locked-in chatbot or proprietary add-on that does not scale. Michael's stance has been the opposite from the start: use Snowflake as the foundation and follow its trajectory as a category-defining company. That position has proven correct.

Michael's earlier career was in Accenture's data management practice. The exposure to enterprise data programs across regulated industries is the foundation everything else is built on, and it is the reason Michael is candid about why pharma keeps getting locked into the same recurring consulting cycles.

At Accolade, a patient services firm using behavioral economics to influence how health plan members use their benefits, Michael took the explicit position that the company would build its own technology rather than lock into Salesforce or Microsoft Dynamics. The reasoning was straightforward: owning the software and harnessing the data was the differentiator, and it preserved acquisition optionality. That same philosophy applies directly to LoxaNova. We build it ourselves, we own the IP, and that is what makes the eventual exit clean.

Veeva is the world leader in customer relationship management and other software for life sciences, including CRM for field sales and CRM for med tech. Michael led US data strategy and data science there and worked directly with the product owners and strategy leaders building those CRM tools. He knows what an effective life sciences engagement platform looks like from the inside, and he knows what a team has to do to build one.

At Intarcia, a Boston biotech, Michael was CDO and CIO during development of an osmotic mini-implantable pump (the ICA 650) that delivered a microdose of a GLP-1 to type 2 diabetics over a ninety-day period. Because it was a REMS product, Intarcia had to train and certify the clinical staff at endocrinologist offices and surgical centers on how to use the kit and place the device subcutaneously, covering scalpel, sutures, depth tool, the entire procedure. The training-and-certification problem Michael solved there is structurally identical to what LoxaNova has to solve for arthroscopic surgeons and OR staff using Daniel's devices. It is on the roadmap by design.

The entire data model lives here. HCP data, territory structures, sales activity, inventory movements, OR case records, patent-linked product catalog, Sunshine Act reporting tables. Michael has already configured this environment for sensitive healthcare data.

NVIDIA Spark handles local inference for anything touching sensitive data. Snowflake Cortex runs ML directly on warehouse data without moving it. OpenAI handles non-sensitive tasks like rep coaching prompts, device recommendation narratives, and sales insight generation.

Streamlit in Snowflake builds dashboards that run directly on warehouse data. Sales manager territory views, Sunshine Act compliance dashboards, inventory reconciliation reports, pipeline analytics, and AI-generated forecasts all live here.

Michael does not like calling it a CRM. It is a healthcare provider engagement tool: collaboration across accounts, contacts at those accounts, territory performance, ASC and group-practice consignment activity, and surgeon onboarding. Softr is the current implementation. It is a no-code environment that gives the team a working engagement layer in front of the Snowflake data model fast, so the May 12 demo is a functioning system the team can actually click through, not a scaffold or a screenshot. Softr is the bridge. We use it to validate the workflow and the data model, then pivot to owned IP.

Once the workflow is validated in Softr, the platform pivots to Node.js, React, and React Native, frameworks that produce IP the entity owns outright. This is the same philosophy Michael set at Accolade: do not lock into Salesforce or Microsoft Dynamics. Build it, own it, and the acquisition stays clean.

Auth0 handles rep and manager login, territory-based role permissions, and enterprise SSO for hospital system integrations. When an acquirer's IT team wants to integrate LoxaNova into their SSO environment, Auth0 makes that a configuration change rather than a rebuild.

A rep on a phone or tablet says something like: "Create a new account for Mayo Clinic at 123 Main Street, add Dr. Michael Barrett as an orthopedic surgeon, Mayo is his primary place of surgery, add them into my territory." The agent listens, distills the prompt into a set of create / update / delete operations, and applies them to the engagement tool. No more clicking "add new account," no more searching menus. The rep talks, the system records.

"What HCP did I call on the most last month?" The rep stops searching through reports and views. The bot reads the data and answers. Embedded directly in the same engagement surface so the experience stays in one place.

The bulk of LoxaNova's data (sales, claims, Medicare, CMS, whatever else gets layered in) stays in Snowflake rather than being copied into the engagement layer. Snowflake Cortex is effectively a coding agent plus the ability to run natural language queries against a semantic layer that Snowflake lets us build without writing code. We expose Cortex as an agent inside the same React / Node framework, so the rep gets analytics in the same interface as the engagement workflow. Asking "Show me my territory totals" or "Who is the next doctor I should talk to" happens in the same place, with one interface and one experience.

There is a company called Hex (H-E-X) doing exceptional work on natural-language-to-visualization. You ask a question and get an answer plus the chart that supports it, without ever opening a report. LoxaNova's engagement platform should be a small, med-device-shaped version of that idea. No more paging through Excel, Tableau, or Power BI to figure out how you are doing. You ask, and the system answers contextually and visualizes alongside.

At Intarcia, the ICA 650 implantable pump was a REMS product. Michael's team had to train and certify the clinicians at endocrinologist offices and surgical centers on how to use the kit (scalpel, sutures, the placement tool that set the pump at the correct subcutaneous depth), then attest that they had completed the certified procedure. That training-and-attestation infrastructure is structurally the same thing Daniel's devices will need in the OR for arthroscopic procedures.

Loic has worked with a company called Acto at a customer who is using LLM and avatar-based training with their sales reps. Approved, medically and legally reviewed materials get delivered conversationally. Reps interact with an avatar instead of paging through hundred-page PowerPoints, ISI documents, and three-page FDA labels. The result is faster ramp, current information, and an audit trail of who has consumed which approved content. This is a strong model for how LoxaNova handles rep enablement, change management, and onboarding as we scale.

The training problem in med tech is broader than the sales team. The surgical team using Daniel's devices needs documented training and proof of certification on the procedure itself. Building that capability into the LoxaNova platform (scalable, auditable, tied to the device and to the case) is the long-arc addition to the roadmap that ties directly back to the work Michael did at Intarcia.