Closing the Loop Around Every HCP Interaction: A Palantir Foundry Case Study in Pharma

The pharmaceutical industry has tremendous investment and various commercial operations teams operating behind the scenes to get the right messages through the right channel to the right healthcare professional (HCP) at the right time. Every team involved is operating in a siloed manner with different systems and variations of data, which increases the complexity and the pace of recommendations provided to the field team.  That fragmentation is part of why pharma companies are increasingly turning to platforms like Palantir Foundry: to unify commercial data and give AI transformation efforts a common foundation to build on, rather than solving each function in isolation. 

The Disconnect Behind the Desk

In most global markets outside the US, prescription drugs cannot be marketed directly to patients. The prescribing decision belongs to the HCP or a payer, hence the need to engage and influence HCPs becomes much more critical. Global brand teams translate dense clinical trial data into HCP-facing content, CLM decks and, visual aid email (VAE) assets, which have historically been built through external vendors: expensive, slow, and difficult to scale. Regulatory and market differences compound the problem; a CLM deck approved in Germany typically has to be rebuilt and re-approved for Spain rather than simply translated, a process that’s slow and prone to inconsistency across markets. 

Once approved, content moves to In-Field Teams (IFTs), who engage HCPs through multiple meetings spanning weeks or months, scheduling calls, preparing tailored talking points, logging notes after every interaction, and rolling it all into reports that track engagement and its impact on prescribing. Multiplied across a field force of hundreds or thousands of reps, this becomes a heavy administrative burden, and a natural place for AI to help reps prepare faster and spend more time engaging HCPs and less on manual admin. Closing that gap, connecting what content teams create with what happens in the field, is where MathCo’s work began. 

From Data Foundation to Production System

MathCo came in as a strategic and business partner for the pharma organization, continuing Palantir’s initiative for the client and working closely with its teams to understand their requirements before building on top of the foundation already in place. The client’s work was already structured around Palantir Foundry, which unified their data, Veeva records, content libraries, brand, and market data under one ontology. 

This is where MathCo’s capabilities carried the greatest weight. Years of hands-on Foundry and AIP delivery experience gave MathCo the platform expertise to translate that ontology into a working system. That experience is backed by a bench of certified Forward Deployed Engineers and continuous upskilling as the platform evolves, and it showed up in the build itself: the interface, the infrastructure, and the intelligence layer. Design sat at the core of every build, and disciplined delivery practices, structured product development lifecycle (PDLC) rigor, paired with AI-assisted engineering, kept both applications moving in lockstep rather than as separate projects. 

IFT Co-Pilot: One Foundation, Two Engineered Applications

Regardless of which team it served, MathCo’s approach stayed consistent: understand the business workflow first, then engineer the platform capability to match it. MathCo’s in-house pharma subject-matter experts, many with over a decade of experience in commercial and field operations, shaped how content and field workflows needed to connect from day one, rather than treating that connection as an afterthought. 

A copilot was built to support representatives through the full engagement lifecycle, before and after every HCP interaction. IFTs already use Veeva CRM to schedule and plan these meetings, so to keep the solution connected to how representatives work, it was embedded directly within Veeva CRM rather than operating as a separate tool that a representative must navigate to. 

Built on Palantir Foundry using OSDK, the application runs on LLMs, and business rules grounded in governed ontology objects, so every read and write move through the same audited, compliant path. In practice, this means: 

• Ahead of a meeting: the tool draws on a representative’s history with a given HCP to generate AI-powered talking points, territory insights, and cycle insights, so reps walk in prepared with context from every prior engagement. 
• After a meeting: free-text notes are converted into a structured, editable call report, validated against compliance requirements, and synchronized bi-directionally with Veeva CRM. 
• Looking ahead: the system recommends the next best action to inform the following visit. 

The application went live in late April and has maintained full uptime since launch. MathCo also built adoption tracking into the application itself, monitoring usage, feature adoption, and token consumption rather than assuming engagement after go-live. Within weeks: 

• 72% of IFT representatives had used the application 
• Nearly a third of all logged call reports were generated through the tool 
• Representatives saved an estimated eight or more hours of administrative time per week 

Outcome: Reduced planning and execution time for IFTs, elevated HCP engagement quality, eliminated redundant studies, and improved compliance and governance across the field. 

AI Content Generator: From Manual Localization to Automated Production

To help brand teams expedite content generation, MathCo built an AI Content Generator for CLM, and visual aid email (VAE) content used in HCP engagement. Rather than starting from scratch for every market, the application lets teams discover existing approved content and localize and adapt it, tuned to market, language, and HCP awareness level. 

Built on Palantir Foundry, the platform works as follows: 

• Content discovery: it extracts structured meaning from approved content cards and maps them by brand and audience, so teams can find and reuse what already exists instead of rebuilding from zero. 
• Content generation: AIP Logic, powered by GPT-5, generates new CLM, and VAE content with HCP persona, tone, and language built directly into the prompt. 
• Review and export: content is previewed and refined in an in-house application before being exported in eWizard-ready format for compliance sign-off and field deployment. 

Based on the current trajectory, anticipated outcomes include: 

• 60% increase in content production efficiency 
• 50% reduction in time to market 
• CLM and visual aid email content is now authored entirely in-house, removing vendor dependency end-to-end 
• n=1 personalization now possible at scale 

The two applications don’t just share a platform; they share a feedback loop. Outcomes captured through IFT Co-Pilot continue to inform what the AI Content Generator prioritizes and localizes next. Building that connection required both pharma domain knowledge, understanding what should link to what, and the platform and AI engineering depth to implement it. 

Built for Scale, Built for Compliance

Every AI-generated output from either application stays editable and is validated against compliance rules before it’s saved or delivered. Behind that, MathCo built a structured feedback loop for the AI layer itself: 

• An evaluator scores model responses against a threshold 
• Sub-threshold responses route to human reviewers 
• Corrected judgments feed into a golden evaluation set used to refine the system and guard against regressions over time 

Output quality improves on a documented basis rather than by assumption. Veeva remains the system of record; the AI layer accelerates the work happening inside it rather than replacing the controls around it. 

The rollout followed the same discipline: 

• Proof of concept in a single region 
• Validation with a working group of representatives and content leads 
• Scale-up across five languages, regions, and departments 
• Productionization, with each stage tested against real usage, not assumptions 

Outcome: Faster content creation, less time lost to manual call administration, more confident HCP conversations, and adoption gains that compound because they come from one connected system. 

From Foundation to Full Value

An ontology gives a Foundry deployment its structure, but structure alone doesn’t produce outcomes. Closing that gap takes a partner who pairs deep platform expertise with pharma-specific domain knowledge, and the discipline to prove value before scaling rather than assume it. 

The question worth asking isn’t whether your Foundry investment has the right foundation. It’s whether you have the right partner to build on it. Reach out to see what that could look like for your organization. 

Learn more about MathCo’s Palantir Foundry capabilities or connect with our experts to start the conversation.