The Forward Deployment Approach That Helped Build $100B+ Companies
What I discovered studying how Palantir, OpenAI, and other B2B tech giants used systematic customer discovery to find breakthrough product-market fit and when this approach makes sense for founders
What I’ve Been Researching: A Powerful Discovery Approach Most Founders Overlook
I’ve been researching how some of the world’s most valuable B2B companies actually discovered their breakthrough product-market fit, and I found something fascinating that doesn’t get talked about enough in typical PMF discussions.
Here’s what caught my attention: instead of relying on traditional customer interviews, some companies embedded their team inside customer organizations for months at a time. They watched how their customers actually worked, discovered problems customers couldn’t articulate, and built solutions that created measurable business outcomes.
This approach has a name: Forward Deployed Engineering (FDE). It was pioneered by Palantir (now worth $65B), and it’s been strategically adopted by OpenAI ($157B), Anduril ($14B), and other B2B tech giants... but not in the way most people think.
Here’s the crucial insight I discovered: Apart from Palantir, these successful B2B companies didn’t all start with the FDE approach. Scale AI, Ramp, OpenAI, and Anduril all began as product-first companies that found initial PMF through scalable solutions. They only added FDE capabilities years later when they started serving complex enterprise, government, and military clients who needed deep customization.
What really surprised me: the real power isn’t about forward deployment per se, it’s about founders systematically discovering what B2B customers actually need. And you don’t need to be technical to do it.
However, this forward deployment approach only makes sense for specific types of B2B startups. Many B2B products work perfectly well with traditional discovery methods and don't need this intensive approach at all. However, every B2B founder can learn from the systematic discovery methods these companies use.
What I Learned: The Two Proven Paths to B2B Success
Path 1: Services-Led Growth (Palantir Model)
Palantir’s (2003) founders embedded with intelligence analysts for months, discovering problems analysts couldn’t articulate (like spending hours manually cross-referencing data). This led to their core data integration platform.
Path 2: Product-Led Growth, Strategic Embedding Later (Modern Model) -
- Scale AI (2016) started with API product for labeled training data → Added FDEs for government deployments (2023+)
- Ramp (2019) started with corporate card + expense management → Added FDEs for enterprise implementations (2022+)
- Anduril (2017) started with defense products (Lattice OS) → Added FDEs for military deployments
- OpenAI (2015) started with API products → Added FDEs to help enterprises move from trial to production (2024+)
The Key Learning: There are two equally valid paths for B2B companies. Some started with services, others achieved initial PMF through scalable products then added embedding strategically. The key is choosing the right discovery approach for your specific B2B context and founder background.
Key Discovery: Why This Creates Breakthrough B2B Products
Traditional discovery is like a detective interviewing witnesses by phone, while customer embedding is like investigating the crime scene for six months.
Why This Creates Breakthrough Products:
1. Discover Unarticulated Problems: Ramp’s embedded teams noticed finance teams’ real problem wasn’t tracking expenses. It was approval workflows creating month-end bottlenecks. This insight led to Ramp’s AI-powered automation features.
2. Validate Through Outcomes: OpenAI’s embedded teams built AI workforce scheduling solutions that cut a telecom company’s field intervention times by 30%.
3. Build Context-Aware Solutions: You see the ripple effects and environmental factors that remote discovery misses.
Analysis: When Each Approach Makes Sense for B2B Startups
Important: This framework applies to specific B2B scenarios. So, not all B2B products need this level of discovery complexity. Many successful B2B companies find PMF through traditional methods and never need customer embedding at all.
Services-Led Growth If:
- You have deep domain expertise in specialized markets and are genuinely unhappy with the status quo
- Customer workflows are highly unique and non-standardizable
- Traditional discovery won’t work due to complexity/regulation
- Customers spend $1M+ annually on the problem and will give you deep access
- You can dedicate 50%+ time for 12+ months per customer
Product-Led Growth If:
- You have product development experience and can build scalable solutions
- Customer workflows have standardizable elements
- You can validate through traditional discovery methods
- Your market includes both simple and complex customer segments
- You want capital-efficient validation with faster revenue
Keep in mind: Neither is better. Palantir succeeded with embedding because it matched their intelligence backgrounds and government market. Scale AI succeeded product-first because it matched their technical backgrounds and API market.
What I've Learned About Implementing Forward Deployment
You don’t need to be an engineer. While original “Forward Deployed Engineers” were technical, the methodology is about systematic problem discovery. Whether you’re a designer, businessperson, or domain expert, you can embed to understand workflows and test solutions.
What you need: Deep curiosity, pattern recognition, systematic measurement approach, domain expertise.
What you don’t need: Coding ability, engineering background (work with contractors for technical prototypes).
The 6-Month Process:
Months 1-2: Access & Understanding
- Identify 3-5 customers with $1M+ problem budgets in your expertise area
- Propose a “discovery partnership,” offering free analysis in exchange for deep access
- Map workflows and quantify current problem costs
Months 3-4: Solution Design & Testing
- Shadow teams daily - watch problems happen in real-time
- Design testable interventions (processes, tools, workflow changes)
- Test with actual data in their environment
Months 5-6: Outcome Validation
- Deploy solutions in production workflows
- Track business impact over 60-90 days
- Document what works and why
Success Criteria: 10x ROI demonstration on their current problem-solving costs
Universal Framework (Apply to Any B2B Business)
Even if you never do full customer embedding, apply these principles:
1. Immersion Over Interviews
Traditional: “Can I interview you about problems?”
Better: “Can I spend a week watching how you work?”
2. Problem Archaeology
Traditional: “What problems do you have?”
Better: “Let's try to identify problems you haven’t noticed yet.”
3. Outcome-Driven Validation
Traditional: “Would you pay for this feature?”
Better: “Did this solution measurably improve your business?”
4. Pattern Recognition
Traditional: Build custom solutions for each customer
Better: Extract reusable patterns that can scale
Conclusion: Choose the Right Discovery Method for Your B2B Business
After studying these billion-dollar companies, my main takeaway is that both approaches can create massive B2B success... neither is universally better.
Key Learnings:
- Match approach to founder-market fit: Leverage your expertise and specific B2B context.
- Understand the tradeoffs: Services-Led Growth = higher costs, deeper insights; Product-Led Growth = capital efficient, may miss complexity.
- Not all B2B products need this: Many successful B2B companies never need forward deployment. Traditional discovery works fine.
- Apply universal principles: Immersion, outcome validation, pattern recognition (regardless of approach).
- Be prepared to evolve: Successful B2B companies eventually use both methods strategically.
Final Insight: Palantir succeeded with customer embedding because it matched their founders’ intelligence backgrounds and government market complexity. Scale AI and Ramp succeeded product-first because it matched their technical backgrounds and more standardizable markets. Your B2B path should match your context.
The goal isn’t to copy anyone’s playbook—it’s to choose the discovery method that gives you the best chance of finding breakthrough PMF in your specific B2B situation.
Want to dive deeper into systematic PMF? This is exactly the kind of framework we explore in my “Learning PMF in Public” approach — taking proven methods from billion-dollar companies and making them accessible for founders chasing product-market fit.
