Every B2B company right now is staring at the same problem: they need more Forward Deployed Engineers, they don’t have enough of them, and they’re looking at their CS team wondering if they can bridge the gap.
Usually, they can’t. Here’s why — and what to do instead.
FDE job postings increased 12x in a single year, according to ICONIQ’s 2025 GTM survey of 205 B2B SaaS executives. That’s not a hiring trend. That’s a structural shift in how AI companies operate. And the companies trying to fill it by repurposing their CSMs are mostly failing.
The jobs are not the same
A CSM manages relationships, mitigates renewal risk, and drives expansion across a portfolio of 8–12 accounts. An FDE embeds with 1–3 customers, writes and debugs workflows, clears deployment blockers daily, and owns whether the AI actually works in production. Those are not the same skill set. They’re barely the same profession.
The deeper structural problem: CS was built to be reactive and relationship-driven, optimized for retention. FDE work is proactive, technical, and optimized for deployment. The required skill set is closer to a solutions engineer or junior PM with strong customer empathy than a traditional CS rep. Most CS teams don’t have it, and retraining takes longer than most companies want to admit.
Why CS structurally can’t fill this gap
The whole premise of AI agents is that they automate work. But deploying an AI agent is itself significant work — and it’s work the agent can’t do for you. Not yet.
Someone has to understand the customer’s workflows deeply enough to know where the agent fits. Someone has to train it on the right data, the right context, the right edge cases. Someone has to test it, catch where it breaks, and iterate. Someone has to get internal buy-in from the people whose jobs will change when the agent goes live.
That’s FDE work. It’s manual, high-judgment, human work.
Here’s how badly it goes wrong when it’s not done: one $6 billion AI company — one of the most prominent in the market — had an agent quoting incorrect pricing to potential customers, telling them they’d need to quadruple their spend. When asked how long the product had been in beta: a year. No one had properly trained the agent. If a company that size can’t deploy its own AI correctly after 12 months, imagine what’s happening at the enterprises buying these tools without FDE support.
The economic reality
FDE work is expensive and slow by nature. A single customer deployment can take 30–60+ days of daily embedded time. Most CSMs can’t do that inside their existing book of business — they’d have to drop accounts or go part-time on implementation. That’s a losing trade on both sides.
The ACV math is unforgiving:
- $50K+ ACV: FDEs are profitable and necessary. Do it.
- $10K–$50K ACV: Hybrid models with some systematized automation can work.
- Under $10K ACV: You must systematize the implementation process or the economics never close. You can’t afford a human doing 30+ days of embedded work per customer at that price point.
The companies that crack the bottom two tiers will win the SMB AI market. Most haven’t figured it out yet.
What the data says about the upfront investment
High-growth AI-native companies are already running a fundamentally different post-sales org than traditional B2B companies:
- Traditional B2B post-sales: roughly 60% CSMs, 20% support engineers, 15% implementation specialists
- AI-native post-sales: roughly 25% CSMs, 40% FDEs and implementation specialists, 15% ML/AI specialists, remainder support and data engineers
That 40% going to FDEs isn’t overhead. It’s what creates time-to-value, which drives expansion, which drives the business.
The payoff is real. One VP of Customer Success at a Series B AI company put it plainly: “We spend 3x as much in the first 90 days as a traditional SaaS company would. But our churn is half, our expansion rate is double, and our customers require 40% less ongoing support after month 3. The math works.”
Compare that to traditional SaaS: 12–18 month adoption curves, expansion that took years, multiple CSM touches to drive adoption that never quite materialized. AI-native companies with heavy Day 1 investment see expansion in quarters, not years. The difference between “We had you getting value in Week 2” and “We had you renewing in Year 2” is the whole business.
Palantir — which essentially invented the modern FDE model — recently announced they’ve reduced deployment times by over 90% using the next generation of automation plus forward deployed engineers. That’s remarkable. It also tells you something important: even the best operator in this model, after years of refinement and at massive scale, still requires humans in the deployment loop. The FDE isn’t a temporary workaround until the product gets easier to deploy. It’s the model.
The 5% who can make the switch:
There are CSMs who can become FDEs. They have actual engineering backgrounds or deep technical AI domain expertise. They’ve spent time in the product, not just around it. These folks can sometimes make the transition into embedded implementation roles.
But they’re rare. And even then, you’re usually better off hiring purpose-built FDEs than trying to retrain relationship managers into builders. The skills that make someone a great CSM — empathy, account management, renewal instinct — don’t transfer into daily debugging sessions and workflow architecture.
What actually works
Hire one strong FDE. Embed them with your top 3–5 customers. Document exactly what they do. Then systematize it.
Once you’ve cracked the implementation process and have a real playbook, you can scale with CSMs or implementation specialists who’ve seen it work. At that point, the CSM’s job is maintaining the relationship after deployment is solved — not fixing deployment itself.
Trying to turn your existing CSMs into FDEs usually tanks both roles. You lose the relationship coverage. You don’t gain real implementation depth. And your best CSMs get frustrated doing work they weren’t hired for.
Get one great FDE first. Build the playbook. Then scale.
