During Intel® Liftoff Days - an acceleration week filled with mentorship, technical deep dives, and pitch prep - one session landed with unusual clarity. In “How to Sell Your AI Product,” Mohamed F. Ahmed, Senior Director of AI Cloud Infrastructure at Intel, spoke to founders not as a tech executive but as someone who had walked their path. Seven times, to be exact.
“You’re not building a product,” he said. “You’re building a business.”
What followed was a practical, honest, and unusually useful breakdown of how to sell AI - without hype, and without trying to copy how SaaS has been sold for the past two decades.
That distinction framed everything that followed.
Get Clear on the Problem You Solve
Sales, Ahmed said, isn’t about forcing a product on someone. It’s about solving a real problem. For technical founders, that means shifting focus from what the product is to what it does and how that helps someone else do their job better.
Before founders start pitching to anyone, they should be able to explain:
- The Problem: Be painfully specific. Who feels it? When? Why now?
- Current Workaround: What are they doing instead?
- Your Product’s Role: Where do you fit into their workflow?
- Measurable Impact: Time saved? Money saved? Revenue gained?
Ahmed framed sales as “a problem-solving motion.” That means your job is to align your product with an existing pain, not invent a new one just to fit your tech.
Exercise: Ask 10 people in your target audience what their biggest time suck is. Don’t pitch. Just listen. You’re doing customer development, not validation.
AI Is Not SaaS. Stop Selling It Like It Is
The way we sell software is changing. AI doesn’t fit neatly into the old “seat-based license” model. That’s not just a pricing issue, it’s a shift in how buyers understand value.
“AI is killing SaaS”
According to Ahmed, AI agents are taking over workflows that used to require teams of people clicking through dashboards. Instead of traditional interfaces, customers want outcomes. So pricing should reflect that.
Instead of pricing by:
- Number of users
- Monthly subscriptions
- Feature tiers
Try pricing by:
- Tasks completed (e.g. number of contracts reviewed)
- Outcomes achieved (e.g. qualified leads delivered)
- Volume processed (e.g. hours of content analyzed)
Founders should ask:
- What part of the business are we helping scale or shrink?
- Can we tie pricing to something they already measure?
This shift favors startups that can move fast and rethink the math. “You’re selling AI labor,” Ahmed said. “So price it accordingly.”
Show the Value (Don’t Just Describe the Features)
If you’re talking to a buyer, they don’t care that your transformer model has fewer parameters or that you wrote the backend in Rust. They care about one thing: what it does for them.
Founders should anchor their messaging in four outcomes buyers care about:
- Scalability – Can I do more with the same team/resources?
- Efficiency – Can I get the same results faster?
- Cost reduction – Will this help me cut overhead?
- Better decisions – Does it improve quality, speed, or accuracy?
If your AI product can check more than one of these boxes, you’ve got a shot. If it doesn’t check any, you’ll struggle to get traction regardless of how smart your tech is.
Advice: If your deck still opens with “We built a large language model that...” rewrite it. Start with the problem and show how you impact the bottom line.
Build Trust Before You Automate
Many AI founders over-index on what their product can do, instead of what users are ready for.
Ahmed shared a cautionary tale from a previous startup. He’d built an autopilot tool for cloud infrastructure that promised huge savings. But adoption stalled. Why?
“Developers told me, ‘I don’t trust a machine with my job.’”
Even when automation is technically sound, adoption is emotional. Trust has to be built and often that means giving users control, even if it slows things down.
Ways to build trust:
- Add manual override steps early on
- Let users review and approve AI suggestions
- Show them your model’s confidence score or audit trail
Don’t rush full automation. Start with decision support, build confidence, then offer autopilot.
Vertical AI Will Win… If You Actually Know the Vertical
AI founders with deep knowledge of a niche market often have an edge. That’s not because their models are better—it’s because they understand the mess behind the scenes.
Ahmed emphasized the power of industry-specific insight:
“If you have something unique about how that business works, that’s what gives you an advantage. Not just better tech.”
To stand out:
- Interview 20 people in the same industry, not just “potential users”
- Learn their language, reporting cycles, compliance burdens, and red tape
- Build your UI and workflow to reflect how they already work
Two Audiences, Two Playbooks
When selling AI, you’re often talking to both a business leader and a technical stakeholder. Their concerns aren’t the same.
Business buyer (CEO, VP, Director):
- Focuses on ROI and outcomes
- Asks: Will this make me money or save me money?
Technical buyer (Engineer, IT lead):
- Focuses on trust and accuracy
- Asks: Will this break anything? Can I control it?
Advice: Build two versions of your pitch. The first should be visual and results-driven. The second should cover model accuracy, risk mitigation, and integrations.
Speed Is the Only Real Advantage You Have
As the session wrapped up, Ahmed asked a simple question: What’s your startup’s most important asset?
Founders answered: ideas, data, people, value prop.
Wrong.
“The only advantage you have is speed. Big companies have more people, more data, more money. You can learn faster. That’s your edge.”
Startups will never outspend, out-staff, or out-data the incumbents. But they can:
- Test pricing models in days
- Change product messaging overnight
- Abandon dead-end features without approval chains
For early-stage AI startups, that means:
- Talking to customers before your product is fully built
- Testing pricing and positioning early (ie ship fast)
- Adapting based on what you learn and not what you assumed
Talk to Strangers, Not Just Supporters
Ahmed recommended The Mom Test, a book that’s made its way through many accelerators. The premise is simple: your mom will lie to protect your feelings. A real customer won’t.
How to run better interviews:
- Don’t ask if they like your product
- Ask what tools they pay for
- Ask what they’ve tried that failed
- Ask what they hate doing each week
If your product doesn’t show up in their top 5 headaches, it’s probably not urgent.
A Framework to Make Faster Decisions
Ahmed introduced the ICE Scoring Framework to founders, which ranks ideas by:
Impact: How much value will this create?
Confidence: How sure are you?
Ease: How quickly can we test it?
The point isn’t to be right. It’s to be fast, learn early, and stay alive long enough to get it right.
And a key takeaway from NSF’s I-Corps playbook: talk to at least ten target customers before building your product. Learn their real problems first.
Final Thoughts
“This is just a teaser,” Ahmed said at the end. “What you do next is what matters.”
Founders left the session with no illusions. Selling AI isn’t about building something cool and waiting for people to come. It’s about listening, adjusting, proving value, and moving fast.
Or, in Ahmed’s words:
“You’re not here to build a product. You’re here to build a business.”
And that means starting with the problem, not the pitch.
Want more insights like this for your AI startup? Join Intel® Liftoff and connect with experts, peers, and the support you need to grow faster.
About The Speaker
Mohamed F. Ahmed is a Senior Director of AI Cloud Infrastructure at Intel, where he leads strategic initiatives to advance cloud-based AI solutions. With over 20 years of experience in technology and several successful startups, Mo brings a founder's perspective to enterprise AI development. His expertise spans cloud infrastructure, SaaS transformation, and AI product strategy. Mohamed holds a Ph.D. in Computer Science and has led multiple technical organizations through periods of rapid innovation and growth.
Related resources
Intel® Tiber™ AI Cloud - Cloud platform for AI development and deployment
Intel® Gaudi® 2 AI accelerator - High-performance AI training processor designed for deep learning workloads
As a Software Tools Ecosystem Specialist at Intel, I’ve had the privilege of working on the dynamic GenAI initiative. My focus is on driving engagement with the software developer audience. I'm a proud team member of the Intel® Liftoff for Startups and the Developer Engagement, Relations, and Studio team.