When Product-Market Fit Isn’t Big Enough: Mpathic’s Second Act
Feb 12, 2025
David Brennan MBA
Most founders dream of product-market fit. But what happens when you actually get it — and then realize the market isn’t big enough to go the distance?
That’s exactly what happened to Grin Lord, co-founder and CEO of Mpathic. Her team built something genuinely important — an AI platform that reviews and flags risk in human conversations — and landed a foothold in the life sciences space. They had real revenue, real use cases, and a killer moat in a market that needed them.
But there was one problem: the TAM wasn’t big enough to support venture-scale growth. And that’s when the next stage of founder thinking had to kick in
Key Takeaways
Mpathic found PMF in the clinical trial space — but the vertical was too narrow to scale VC-style
Their early traction came from the psychedelic medicine space, where long-form patient recordings were manually reviewed for safety
They realized the same manual-review problem existed further down the pipeline — in endpoint assessment
Now they’re building new models and expanding into broader life sciences, where human bottlenecks are everywhere
Their strategy: go where humans can’t scale, and let AI do the heavy lifting — with oversight, not replacement
From Psychology to Product-Market Fit
Grin started her career as a clinical psychologist, training therapists how to listen and communicate more effectively. But she knew early on: you can’t teach empathy in a 2-day workshop. It requires feedback, reflection — and repetition.
So she built a speech analysis pipeline back in 2008 — way before LLMs or AWS Transcribe — and used it to coach clinicians on how to improve. That evolved into her first startup, and then into Mpathic: a company that uses AI to analyze real human conversations and flag what's actually happening — emotionally, behaviorally, and clinically.
The First Win: PMF in Psychedelic Trials
When they launched Mpathic, they sold the product horizontally. Sales, HR, healthcare, coaching — anywhere conversations happen. And it could work everywhere. But early on, Grin and her team realized something founders learn the hard way: broad applicability doesn’t equal focus.
So they picked a lane: clinical trials.
Specifically, the psychedelic medicine space — where a single patient might have 50+ hours of audio or video recordings that need to be reviewed for safety and compliance. Human reviewers were doing that manually. Slowly. Inconsistently.
“Only 10% of recordings were getting reviewed — and even that was too much for teams to keep up with.”
Mpathic stepped in with automated models trained by real psychologists. It worked. Sales started repeating. Product-market fit was locked in.
The Problem: PMF, But Not Enough TAM
This is the part most founders don’t talk about.
You do the hard thing — you find fit. But then the ceiling shows up. The psychedelic trial space was just too small.
“It was a $6B market. Big, but not venture-big. VCs want more.”
So now what? Walk away? Rebuild again?
Not for Grin. Instead, she zoomed in — and found another bottleneck hiding in plain sight.
The Next Move: Endpoint Assessments
One of Mpathic’s early customers said the quiet part out loud: “This is great for safety monitoring. But where we really need help? Endpoint review.”
That was the unlock.
In clinical trials, endpoint assessments are structured interviews that help determine whether a treatment is working. Think: Is this patient showing signs of depression? Anxiety? Cognitive decline?
These assessments are:
Conducted by humans
Scored manually
Prone to inconsistency and rater bias
Often too long or too messy to review fully
“One rater says this person’s a 3. Another says they’re a 1. That variability can sink a trial.”
Sound familiar? It was the same problem as their first use case — just further down the chain. More manual review. More fatigue. More human variability.
So Mpathic expanded. Now they provide real-time quality assurance on structured clinical interviews. Not replacing raters — just flagging when things go off track.
The AI Pattern: Go Where Humans Break Down
This is the playbook Mpathic is running:
Look for long recordings
Spot the human bottleneck
Build AI to augment, not replace
Scale with oversight, accuracy, and auditability
“We’re not taking jobs. We’re letting doctors do their real jobs — while AI handles the stuff no one wants to do anyway.”
And they’re not guessing. Mpathic's models are trained and validated by domain experts. They’ve gone from a team of 30 psychologists to a core group of specialists — and the AI is only getting stronger.
What Other Founders Should Take From This
“PMF is a milestone — not a finish line. And finding it in the wrong-sized market can be just as dangerous as not finding it at all.”
Grin didn’t panic. She listened. She found a near-adjacent use case. She validated it with enterprise buyers. And now she’s back on a path that could take Mpathic from specialist vendor to category leader.
Final Word
“AI shouldn’t replace clinicians. But it should replace the parts of their job they don’t want — and can’t do at scale.”
Book a Free AI Assessment if you’re building in healthcare, running an AI company, or navigating the jump from PMF to market expansion. I’ll help you figure out where the real bottlenecks are — and how to turn your product into a platform that actually scales.
