My friend asked me to roast some beans for her. No pressure, right? Except… she had very specific ideas about what she wanted, and I had limited beans to work with. This PHU615 from Chiang Mai turned out to be way more interesting than just “help me roast this”—it was actually about learning to roast someone else’s vision instead of my own.
The Bean (and the Gentle Request)
So here’s the deal: PHU615 is a natural fermentation bean from the Phupanna farm in Chiang Mai. When you crack open the bag, the aroma promises some wild stuff:
- Lychee
- Rose
- Strawberry
- Citrus fruit
- Sweetness
Pretty fancy, right? But here’s the thing my friend emphasized: she wanted medium-light roast. Not because she’s trying to be difficult, but because she actually wants to taste those fruity, floral notes. Darker roasts? They tend to bury all that delicate stuff under chocolate and bitter roast character.
The beans themselves were pretty normal—10.8% moisture, 0.72 density. Nothing wild. Standard 500g batch.
The Shortcut (And Why I Took It)
Here’s where I broke my own rules: I didn’t do a baseline roast.
Normally I’m all about capturing how a new bean behaves—dry end, first crack, all that data. But honestly? With only 500g to work with, using half the beans for testing felt wasteful. Plus, I had old roast data from PHU915, another bean from the same farm.
The logic: same farm, similar processing, probably similar behavior, right?
My Old PHU915 Numbers:
- DE: 149°C
- FC: 181°C
- SC: 207°C
I basically said “same farm, I’m using this profile” and went for it. Risky? Maybe. But it felt smart given the circumstances.
The Roast: Following Yesterday’s Recipe
I used the old PHU915 as my template and adapted it for medium-light:
- Dry end around 149-150°C? ✓ Got it
- First crack landed around 181-184°C? ✓ Check
- Kept Maillard fairly short to preserve brightness
- Dropped early (before 199°C) to stay light
Final reading: Agtron 59 = medium-light. Exactly what she wanted.
Not gonna lie—it felt good to hit the target roast level. But there was this nagging feeling…
The Maillard Thing (My Regret)
Here’s where I get a bit in my own head: for a natural fermentation bean with lychee, rose, and strawberry notes, I kinda wanted more Maillard time to let those fruity compounds get properly sweet and integrated.
But here’s the reality: I still can’t seem to extend Maillard without accidentally sliding into medium roast (yes, this is still my recurring problem). So I kept it conservative to preserve that medium-light level she wanted.
Trade-off: I got the roast level right, but the sweetness complexity probably isn’t as rounded as it could be.
Is that 100% success? Nah. More like… 70% happy with my own execution, even though the result hit her target.
The Cup Test: Where Things Got Real
My friend tried it as a latte—and she actually liked it. Said it works well with milk, which tracks for a medium-light roast.
She wasn’t sure about espresso, which… yeah, medium-light can be tricky in espresso. Might pull thin or taste under-extracted depending on your machine.
I grabbed the V60 to see what was really going on with this bean.
Verdict: A bit sour (totally normal for medium-light), but—and this is the important part—I could clearly taste strawberry. Like, unmistakably fruity. The natural fermentation character came through.
That’s the win right there. Forget my Maillard regrets. The fruit showed up. My friend was hunting for lychee and strawberry, and the strawberry at least made an appearance. Mission accomplished.
The Real Learning: Farm Data Actually Works
Turns out using old data from the same farm paid off. PHU615 behaved similarly to PHU915 thermally, which meant my old profile wasn’t some random guess—it was actually relevant.
That said… if I’m being honest? A proper baseline roast would probably have unlocked a longer Maillard window without overshooting medium. Maybe an extra 30-45 seconds that could have made the sweetness pop more.
But here’s the thing: my friend is happy. She got her medium-light with fruit notes. She didn’t ask for “optimal sweetness integration.” She asked for fruity and bright, and that’s what she got.
Roasting for Someone Else (A Mindset Shift)
This whole experience taught me something: when you’re roasting for someone else’s vision, your standards change a little.
Like, my brain kept thinking “this could be better with longer Maillard” and “I rate this 70% because of X and Y.” But my friend actually enjoyed it. In her world, this roast is 100% successful because it’s what she wanted.
That’s a weird mental shift from roasting purely for myself where I’m chasing 4/5 or 5/5 ratings. Sometimes “good enough for what they wanted” beats “perfect by my standards.”
Would I Do It Again?
Yeah, definitely. Though next time I’d probably do a quick baseline roast if there’s enough beans. That 30-second Maillard extension could unlock better flavor, and it’s worth the investment if someone’s actually requesting something specific.
But the farm data shortcut? Smart move when you’re tight on beans and working with a known source. Would do again.
Roast Vibe: Medium-light hit target, strawberry came through, friend happy
My Rating: 70% (solid execution, could’ve optimized more)
Friend’s Rating: Seems like 90%+ (enjoys it as latte, enjoys the fruity notes)
Lesson: Sometimes hitting someone else’s goal is more important than hitting your own perfectionism
Quick Profile:
- Bean: PHU615 (natural fermentation, Phupanna farm, Chiang Mai)
- Amount: 500g
- DE: ~150°C (based on old PHU915 at 149°C)
- FC: ~183°C (old PHU915 was 181°C)
- Development: Short (aiming to stay light)
- Drop: ~195-196°C (before the old SC marker of 207°C)
- Final: Agtron 59 (medium-light) ✓
- Tasting: Strawberry notes clear, slight sourness (medium-light normal), great hand drip expression
- Friend says: Loves it as latte, tasty fruity vibe
