Fourth cycle done, fifth one coming on June 10th. The detailed protocol note will land that day. This post is the part that actually matters: what changed in the design, and why.
The short version: 5.0 Light is the first cycle where I am running fewer compounds on purpose. Not because of budget, not because something went wrong. Because the previous stack had drifted, and the drift was instructive.
The drift
NeuroForge has always officially been about structural building rather than stimulation. But each cycle I added something. Cycle 4 was the maximum version: Lion’s Mane back, DHA doubled, Rhodiola swapped in for Ashwagandha, plus Alpha-GPC, Bacopa, Thorne 2/Day, Gotu Kola in the mix. Each compound had a plausible mechanism and a paper. Adding is easy.
Cycle 3 had already taught me something I did not fully metabolize at the time. It ran through autumn 2025 alongside a heavy training block and the Berkeley exec AI program, and it was subjectively the weakest cycle I had done. Not because the stack stopped working. Because under high physical and cognitive load, day-to-day signal flattens. The body has bigger inputs than anything you swallow in capsules. The structural work continues quietly, but the perceived effect drowns in noise.
That points at a sharper question: if half of the stack is modulating state (adaptogens, acute cholinergic levers, a strong multi), how do I separate what is structural building from what is just lever-pulling? Answer: I cannot. Not from inside the cycle.
5.0 Light is the experiment in subtraction.
Three layers, not one stack
The cleanest reframe I got out of the last few weeks of thinking was to stop treating “NeuroForge” as a single category and split it into three.
Structural brain support is what neurons are literally made of. Phospholipid membranes, synaptic density, myelin, dendrites. Slow, quiet, cumulative. You do not feel it day-to-day. It either compounds or it does not. Garden metaphor: soil, roots, plant structure.
Mitochondrial infrastructure is the cellular energy capacity itself: the components of the electron transport chain, the signals to build or maintain mitochondria. Also slow, also structural, but in a different dimension. Affects everything that demands ATP, which is brain, heart, kidneys, and muscle. Garden metaphor: the power plant on the farm.
Fuel and activation is the transport and acetyl-group logistics that feed those mitochondria. More immediately “felt”, more activational, more like classic supplementation. Garden metaphor: getting fuel to the generators.
Separate from all three sits the fourth category that 5.0 Light removes almost entirely: state modulators. Adaptogens, acute cholinergic levers, anything that shifts how I feel rather than what I am built from. These are useful tactically. They are noise in a structural cycle.
What stays, by layer
Structural brain support
| Compound | Dose | Timing |
|---|---|---|
| DHA / EPA (ProDHA) | 2 caps | Evening, with fat |
| Uridine / UMP | 250 mg | Morning, with fat |
| Phosphatidylserine | 100 mg | Evening |
| Magnesium L-Threonate | 3 caps (144 mg Mg) | Evening |
These are the spine of every cycle I have run. They never come out. The reason I am explicitly listing them under “structural” is to stop pretending they belong in the same conversation as Alpha-GPC.
Mitochondrial infrastructure
| Compound | Dose | Timing |
|---|---|---|
| CoQ10 | 100 mg | Morning, with fat |
| PQQ | 10-20 mg | Morning, 4-5x per week |
Both stay. PQQ in particular nearly got cut in an earlier draft because it sounded like an “activator”, but mechanically it is closer to mitochondrial biogenesis signaling than to state modulation. It belongs with the structural side, even if it lives one layer over. Cutting it would have been ideological, not principled.
Fuel and activation
| Compound | Dose | Timing |
|---|---|---|
| ALCAR | 500 mg | Morning, 3-4x per week, optional |
| Creatine | standard | Restarting in ~4 weeks (currently on break) |
ALCAR is the most “felt” of the remaining compounds and the one I hold most loosely. Not every day, not on full rest days where I want a clean read on actual fatigue. Creatine restarts roughly in sync with the cycle. That coincidence is worth flagging on its own.
Bacopa, separately
Bacopa is not on the cycle. It is its own 12-week memory block.
| Compound | Dose | Timing |
|---|---|---|
| Bacopa | 300 mg | Daily, June through end of August |
What got cut, and why
Alpha-GPC. The most useful one to remove, precisely because it works. A strong acute cholinergic lever is exactly the kind of thing that distorts the baseline I am trying to measure. If I want a sharper morning, I can take it tactically on a specific day. I do not need it daily as part of a “structural” stack.
Rhodiola and Ashwagandha. Out together. Adaptogens modulate the HPA axis and stress response by design. That is their entire point. Which means they actively shift the homeostasis I am trying to preserve and observe. Removing them is the riskiest cut psychologically (Rhodiola felt good in Cycle 4) and the cleanest one philosophically.
Thorne 2/Day. I saturate quickly on B12, D3, iron. Running a strong multi every cycle is insurance against a problem I do not have.
Lion’s Mane and Gotu Kola. Not bad compounds. Just not load-bearing. If the cycle works without them, that is information.
Five compounds removed. Zero added. The stack is roughly half its previous size.
Bacopa: not an adaptogen, not on the cycle
The reframe I owe myself a note on. I was treating Bacopa as an adaptogen because it sits next to Rhodiola and Ashwagandha culturally (Ayurveda), but the mechanism is different. Bacosides act on membrane lipid composition in the hippocampus, dendritic arborization, BDNF modulation, antioxidant activity in neurons. Mild cholinergic, but not via receptor manipulation. Those are structural building mechanisms. No tolerance buildup in the cyclical sense, no HPA modulation.
The implication is that Bacopa does not belong in NeuroForge cycle ontology at all. It belongs with DHA, magnesium, creatine: chronic substrate. The RCT evidence is concentrated at 12-week durations at 300-450 mg/day, with the strongest signal on memory free recall.
So: Bacopa runs continuously from June through end of August, then washes out. The cycle starts and stops without it. Cleaner that way.
The creatine attribution problem
Creatine restart lands roughly when 5.0 Light starts. Two structural changes beginning at the same time, plus the implicit “new cycle” expectancy effect, makes attribution noisy for the first two to three weeks. Worth flagging so I do not credit (or blame) the wrong thing later.
Timeline
- June 10th: start NeuroForge 5.0 Light, start Bacopa block, creatine restart roughly coincides.
- June and July: full tactical block.
- End of August: finish Bacopa to close the 12-week window. Most of the rest comes off around the same time.
- September and October: reset. Homeostasis window.
- Q4 2026: if a strategic cycle returns, it gets redesigned from this baseline, not built by reinstating the old stack.
What I am actually testing
Can I keep the structural and mitochondrial benefits of NeuroForge while removing the compounds most likely to distort homeostasis?
If yes, 5.0 Light becomes the new default and the bigger stack becomes a tactical option for specific seasons. If no, that is also useful: it tells me which removed compounds were doing structural work I underestimated, and they come back one at a time. Not the whole old stack reinstalled by reflex.
Either outcome beats the current state, which is “I run a big stack and trust the story.”
Not more inputs. Cleaner signal.