THE BUILD UNIT
Build the smallest complete thing that someone would actually use on its own.
How to ship new software when product–market fit is unknown.
Understand 1G in 60 seconds1G IN 60 SECONDS
Build the smallest complete thing that someone would actually use on its own.
Your big vision — the full platform you dream of building.
If a feature can't stand alone as a tiny, useful product → don't build it yet.
WHY
Most MVPs fail because they're 30% of 10 things instead of 100% of 1 thing.
A classroom platform
Don't build "a full classroom platform" (10G).
Build "teachers can post one assignment" (1G).
If teachers use it → add the next 1G. If not → pivot.
That's it. The rest is details.
THESIS
One-Gram of Tech (1G) is a methodology for building new software products where product–market fit is unknown.
The core claim:
Ambitious "10G" products should be built as a stack of 1G units – tiny user-facing features that could stand alone.
If it cannot stand alone, it is support work or waste.
10G is the narrative; 1G is the build unit that prevents all-or-nothing releases.
ASSUMPTIONS
If these premises are rejected, the methodology does not apply.
Engineering time is scarce capital, regardless of cash on hand.
User value is uncertain for genuinely new ideas – many guesses miss.
Being wrong is expensive, and extra scope multiplies the cost.
THE 1G GATE
"If this were shipped on its own, as a tiny standalone product for a specific segment, would it have a plausible path to adoption and revenue on its own?"
A feature is only a 1G if the reward for using it occurs in the same cycle as the effort required to use it. If the user must perform an action today for a reward in six months, it is not a 1G; it is a 10G dependency.
A candidate feature counts as a 1G only if all of the following are true:
There is a clearly defined type of user; not "everyone" but a concrete segment.
The problem has a visible cost in time, money, risk, or missed opportunity today.
There is a precise definition of "done" in which that problem is solved end-to-end for that user, without requiring several other unfinished pieces to be useful.
It is realistic to imagine a simple landing page that promises only this outcome, and a segment that would adopt it on its own and plausibly pay for it, even if nothing else existed.
Informal version: "Would this make sense as a micro-SaaS if it had to?"
GROUND-ZERO RULE
Every product has a ground-zero unit of value. This is not a design preference. It is Dependency Logic. If Feature B depends on Feature A, building both at once introduces technical noise. We isolate the foundation so your feedback is 100% interpretable.
the user must be able to apply.
the user must be able to create a profile.
the user must be able to upload.
users must be willing to give their email.
If this foundation does not work, nothing else can be validated.
When this step is skipped, you lose the ability to diagnose problems. The methodology avoids this failure mode entirely. Each version of the product is designed to answer a single yes/no question:
10G VS 1G
Your 10G vision is your direction, not your first thing to build.
10G statements are useful as direction and narrative. The problem arises when a 10G vision is treated as the first unit of execution.
1G does not reject 10G. Instead:
Start from the 10G statement.
Break it into concrete pains with clear costs.
Define the smallest complete outcomes for each pain.
Select one 1G to build and ship.
Repeat, stacking additional 1Gs over time.
The 10G product becomes the accumulation of validated 1Gs plus the necessary support/glue, not a single massive first release.
CAPITAL PROFILE + HIT-RATE
The difference is in how capital is deployed.
"One broad, entangled experiment. High cost to reach meaningful validation."
"Several narrow, explicit experiments. Cost to reach validation often drops to this range for a single, sharply scoped 1G."
Failure pinpoints a specific hypothesis rather than "the platform". Success is anchored in a specific behaviour users adopt or pay for.
The 1G-first mindset turns engineering time into Learning Capital.SUPPORT WORK, COMPLEMENTS + GLUE
Not all important work can or should pass the 1G gate. Examples include infrastructure, security, compliance, and refactors.
"Does this materially increase the expected value, robustness, or coherence of one or more real 1Gs, and therefore of the 10G product they form?"
The methodology claims that only work justified as a 1G or support/glue that clearly raises the expected value of specific 1Gs should receive engineering capital in the early stages.
THE OPERATIONAL QUESTION
In practice, the methodology reduces to repeatedly asking:
"What is the smallest complete problem being solved here, for which specific users, that they would genuinely care enough about to adopt or pay for on its own – and does this feature meet that bar?"
It becomes a 1G building block of the 10G vision.
It is either support work for a different 1G, or it should not be built.
THE MATHEMATICAL SPINE
The technical depth is not a separate methodology. It is the machinery underneath the same rules you just read.
G → {G₁, G₂}We treat the ambitious "10G Vision" as the parent graph and decompose it into "1G" subgraphs. Each unit is a self-contained module.
Partitioning a complex graph (G) into edge-disjoint subgraphs (H).
Toposort(G)The "Ground-Zero Rule" is a literal DAG implementation. We identify the Root Node, the foundation upon which everything else depends.
A DAG is the mathematical requirement for Topological Sorting.
Cut(V) → isolate(vᵢ)The 1G Gate acts as a Vertex Cut. By shipping one atomic unit at a time, we "cut the edges" to external dependencies.
In dense networks, errors propagate through adjacent edges.
G = ⋃ᵢ CliqueᵢIterative Graph Synthesis. The final 10G product is the accumulation of validated 1G units. Stable by design.
Mimics Clique-Based Reconstruction, building a Supergraph from cliques.
Nov 21, 2025
Start free with 15 credits. End up with something that actually works, that you own, that you understand.
Your code, your server, your domain. Not ours.
Live on the actual internet, built with real tools. Not a demo.
Every decision and every step comes with the why.