Carbon paste comes in a tube and looks like grease. It is a friction enhancer, and it exists so you can use less torque rather than more.

Slipping at spec means more friction, not more torque

The single most useful thing on this page. Carbon assembly paste is a friction enhancer rather than a lubricant: microscopic abrasive particles in a non-greasy base, whose purpose is to increase friction between two mating surfaces so components can be secured firmly at lower and safer torque settings. That means the correct response to a slipping carbon seatpost or handlebar is to improve the interface, not to exceed the specification. A practitioner who has run into it repeatedly says it plainly: if you are at the suggested torque and slipping, you need to increase the friction. The torque figure printed on a carbon part is a ceiling rather than an opening bid.

Grease is wrong here, and the worst part is the feedback loop it creates

The consensus stated in the sourcing is unambiguous, that no grease should be used on carbon seatposts, and there are three separate reasons. Grease reduces friction, which is exactly backwards when you need a post to hold, and it can cause a seatpost to gradually slip down into the frame even with the clamp at maximum torque. Carbon components have specific lower torque specifications, and grease makes adequate clamping force harder to achieve without over-tightening, which crushes or weakens the fibers. And some lubricants can, in rare cases, interact negatively with the resin matrix over time. The middle one is the dangerous one because it is a loop: grease makes the part slip, slipping invites more torque, and more torque is what damages carbon. The grease does not break the part directly, it manufactures the temptation that does.

Mark your insertion depth with tape before the paste goes anywhere near it

A practical warning that follows directly from what the paste is, and that nothing on a package mentions. Because the compound is deliberately abrasive, a practitioner warns to be wary of using it with anodized components, noting that it does not take much to scratch the anodizing, and recommends working out your install depth and wrapping tape at that point before applying paste, so you do not accidentally go too far down the seat tube, have to pull back up, and find visible scratches. The tape prevents over-inserting past where you need to be. It costs two minutes and it is the difference between a clean install and a permanently marked post, on a component where the marks are exactly where you will see them every time you adjust the saddle.

Paste may be a one-time application, so reapply after any disassembly

A practitioner finding rather than a manufacturer claim, and it is presented as such. Somebody reporting their own experience says the compound works to some extent but that it is a one-time deal: once you torque down bars or a post, if you subsequently loosen the connection or it slips, you need to clean and reapply or it will be more likely to slip in the future. Their theory is that the micro silica gets crushed after a few compressions and can no longer provide the same friction. That is a theory rather than a measurement, and it is mechanically plausible enough to be worth a habit: after any disassembly, clean the old paste off both surfaces and apply fresh rather than retightening onto particles that may already be flattened.

Experienced people disagree about whether to use paste at all

Worth knowing so that a shop recommending one approach or the other is not treated as suspect. One practitioner advises using nothing initially: tighten the seatpost clamp to no more than the recommended maximum torque, ride the bike for a while, see whether it slips, and only if it does apply a little assembly paste. Another has moved to dry carbon-on-carbon for most things, with both surfaces spotlessly cleaned with alcohol. Both are people who work on these regularly, and paste-by-default is clearly not universal practice. What both camps agree on, and what actually matters, is the same thing: not grease.

The right answer depends on the material pair and the load

The reason grease exists at all in bicycle assembly is galvanic corrosion between dissimilar metals, which is why it is correct for an alloy post in a metal frame and wrong in carbon. Practitioners split it further by load: one uses a tiny bit of grease on an aluminum dropper post in a carbon frame, and on carbon bars in an aluminum stem where there is not much rotational force, specifically for corrosion control, while treating high-rotational-force joints differently. So the honest version is not a single rule but a two-part question: what metals are meeting, and how much is the joint being asked to resist turning. And carbon paste does double duty here, since it is described as protecting the post and frame from corroding together inseparably as well as preventing slipping.

Clean means clean, and a torque wrench is not a refinement

Two things that everything above depends on. Friction is a property of the interface, so the interface has to be right: both surfaces cleaned with alcohol, old paste removed, old grease removed, grit removed. A dry carbon-on-carbon joint only works if it is genuinely spotless, and paste applied over contamination is paste doing nothing. And the wrench is not about precision for its own sake. It is because the carbon failure mode is crushing rather than stripping, which means the part does not tell you when you have gone too far. Feel works on a fastener that protests. It does not work on a material that quietly gives up some of its strength and continues to look correct.

Recheck after the first rides rather than pre-tightening

A habit that resolves the temptation this whole page is about. Clamped joints settle in the first hours of use, which is the same phenomenon as the cable settling described on the drivetrain page in this category. The wrong response is to anticipate it by adding torque at assembly, because that is exactly the over-tightening that damages carbon. The right one is to torque to the printed figure and then recheck after the first rides, adjusting back up to the same figure if the joint has relaxed. That way the number is never exceeded and the joint still ends up correct. And if a joint keeps needing attention, or starts creaking, that is a reason to have somebody look at it rather than a reason to keep going.