I've always had a vague-ish idea of how Alchemy would work, but after hammering poisons into an acceptable, poisony shape, I think it's viable to take a similar tack with Alchemy as well. It was always sorta my intent all along -- honest! -- to make Alchemy a mix of Magecraft (adding effects together to create a potion) and Craft/Artifice (making stuff beforehand that's good just about anytime), and now it looks like we're in a good position to do that.
The only trouble is that, as written, making a poison is way harder than making a potion. If you want to make a potion that lets you see in the dark, that's just one or two effects (+2 Alertness in dim light and/or Aspect: "Night Vision"), which means only an Average or Fair effort is required to make it. Meanwhile, a poison of similar potency (let's say Fair Power, Mediocre Endurance, Mediocre Stealth, for a total of +2) is almost prohibitively weak. One attack at +2? Bah. They're hardly equivalent.
But poisons are made of mundane stuff, for the most part. There are weak poisons out there that are both cheap and, given a modicum of training, easy to make. That's as it should be, if you ask me.
Potions, on the other hand -- philters, elixirs, unguents, etc. -- are frickin' magical. They're a whole other ball of wax. There are no "Average" or "Fair" potions. What is this, K-Mart? No, even the humblest potion is at least Great (+4).
To build a potion, assemble effects just like you would for a spell, then add three. That's the total quality of the potion, and the target number to meet or beat with your Physik roll. Alternately, you can just spend a Fate Point, and voila: potion.
Laboratory quality and extra time bonuses work the same way here as they do when concocting a poison, and potions take just as long to make, as well. I.e., the basic time required starts at An Hour (A Few Minutes + 3), +1 step on the Time Increments table per effect added. Shifts on the Physik roll can be spent to reduce the time required (-1 step/shift), and extra time can be taken to increase the odds of success (+1 to roll/+1 step), either before the roll or retroactively. Add the quality of the lab to the roll as a bonus.
Even with three or four effects, if you take your time and have a decent lab, you're probably automatically succeeding even without spending a Fate Point (the Fate Point just lets you do it in the minimum time required, although getting a number of shifts on a good roll could conceivably mean doing it even faster).
This strikes me as pretty balanced with Magecraft. Both fields of magic let/make you spend a Fate Point to succeed without making a roll/taking Magic Stress. Magecraft lets you get off one spell for free, while Alchemy makes you pay for everything one way or another -- either in time or in Fate Points -- the trade-off being that Alchemy conceivably lets you pack a lot of effects into a single potion, to be used anytime, but you have to decide what that potion is ahead of time.
Speaking of which, there should probably be a way to limit just how many effects you can get into a single potion -- maybe a cap of Physik +1 effects/potion. As for how many potions you can have at once, that's a whole other issue. Perhaps that cap of Physik +1 could apply to how many effects total you can have across any number of potions. So with a Great (+4) Physik skill, for example, you could have five one-effect potions, two two-effect and one one-effect potions, one two-effect and one three-effect potion, and so on. This may be a little too game-balancey, but seriously, you know that if there isn't a limit, someone's going to show up with a hundred potions, and then you'll rue your player-trusting, story-gaming, group-hugging ways. Better to have a guideline that can be bent or broken than to not have anything at all, if you ask me.
Duration for most potions would probably be either Instant (e.g., healing) or 15 Minutes (e.g., a Potion of Strength) -- long enough to last through a typical scene.
Oh, and stunts. There'd have to be a few stunts to back up Alchemy, like getting the most out of a lab (treat lab as if its quality were one higher), being able to work faster (base time required is Half An Hour instead of An Hour), specializing in certain kinds of potions (pick one effect that doesn't count towards your normal effects cap), and so on. Have to work on that later.
(Just for laughs, maybe rolling four negatives on your Physik check could result in an explosion. These matters do involve risk, y'know.)
Now, the question arises: Can this be done for Artifice as well? I'd like to think so. It would be oh-so awesome to have consistency between those two, but I'll have to give it more thought. Off the top of my head, I'd likely set all durations at a base of 15 Minutes. Permanent items can also be created, but that requires spending a point of Refresh. The basic idea behind Artifice is that you can temporarily (or permanently) create or enchant items -- you need that tangible focus to work your magic. Maybe you inscribe a rune on a sword to make it sharper for a short time, or weave magical thread into a piece of cloth to enable whoever wears it as a blindfold to see in the dark. You get the idea.
Just to refresh your memory, or introduce you to it for the first time, here's the original post on Magcraft, Alchemy, and Artifice.
One last thing. I'm considering dropping all trappings for Magecraft that aren't elemental in nature, and then calling it, well, Elementalism. That means cutting out a lot of common effects (magical force shields, charm spells, etc.), but it also means a much tighter focus for mages -- and the latter really appeals to me. I'd rather build a limited system that models a particular thing than a one that has broad applications but no personality. Jury's still out.