Precon Upgrade (Pro): from $99 out-of-the-box to a deck you actually play

If you've ever opened a Commander precon and thought "this is fun but it's slow," you're not wrong.

Wizards designs precons to be playable out-of-the-box and reasonably balanced against each other. That's the right design goal — but it also means the strategy is intentionally diluted, the mana base is intentionally clunky, and the win conditions are intentionally telegraphed. The best part of buying a precon isn't the deck you got. It's the commander you got, and the 60-70% of the deck that you'll actually keep.

The other 30-40% is filler. Replacing that filler smartly is the difference between "this precon is fun" and "this is one of my favorite decks."

That's the loop we built AI Precon Upgrade for.

How it works in three clicks

1. Go to Decks → Upgrade a Precon.
2. Pick from the catalog. We pulled 700+ official preconstructed decks via MTGJSON — Commander products, Planeswalker decks, Theme decks, even older Intro Packs. Featured row highlights recent Commander releases (Tarkir Dragonstorm, Bloomburrow, Duskmourn, etc.).
3. Click. We resolve every card via Scryfall, save it as a new deck on your account, and immediately drop you into the Improve Deck panel — already pre-primed with the precon's context.

What the AI knows when you click "Analyze Deck":

• This is the unmodified <precon name> from set <X>, released <year>.
• The user's commander.
• The precon's strategy (inferred from the commander + theme).
• Your collection (so it can prefer cards you already own).
• Your budget cap per add (optional).
• The commander color identity hard constraint.

What the AI is told to do:

• Cut 5-10 obvious filler cards. Vanilla creatures, narrow situational removal, mediocre mana rocks. Keep signature theme cards.
• Add 5-10 high-impact replacements that respect the precon's strategy. Don't try to convert a tokens deck into a control deck.
• Stay within budget. If you say $10/add, no $40 mana rocks.
• Stay in your collection if you ask it to. Three modes: any cards, prefer-my-collection (default for precon imports), or only cards I own.

A worked example: "Sultai Arisen" from Tarkir Dragonstorm Commander

Brand-new precon. Commander is Teval, the Balanced Scale — a graveyard-recursion deck that wants to mill yourself, fill the yard with creatures, and bring them back.

Out of the box, this deck has:

• Some flavor-fail filler creatures (low-impact 3- and 4-drops).
• A few "exile your graveyard" cards that actively work against your strategy.
• Mediocre mana rocks (the precon staples — they're fine but not great).
• Underwhelming removal.

I ran the upgrade flow with prefer my collection and $5/add budget. AI suggestions came back in about 8 seconds:

CUTS (representative):

• Necropolis Fiend — exiling your own graveyard is counterproductive when your whole gameplan is recursion.
• Diviner of Mist — 4-mana 3/3 flyer with no payoff.
• Lord of the Forsaken — mana reduction is too restrictive to justify the cost.

ADDS (representative):

• Dread Return — a graveyard-recursion staple with built-in flashback synergy.
• Tragic Slip — high-impact removal that's trivially enabled by your sacrifice-heavy gameplan.
• Wayfarer's Bauble — better early ramp than the filler creatures.
• Witherbloom Charm — flexible utility (removal, drain, land recursion).

Every add is a card that costs less than $5, came from my collection where possible, and fits the precon's strategy rather than trying to retheme it.

Why this is different from "ChatGPT, upgrade my deck"

Three things that make the difference:

The AI's job is constrained, which is exactly the kind of task LLMs are good at. It's pattern-matching against thousands of upgrade discussions in its training data, filtered through your specific deck and your specific wallet.

It's also why the suggestions feel grounded — it's not making up cards or suggesting Sol Ring (the deck already runs Sol Ring; we tell it that). It's pruning the actual filler in your actual deck.

When to use it (and when not to)

Good fits:

• You just bought a precon and want to know what to swap before sleeving up.
• You want a budget-conscious upgrade path — $5/add is a sweet spot.
• You don't have time to read three EDHREC pages and a Reddit thread.

Not great fits:

• You want to convert a precon into a fundamentally different archetype. The AI is told to preserve the strategy; it won't turn your hugs deck into stax.
• You're optimizing for cEDH. The training data leans casual; competitive cuts often want to gut the precon entirely.
• You want the meme deck. Sometimes the fun is the bad cards.

For everything else — pick a precon, click upgrade, get a deck you'd actually shuffle up.

What's coming next

A couple of things on the roadmap:

• Diff view for the suggestions. Right now you accept cuts and adds individually. We're adding a "show me what the deck looks like with all suggestions applied" preview.
• Multi-pass refinement. Run the AI a second time with "I already accepted these cuts/adds, now what?" Useful when you've already done the obvious upgrades and want round 2.
• Buy list export. If 8 of the 10 suggested adds are cards you don't own, generate a TCGplayer mass-entry order for them in one click.

Try it now: Decks → Upgrade a Precon. Featured precons load instantly; full catalog (700+ decks) is searchable.