PokePilot Help
Build a team, compare damage, and refine your sets with analysis. Calculations and recommendations support your decisions; they do not guarantee a win.
PokePilot analysis examples show real Pokemon, team, open-slot recommendation and set analysis results for a Kabamanda team, plus a Luna low versus medium comparison. No sign-in or API key is required to read them.
Hippowdon: Defense or Special Defense? compares damage from Mega Salamence and Primarina to show when a different spread buys a turn for Yawn.
1. Getting started
Select an empty team slot, then search for a Pokemon in the selector. When usage data is available, the app fills in a sample build, including settings such as its item, nature, effort values, and moves. This is a starting point, not a mandatory set. Edit each field to suit your plan. Without usage data, review the defaults and fill in the build yourself.
Choose Singles or Doubles in the header before comparing sets. The format can affect calculation conditions and relevant usage statistics. Use the team builder/calculator toggle to switch workspaces.
First visits default to Singles and the system color theme. Without a saved language preference, browser language determines whether Korean or English is selected. This does not identify your nationality or location. Change language and theme in the header menus.
2. Teams and builds
Switch Pokemon tabs to choose the member you are editing. Select an item, ability, nature, or move to change it. Items and moves can also be cleared. For Pokemon with supported forms, check the available form and its associated abilities, stats, and moves.
Effort values and nature
Adjust effort values with sliders or numeric inputs. The app uses Champions values: up to 32 in one stat and 66 in total. Do not compare these numbers directly with the older 252/510 EV notation. Read the nature's increase and decrease indicators alongside the resulting stats.
More Attack does not always change the number of hits needed to KO an opponent. However, the same hit-count range can still have a different KO probability. Speed investment should reflect the Pokemon's team role, not just the opponent currently selected in the calculator.
Bench and ordering
The active team holds up to six Pokemon, and the bench holds up to six alternatives separately. Save a recommendation to the bench when you want to compare it before replacing a team member. Drag Pokemon or moves to reorder them.
3. Defenses and coverage
The defensive type chart is a starting point for identifying shared weaknesses and useful resistances. Its resistance count includes immunities. If several members share a weakness, consider which teammate can realistically take that attack.
Offensive coverage describes which types your selected attacks can hit super effectively. Covering a type does not mean you beat every Pokemon of that type. Dual typing, abilities, items, actual damage, and turn order still matter.
Example: does a shared weakness require a replacement?
A team with multiple Ground weaknesses might still have a reliable switch-in or a way to defeat a threat before it attacks. Instead of only reducing a number in the chart, calculate who takes the hit. A resistant teammate with low HP or a weakness to the opponent's other move may not be a reliable answer.
4. Damage calculator
Select your Pokemon and an opponent, then match both builds to the situation you want to test. Use the attack/defense toggle to change whose moves are calculated. Check current HP, status, stat stages, weather, terrain, and the additional conditions available in the interface.
Weather- and terrain-setting abilities can update the environment automatically. Recheck the displayed environment after changing a Pokemon or ability. Spread attacks and support effects can affect Doubles calculations.
Reading percentages and KO labels
A damage range shows possible damage under the selected conditions. A guaranteed OHKO does not account for every way an attack can fail, such as missing, Protect, or a switch. A probabilistic OHKO means only some damage rolls reach the KO threshold.
Example: 90-106% damage and a guaranteed 2HKO
A hypothetical 90-106% range can KO a full-health target, but does not guarantee it. The exact chance depends on the actual damage rolls, not the length of the percentage interval. A guaranteed 2HKO also does not promise victory within two turns: healing, Defense increases, and turn order can change the outcome.
For example, Stamina can raise Defense after the first hit, so repeating first-hit damage may be misleading. A move that ignores Defense stages can behave differently. Last Respects also depends on battle history. Do not assume the calculator simulates conditions that are neither displayed nor configurable.
5. PokePilot analysis
AI analysis requires Google sign-in and a personal OpenAI API key. Register it in the top-right profile menu; follow the API key setup guide for instructions. Team building and damage calculation work without a key.
Select Pokemon, Team, Find, or Sample. Pokemon analysis requires a Pokemon in the selected slot; team analysis and Pokemon recommendations require at least three active members in Singles or four in Doubles. Choose the default Luna low or Luna medium for deeper reasoning at increased time and cost at the bottom-right of the panel. Press the analysis button to prepare the input and review approximate time, total-token and cost ranges. Opening this confirmation does not call OpenAI or save your team. After you confirm, the team is saved before the paid analysis starts. Estimates do not guarantee actual charges or completion time, and higher reasoning effort does not guarantee accuracy.
Confirming analysis automatically saves the current team before the AI request. If saving fails or a sync update needs your attention, analysis does not start and no API cost is incurred. An unchanged saved team is checked without saving another copy. The team remains saved even if the AI request fails.
During analysis, the button shows elapsed time and the result area shows team saving, input preparation, and AI response waiting status.
After editing a team or build, refresh the analysis: an existing result may describe the old conditions. Use history to revisit previous results. Usage-based results retain their original statistics season and date; old records without this information are marked as not recorded. A statistics update does not recalculate earlier analyses.
Team and Pokemon analysis
Team analysis discusses interactions, a game plan, strengths, and weaknesses. Pokemon analysis focuses on the selected member's role and how its current build works within the team.
Pokemon recommendations
With open slots, recommendations consider additions. With a full team, they consider replacement candidates. Check whether the proposed replacement would remove an essential role or combination, not just whether the newcomer is individually strong. Apply a candidate or save it to the bench for comparison.
Build recommendations
Build recommendations examine the selected Pokemon's item, nature, effort values, and moves. Keeping the current build can be a valid recommendation. Expand calculation evidence when available and compare the before-and-after conditions. A useful candidate is not necessarily superior against every opponent.
Evaluate moves and items together. For example, using a support move while retaining a Choice item creates a move-lock constraint. Replacing a move used in a team combination can also cost more than it gains. Review those tradeoffs before applying the build.
Warnings, limits, and privacy
OpenAI charges your account. PokePilot has no daily/monthly analysis allowance or cooldown, but OpenAI balance, model access, and rate limits still apply. Failed requests show an explanation instead of a rules-based replacement. Only failures confirmed to occur before the provider call receive a no-analysis-cost notice. AI can misinterpret move interactions or battle plans. Requested analysis data is sent to OpenAI. See the privacy notice for details.
AI response waiting is limited to 2 minutes 30 seconds for Luna low and 6 minutes 30 seconds for medium. The browser does not automatically resubmit after a timeout, but processing may continue on the server or OpenAI and charges may occur. Check your analysis history and OpenAI usage before retrying.
6. Saving and sharing
Name your team and use Save to keep it. Open the team menu to load and manage saved teams. Save edits before switching teams if you want to preserve them.
Signing in synchronizes saved teams and bench sets, analysis history, language, theme, default battle format, tutorial completion, and the last selected analysis tab/model settings across signed-in devices. Unsaved edits and current screen state are not synchronized. Signing out clears this browser's team and history copies but retains the account copies. Guest teams stored only in the browser can be lost when site data is cleared. Keep important teams separately as Showdown text.
Import Showdown text through the new-team menu and use the Showdown text control to export a build or team. After importing, verify forms, items, moves, effort values, and legality indicators. Image export is for visually sharing builds and teams; an image is not a substitute for an importable text backup.
7. Data and regulations
Eligibility data and usage statistics are different datasets. Supporting a new regulation does not mean usage statistics for that environment are available immediately.
Battle rules use Regulation M-C. Rankings and automatic samples use daily Singles or Doubles statistics from Pokémon Champions Battle Data. Read the source season, date, and warnings in the analysis panel. During outages, recent cached statistics may remain available; missing statistics do not mean a Pokemon is illegal.
Automatic samples combine popular individual choices. Nature, Stat Points, items, and moves are measured separately; their combination is not necessarily a commonly played complete set. Review and adjust the sample for your team.
Game data and calculations use Pokemon Showdown-related resources. Korean data and images also use sources including PokeAPI. Missing translations or form images may fall back to English or an alternative image. App legality indicators do not replace the latest official tournament rules.
8. Common questions
Why was no sample loaded?
Usage data may be unavailable for the Pokemon, form, or battle format, or a data request may have failed. Check loading and error notices. If statistics are missing, set up the build manually.
Why does another calculator show different damage?
First match the form, item, ability, nature, effort values, and resulting stats. Then compare HP, stat stages, weather, terrain, Singles/Doubles, and support effects. One item, such as Life Orb, can make a substantial difference. Report the full reproduction conditions if the results still disagree.
Why does a recommendation keep my current build?
A change may offer little benefit or compromise the member's role. Review the explanation and evidence in the context of your team rather than judging the result only by how much it changes.
Where can I report an issue?
Use GitHub feedback and include reproduction steps, the battle format, builds, and expected result. Reports are public: do not include API keys or personal information.