Platform
From story to deployment — and back.
01 / Build
What you had. What KnownForge generated.
A query inside a trigger loop runs once per record — past 100 records it hits the SOQL governor limit. Scrub to watch it bulkified: IDs into a Set, one query, a Map for lookup.
1trigger AccountTrigger on Account (after update) {2 for (Account a : Trigger.new) {3 List<Contact> cs = [SELECT Id FROM Contact✓ fixed4 WHERE AccountId = :a.Id];5 for (Contact c : cs) {6 c.MailingCity = a.BillingCity;7 }8 update cs;✓ fixed9 }10}
1+trigger AccountTrigger on Account (after update) {2+ Set<Id> accountIds = new Set<Id>();3+ for (Account a : Trigger.new) {4+ accountIds.add(a.Id);5+ }6+ Map<Id, Account> accounts = new Map<Id, Account>(Trigger.new);7+ List<Contact> toUpdate = new List<Contact>();8+ for (Contact c : [SELECT Id, AccountId FROM Contact9+ WHERE AccountId IN :accountIds]) {10+ c.MailingCity = accounts.get(c.AccountId).BillingCity;11+ toUpdate.add(c);12+ }13+ update toUpdate;14+}
- No SOQL inside loopspassed
- No DML inside loopspassed
- One query, one update — at any batch sizepassed
- Check-only validation passedpassed
02 / Stages
Five stages. One live picture underneath.
01 / 05 · Capture
The story arrives with context.
Work flows in from Jira, Linear, or GitHub Issues. Preflight analysis identifies every Salesforce component the story will CREATE or MODIFY — and asks only the questions that genuinely need a human, with answer options pulled from live org metadata (real layout names, record types, classes).
Intelligence tie-inReads the org's live metadata catalogue. Nothing is guessed.
preflight · KF-2481
MODIFY Flow Opportunity_Pricing_UpdateMODIFY ApexClass PricingServiceCREATE CustomField Opportunity.Discount_Tier__c? Which page layout should show Discount_Tier__c?○ Opportunity Layout○ Opportunity (Sales) Layout02 / 05 · Design
Reviewed before a line is written.
A technical design doc generated from actual org metadata and codebase patterns — streamed live — with preflight answers injected as hard constraints. You approve the plan before any code exists.
Intelligence tie-inComponent types come from the plan, not from parsing or guesswork.
TDD · draft
## 1. ScopeDiscount tier drives approval on Opportunity.## 2. ComponentsFlow · ApexClass · CustomField · Layout## 3. ConstraintsLayout: Opportunity (Sales) Layout ← preflight▍ streaming…03 / 05 · Build
Code that reads like your team wrote it.
Generation reads existing automations, follows conventions, and enforces curated Salesforce best practices (trigger architecture, SOQL, DML, governor limits, security) at prompt time. Packaging is automatic — correct package.xml, every time.
Intelligence tie-inBest-practice guardrails + org naming conventions injected into every generation.
package.xml
<types><members>PricingService</members><name>ApexClass</name></types><types><members>Opportunity.Discount_Tier__c</members><name>CustomField</name>04 / 05 · Validate
It doesn't ship until it's clean.
A two-step loop before anything deploys. Local check first — static analysis (zero tokens), then an Apex pre-review only if that's clean. Then a check-only deploy against your org via the Metadata API. Failures go to an AI fixer that repairs the files, rebuilds the package, and re-validates — up to three rounds, automatically.
Intelligence tie-inEvery validation error stored structurally, so the fixer works from evidence.
validation · round 2 / 3
local static analysis cleanlocal apex pre-review cleanorg check-only deploy 1 errorPricingService:42 Variable does not existfix files repaired · package rebuiltorg check-only deploy passed05 / 05 · Ship
Deployed, committed, and remembered.
The deploy runs; branch, commit, and pull request are created automatically. A post-deploy sync refreshes the intelligence layer — the next story starts from a current picture of the org, not last month's.
Intelligence tie-inThe layer auto-refreshes on every deploy.
ship · KF-2481
deploy staging donebranch feature/KF-2481commit a1c9e0f KF-2481 pricing tierspull request openedsync intelligence layer refreshed
01 / 05 · Capture
The story arrives with context.
Work flows in from Jira, Linear, or GitHub Issues. Preflight analysis identifies every Salesforce component the story will CREATE or MODIFY — and asks only the questions that genuinely need a human, with answer options pulled from live org metadata (real layout names, record types, classes).
Intelligence tie-inReads the org's live metadata catalogue. Nothing is guessed.
preflight · KF-2481
MODIFY Flow Opportunity_Pricing_UpdateMODIFY ApexClass PricingServiceCREATE CustomField Opportunity.Discount_Tier__c? Which page layout should show Discount_Tier__c?○ Opportunity Layout○ Opportunity (Sales) Layout02 / 05 · Design
Reviewed before a line is written.
A technical design doc generated from actual org metadata and codebase patterns — streamed live — with preflight answers injected as hard constraints. You approve the plan before any code exists.
Intelligence tie-inComponent types come from the plan, not from parsing or guesswork.
TDD · draft
## 1. ScopeDiscount tier drives approval on Opportunity.## 2. ComponentsFlow · ApexClass · CustomField · Layout## 3. ConstraintsLayout: Opportunity (Sales) Layout ← preflight▍ streaming…03 / 05 · Build
Code that reads like your team wrote it.
Generation reads existing automations, follows conventions, and enforces curated Salesforce best practices (trigger architecture, SOQL, DML, governor limits, security) at prompt time. Packaging is automatic — correct package.xml, every time.
Intelligence tie-inBest-practice guardrails + org naming conventions injected into every generation.
package.xml
<types><members>PricingService</members><name>ApexClass</name></types><types><members>Opportunity.Discount_Tier__c</members><name>CustomField</name>04 / 05 · Validate
It doesn't ship until it's clean.
A two-step loop before anything deploys. Local check first — static analysis (zero tokens), then an Apex pre-review only if that's clean. Then a check-only deploy against your org via the Metadata API. Failures go to an AI fixer that repairs the files, rebuilds the package, and re-validates — up to three rounds, automatically.
Intelligence tie-inEvery validation error stored structurally, so the fixer works from evidence.
validation · round 2 / 3
local static analysis cleanlocal apex pre-review cleanorg check-only deploy 1 errorPricingService:42 Variable does not existfix files repaired · package rebuiltorg check-only deploy passed05 / 05 · Ship
Deployed, committed, and remembered.
The deploy runs; branch, commit, and pull request are created automatically. A post-deploy sync refreshes the intelligence layer — the next story starts from a current picture of the org, not last month's.
Intelligence tie-inThe layer auto-refreshes on every deploy.
ship · KF-2481
deploy staging donebranch feature/KF-2481commit a1c9e0f KF-2481 pricing tierspull request openedsync intelligence layer refreshed
03 / Validation loop
Up to 3 rounds, zero human clicks.
01
Local check
Static analysis, then Apex pre-review
02
Org validation
Check-only deploy via the Metadata API
03
AI fix
Repairs files from structured errors, rebuilds the package
04
Re-validate
Back through the checks — up to 3 rounds
04 / Start
Start shipping Salesforce with confidence, today.
No credit card. No sales call.