Kindgi
DocsSign In
GuidesPromptsIntegrationsVideos
All Guides

Optimizing Your Workflow

After a successful test run, Kindgi can optimize your workflow — locking in exactly how each step worked. Future runs are faster, more cost-efficient, and much more consistent. This guide walks you through everything: optimizing, managing runs, switching between optimizations, and disabling when needed.

Getting Started·5 min read
1Set Up for a Clean Run1/8
Phase 1Prepare for and execute a clean test run
01

Set Up for a Clean Run

Optimization requires a fully successful test run — all steps green. These tips will maximize your chances of getting one on the first try.

Use exact resource names
When referencing repos, channels, sheets, or boards — use the exact name as it appears on the connected account. E.g., "Brown-and-Beatty-Solutions/grievance-management" not "brown and beatty repos".
Verify connections first
Go to the Build tab and confirm all integrations show the correct connected account. A misconfigured connection is the most common cause of step failures.
Have real data available
Run with actual data flowing through — real emails in the inbox, real issues in the repo. Steps that process empty results can't be optimized effectively.
Keep prompts specific
Precise instructions produce better plans. "Fetch issues from the last 24 hours" is better than "check for new issues."
Phase 2Lock in your workflow from a successful run
02

Run, Review, and Optimize

Hit Test Run and watch the steps execute. If all steps complete successfully, the amber Optimize Workflow banner appears at the bottom of the results. Review each step's output, then click Optimize Workflow on the banner.

Before clicking Optimize, review the results. Open each step and check the output — make sure it's what you expect. Optimization locks in the behavior from this exact run, so the data and responses should look right.

Faster
Steps execute in under 2 seconds instead of 5-15 seconds — no AI reasoning overhead.
Cost-efficient
Optimized workflows typically use 3–6x fewer credits per run. Your subscription goes further — run more agents with the same plan.
Consistent
The same actions run the same way every time. Everything is locked in from your successful run.
All steps completed — the Optimize Workflow banner appears
Done — steps are now optimized for future runs
03

The Optimize Dropdown

Once you've optimized, a Zap button appears in the Build tab toolbar showing how many steps are optimized (e.g., "3 steps optimized"). Click it to open the dropdown — this is your control center for managing optimization.

The dropdown lists your most recent completed test runs. Each run shows:

Active badge
The run your workflow is currently optimized from. Highlighted in your brand color with a Zap icon.
Compiled badge
Runs that have been optimized before. You can switch to them instantly — no need to re-optimize.
Live label
Runs from the version currently in production. Helps you know which runs match what's actually running.
The optimize dropdown shows all your completed test runs
04

What Gets Optimized

Kindgi optimizes two types of steps from your workflow. Steps that require creative thinking continue to use AI.

Integration steps
Fetch emails, create drafts, send messages, create cards, update sheets — any step that talks to an external app through a connected account.
Data extraction steps
When your workflow processes items in a loop (e.g., extracting order details from each email), Kindgi learns the extraction pattern and applies it directly.

Steps that require genuine creative thinking — like writing a reply in your voice or deciding which emails are important — continue to use AI. Those are the steps where flexibility matters.

Downstream steps (send email, create card) optimize best because they receive all their data from previous steps. Source steps (fetch emails, list issues) also optimize well when run with real data.

Phase 3Switch, re-optimize, or disable
05

Switching and Re-optimizing

The optimize dropdown gives you full control over which run powers your workflow. Click any run in the list to take action.

Switch instantly
Runs with a **compiled** badge were optimized before. Click one to switch — it activates immediately with no recompilation.
Re-optimize
Click the **active** run to rebuild the optimization. Useful after changing step settings — same run, fresh optimization.
Optimize from a different run
Click any other run to optimize from its results. Your current optimization isn't lost — you can switch back anytime.

You can also trigger optimization from the Test tab — when viewing a completed run that isn't the current source, you'll see either "Optimize from this run instead?" or "This run has an existing optimization" with options to switch or rebuild.

Re-optimize after changing step settings — same run, fresh optimization
06

When a Step Fails

If a step fails during a test run, you don't need to start over.

Click Re-run on the failed step. The AI makes fresh decisions each time, and the next attempt often succeeds. You can also check the step's error details to understand what went wrong.

A failed step doesn't affect optimization — you only need all steps green for the initial optimization. After that, if an optimized step encounters an issue during a run, Kindgi automatically falls back to AI execution for that step.

07

Disabling Optimization

If an optimized workflow isn't producing the results you expect, you can disable optimization entirely.

Open the optimize dropdown in the Build tab and click Disable optimization at the bottom (shown in red). Your workflow will go back to using AI for all steps.

When to disable
If results seem wrong, outdated, or you've changed how a connected service is set up (e.g., different repos, different sheet structure).
What happens next
Steps run through AI again on the next execution. You can re-test and optimize again once you get a clean run you're happy with.

Disabling optimization doesn't delete anything else — your workflow, connections, and history are all preserved. You're just telling the system to figure things out fresh on the next run.

08

After Optimization

Optimized workflows run automatically — no action needed from you. The system handles edge cases gracefully.

Different data
Works fine. Optimized steps handle any data with the same structure — different emails, different orders, different recipients.
Automatic fallback
If an optimized step encounters an issue, Kindgi automatically falls back to AI execution for that step. Your workflow still completes.
Connection changes
If you switch to a different connected account, optimization is automatically cleared so the AI can figure out the new setup.
Quick checklist
Verify all connections on the Build tab
Use exact resource names in your prompt
Run with real data in your connected accounts
Review the results — make sure output looks correct
Click Optimize Workflow when the banner appears
Use the optimize dropdown to switch between runs or disable
After changes to connections or workflow, re-test and optimize again

You're all set

Your Optimizing Workflows guide is complete.