March 14, 2026

Optimize Your Agent Workflows

Kindgi lets you optimize your workflows after a successful test run. The system learns exactly how each step worked and locks it in — so future runs are faster, more cost-efficient, and much more consistent. No technical knowledge required.

Why optimize?

When your agent runs for the first time, each integration step uses AI to figure out what to do — which action to take, how to format the request, how to read the response. This works, but it takes time and costs credits. And because AI thinks through the problem fresh each time, results can vary slightly between runs.

Optimization solves all three problems at once. After a successful run, Kindgi captures exactly what worked and replays it directly on future runs. The AI thinking step is removed entirely for those steps.

What changes after optimization?

Three things improve immediately:

• Speed: Optimized steps finish in under 2 seconds instead of 5–15 seconds. The AI reasoning overhead is gone.
• Cost: Optimized workflows typically use 3–6x fewer credits per run, depending on how many steps can be optimized. That means your subscription goes further and you can run more agents with the same plan.
• Consistency: The same actions run the same way every time. No variation, no surprises.

How does it work?

The process is simple and takes just a few clicks:

1. Run a test. Go to the Test tab and run your agent. Make sure all steps complete successfully.
2. Review the results. Open each step and check that the output looks right. Optimization locks in the behavior from this exact run, so take a moment to verify.
3. Click Optimize. A banner appears after a successful run. Click it, and Kindgi learns from your run. This takes a few seconds.
4. Run again. Your next run uses the optimized path — faster, using significantly fewer credits for optimized steps.

What gets optimized — and what doesn't?

Kindgi optimizes two types of steps:

• Integration steps — the ones that talk to external apps like Gmail, Google Sheets, Trello, Slack, or GitHub. These follow a predictable pattern (fetch data, send a message, create a card), so locking them in makes sense.
• Data extraction steps — when your workflow processes items in a loop (e.g., extracting order details from each email), Kindgi learns the extraction pattern from your successful run and applies it directly on future runs.

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.

What if something changes?

Optimization is designed to be flexible. Here's what happens in common scenarios:

• Different data: Works fine. The optimized steps handle any data with the same structure — different emails, different orders, different recipients.
• Changed connections: If you switch to a different connected account, optimization is automatically cleared so the AI can figure out the new setup.
• Updated workflow: Changed your steps in the Build tab? Just run another test and optimize again.

Can I turn it off?

Yes. If an optimized workflow isn't producing the results you expect, you can clear optimization from the Build tab at any time. Your workflow goes back to using AI for those steps — no data is lost. You can re-test and optimize again whenever you're ready.

And if an optimized step ever encounters an issue during a run, Kindgi automatically falls back to AI execution for that step. Your workflow still completes — it just takes the longer path for that one step.

When should you optimize?

Optimize when you have a test run that produces exactly the results you want. Think of it as telling Kindgi: “This is how I want it done from now on.” The best time to optimize is after you've fine-tuned your agent and confirmed the output is right.

Frequently asked questions

What does optimizing a workflow do in Kindgi?

Optimizing locks in exactly how each step worked during a successful test run — both integration steps and data extraction steps. Instead of relying on AI to figure out what to do each time, the system replays the proven pattern directly — making future runs faster, more cost-efficient, and much more consistent.

How much faster are optimized workflows?

Optimized steps execute in under 2 seconds instead of 5–15 seconds. The AI thinking time is completely eliminated. For workflows with multiple integration steps, this can cut total run time by 50–80%.

Do optimized workflows cost less to run?

Yes. Optimized workflows typically use 3–6x fewer credits per run. The system replays the learned pattern directly instead of using AI for each step — so your subscription goes further.

Can I undo workflow optimization?

Yes. You can clear optimization at any time from the Build tab. Your workflow reverts to AI-powered execution with no data lost. You can re-test and optimize again whenever you get a clean run you're happy with.

What happens if an optimized step fails?

Kindgi automatically falls back to full AI execution if an optimized step encounters an issue. Your workflow still completes — it just uses the AI agent for that step instead of the optimized path. No manual intervention needed.