Dataset Optimization: Improve Prompts in Your Dataset

If you already have a dataset with a Run Prompt column, skip to Step 2.

Go to app.futureagi.com → Dataset (left sidebar) → Add Dataset → create a dataset with input columns (for example question, context).

Add a Run Prompt dynamic column:

1. Click Add Column → select Run Prompt
2. Write a prompt template referencing your input columns, for example: Answer this question using the context: {{question}} Context: {{context}}
3. Select a model (for example gpt-4o-mini)
4. Run the prompt to generate outputs for all rows

You should see every row filled in with a generated answer under the Run Prompt column. The column stores both the prompt template and the outputs, which is what the optimizer improves.

Navigate to your dataset → click the Optimization tab (third tab, after Data and Experiments, before Summary).

You should see the run list for this dataset. If no runs exist yet, an empty state shows a Run Optimization button. Once runs exist, the list view shows an Optimize Prompts button in the header instead.

Click Run Optimization (empty state) or Optimize Prompts (list view header) to open the configuration drawer.

Field	Value
Name	Built from the column name, the optimizer, and a timestamp (for example answer-GEPA-Mar4-1430), edit if needed
Choose Column	Select a Run Prompt column from the dropdown
Choose Optimizer	Select an optimization algorithm (see table below)
Language Model	The LLM used during optimization (for example gpt-4o)
Optimizer Config	Parameters specific to the selected optimizer, auto-populated with defaults
Evaluations	Select one or more evaluation templates to score candidates

Available optimizers

Optimizer	Config parameters	Best for
Random Search	num_variations	Quick baseline, generates random prompt variants
Bayesian Search	min_examples, max_examples, n_trials	Few-shot example selection and ordering
ProTeGi	beam_size, num_gradients, errors_per_gradient, prompts_per_gradient, num_rounds	Targeted prompt edits based on error analysis
Meta-Prompt	num_rounds	General-purpose prompt rewriting
PromptWizard	mutate_rounds, refine_iterations, beam_size	Multi-stage mutation, scoring, and critique-refinement
GEPA	max_metric_calls	Evolutionary exploration of diverse prompt styles

Every optimizer also takes a task_description.

Click Start Optimization to launch the run. You should see the drawer close and a new row appear at the top of the run list with status Pending.

After launching, the Optimization tab shows the run with its current status.

The run moves Pending, Running, Completed; while it’s Running the tab auto-refreshes every 5 seconds. Failed and Cancelled stop it early.

Click the run to see the detail view with a Steps panel showing progress through the optimization stages, a results graph showing score progression across trials, and a trials grid listing each trial’s score and prompt variant.

You should see the status move from Pending to Running to Completed, and the results graph fill in as each trial finishes.

Click any trial in the grid to open the trial detail view. The detail view has two tabs.

The Prompt tab shows a side-by-side comparison: AGENT PROMPT is the baseline prompt from your Run Prompt column, OPTIMIZED AGENT PROMPT is the variant the optimizer generated for this trial. Toggle Show Diff to highlight the changes between the two.

The Trial Items tab shows the individual iterations the optimizer ran to produce this trial’s prompt, with input, output, and evaluation score per row.

Review a few trials to see how different optimization paths produced different prompt structures.

You should see a clear score gap between the best and worst trials, with the diff view showing exactly what changed. For example, the baseline scored 0.61 and the best trial 0.79 (illustrative, from an example run). The best-scoring trial’s prompt is the one you carry back to your column.

Once you’ve identified the best trial:

1. Copy the optimized prompt from the trial detail view
2. Paste it over your Run Prompt column’s template, or into a prompt template in Prompt Workbench for version control and production serving

To re-run optimization with different settings, for example a different optimizer or metric, click Optimize Prompts again from the Optimization tab.

You should see the previous run stay in the list, so you can compare its trials against the new run.