Skip to main content

How It Works

Describeit.Yourteammatesbuildit.
Youapproveit.

You talk to AI teammates in plain English. The Manager hands each part to the right one. Your data person reviews the work before it ships.

The Workflow

Four steps from source to dashboard.

Connect, describe, review, ship. Here is the flow, end to end.

01

Connect a source

Point OptimaFlo at a database, a cloud bucket or an API. The Ingestion Engineer reads the structure and gets it ready.

02

Describe what you need

Type it in plain English: "load my Stripe data and show revenue by region." No ticket queue.

03

Review the work

Your teammates write the pipeline and the SQL. You read it, change it, or approve it. Quality checks are attached to every step.

04

Ship it in your cloud

The approved work runs in your own AWS or GCP account and feeds a live dashboard. You end up with a raw copy of each source with full history, clean tables from SQL your team approved, and ready tables that power your dashboards, all stored as open Apache Iceberg tables you own.

Your AI Teammates

AI teammates do the legwork, you make the calls

Extra hands for the data person or team you already have. Describe what you need in plain English and a Manager delegates to specialists who handle ingestion, modeling, analysis, dashboards, and quality. Your people review and approve what ships. Big data platforms assume a platform team to run them. OptimaFlo handles the platform work, so your team can take on more.

1
Connect GCS bucket
→ Ingestion Engineer
2
Build end-to-end pipeline
→ Data Engineer
3
Generate dashboard
→ BI Developer
4
Attach quality rules
→ Quality Engineer

Manager

Describe what you need in plain English. The Manager breaks it into steps and delegates to the right teammate.

✓
Scanning bucket
gs://analytics-prod/
✓
Found 3 folders
events/ users/ transactions/
✓
Detected format
Parquet (snappy)
✓
Inferred schema
6 columns, 2.4M rows
Inferred Schema
user_idINT
emailVARCHAR
created_atTIMESTAMP
revenueDECIMAL

Ingestion Engineer

Connects to your data. Reads its structure. Recognizes file types on its own.

SRCConnect
RAWIngest
CLNClean
KPIModel

Data Engineer

Breaks your goal into pipeline steps. Generates SQL for each. Lays them out on the visual canvas.

clean_transform.sql
WITH cleaned AS (
SELECT
CAST(user_id AS INT),
TRIM(LOWER(email)),
SUM(revenue) AS total
FROM raw.events
GROUP BY 1, 2
)

Analytics Engineer

Writes and cleans the SQL behind every model. Checks for security issues, then re-runs transforms when your data changes.

MRR$42k+12%
DAU8.4k+5%

BI Developer

Builds interactive dashboards from your data, then edits them through conversation. Adjust filters, swap chart types, add widgets by asking.

Show me revenue by day
Mon
Tue
Wed
Thu
Fri
Sat
Sun

Analyst

Query your data in plain English. Generates SQL, runs it, returns charts and tables. No code required.

Completeness
95%
✓
Validity
90%
✓
Uniqueness
88%
✗
Consistency
94%
✓
Timeliness
97%
✓

Quality Engineer

Builds quality checks: completeness, consistency, uniqueness, accuracy. Attaches them to every step.

AI Instructions

Teach the AI agents your rules once.

Provide custom rules that the AI agents follow on every build, scoped to a workspace or a single pipeline. No repeating yourself.

Your rules, always followed

The AI agents apply them on every build, automatically. No reminders, no drift.

Workspace or pipeline scope

Set defaults for everything, or rules that apply to just one pipeline.

Reuse or override

Save instruction sets and reuse them across pipelines, or override for a one-off run.

AI Instructions
Workspace

Exclude test and internal accounts.

Revenue is in USD; convert other currencies first.

Match our naming: snake_case, gold_ prefix.

Applied to every build

Under the hood: proven Apache open standards

Apache Iceberg: Table format
Apache Airflow: Orchestration
Apache Polaris: Catalog

More data work than people? Add AI teammates.

From raw data to live dashboards in one conversation.

Now in early beta. One flat plan, no per-query tax. Runs in your cloud, and your tables stay in your account.