From zero to incremental in a few lines.

Create a view, change the base table, refresh. Only the new rows are processed.

CREATE TABLE sales (region VARCHAR, product VARCHAR, amount INT);
INSERT INTO sales VALUES ('US', 'Widget', 100), ('EU', 'Gadget', 200);

CREATE MATERIALIZED VIEW regional_totals AS
    SELECT region, SUM(amount) AS total, COUNT(*) AS cnt
    FROM sales GROUP BY region;

-- Changes are captured as deltas automatically
INSERT INTO sales VALUES ('US', 'Bolt', 50), ('JP', 'Gear', 300);
DELETE FROM sales WHERE product = 'Gadget';

PRAGMA refresh('regional_totals');

SELECT * FROM regional_totals ORDER BY region;
-- JP  | 300 | 1
-- US  | 150 | 2        (EU is gone: its only row was deleted)

Views with REFRESH EVERY are maintained by a background daemon.

-- Refresh in the background every 5 minutes
CREATE MATERIALIZED VIEW regional_totals REFRESH EVERY '5 minutes' AS
    SELECT region, SUM(amount) AS total, COUNT(*) AS cnt
    FROM sales GROUP BY region;

PRAGMA refresh_status('regional_totals');   -- interval, last / next refresh

-- Let OpenIVM pick incremental vs full per refresh
SET openivm_refresh_mode = 'auto';
SET openivm_adaptive_refresh = true;          -- experimental learned cost model
PRAGMA refresh_cost('regional_totals');       -- incremental vs recompute estimate

Over DuckLake, deltas come from native snapshots, with no duplicated storage.

INSTALL ducklake;
LOAD ducklake;
ATTACH ':memory:' AS dl (TYPE ducklake);

CREATE TABLE dl.orders (id INT, product VARCHAR, region VARCHAR, amount INT);

-- Views stack into pipelines; DuckLake snapshots provide the deltas
CREATE MATERIALIZED VIEW dl.product_totals REFRESH EVERY '5 minutes' AS
    SELECT product, SUM(amount) AS total, COUNT(*) AS cnt
    FROM dl.orders GROUP BY product;

CREATE MATERIALIZED VIEW dl.top_products REFRESH EVERY '10 minutes' AS
    SELECT product, total FROM dl.product_totals WHERE total > 1000;

-- Refreshing product_totals also refreshes top_products
SET openivm_cascade_refresh = 'downstream';   -- or 'upstream', 'both', 'off'

OpenIVM is a compiler: you can read exactly what it runs.

-- Write every compiled statement to disk
SET openivm_files_path = '/tmp/openivm';

CREATE MATERIALIZED VIEW regional_totals AS
    SELECT region, SUM(amount) AS total FROM sales GROUP BY region;
PRAGMA refresh('regional_totals');

-- /tmp/openivm/openivm_compiled_queries_regional_totals.sql   (DDL at CREATE time)
-- /tmp/openivm/openivm_upsert_queries_regional_totals.sql     (the refresh SQL)

-- Peek at pending changes
SELECT * FROM openivm_delta_sales;

The Spark integration, with quickstarts and best practices for dbt workloads, is in development. Open an issue to get involved.