This T1A Alchemist Quickstart Guide will demonstrate how to leverage Alchemist to accelerate the migration of your legacy SAS code to Snowflake. We'll be covering the 3 most frequently occurring types of SAS assets:

By the end of this guide, you will know how to:

SAS has been at the core of enterprise data environments for decades. While it started as a statistical analysis platform, many organizations have come to rely on it far beyond analytics. It's common to find SAS powering full-scale data warehouses, complex ETL pipelines, and regulatory reporting workflows — often through tools like SAS Data Integration Studio (SAS DI) and SAS Enterprise Guide (SAS EG). Over the years these environments have grown into mission-critical systems with hundreds or even thousands of jobs running daily.

However, maintaining a legacy SAS estate comes at a growing cost. Licensing is expensive, the talent pool is shrinking, and the platform was not designed for the modern cloud-native world. Snowflake offers a compelling alternative: near-unlimited scalability, separation of storage and compute, and all of that at pay-per-use pricing. Migrating to Snowflake lets your team consolidate data engineering, analytics, and data science workloads onto a single modern platform, while significantly reducing infrastructure overhead and operational complexity.

Migrating from SAS to Snowflake might sound straightforward on the surface, given that SAS is just code and data flows. In reality, most organizations significantly underestimate the scope. Years of organic growth leave behind a tangled mess where the actual code is just the tip of the iceberg.

Beneath the surface lie hidden dependencies between jobs, undocumented macros, duplicate assets scattered across teams, and compliance constraints. Without full visibility into these layers, migration projects quickly spiral into missed deadlines and budget overruns. The bulk of the complexity of a migration project comes from uncovering the right scope for migration and coming up with a robust migration plan that takes all the different aspects into account.

Most of this hidden complexity can be uncovered and reduced with the right tooling. Rather than attempting to migrate everything at once, a structured approach lets you filter out what's no longer needed, understand what remains, and convert only what matters.

This is exactly the workflow that our toolkit is built around. SAS Logs Analyzer and Metascope help you see the full picture. Alchemist Code Analyzer tells you what you're actually dealing with. And Alchemist Converter turns it into production-ready Snowflake code.

With the rise of AI, it may be tempting to paste your SAS code into an LLM and ask for a translation. While LLMs are great for explaining code or quick prototyping, they often introduce subtle logic differences — a changed join order, unexpected null handling, slightly different rounding. In regulated industries, even minor deviations can mean a compliance violation. Additionally, with not a lot of SAS code available publicly, LLMs were not trained to understand it on a deep enough level - which leads to conversion mistakes.

This is why Alchemist uses a combinaion of deterministic conversion logic and AI in a broader, structured conversion pipeline — not AI as the sole engine.

If you haven't already, register for a Snowflake free 30-day trial. The rest of the sections in this lab assume you are using a new Snowflake account created by registering for a trial.

After your first login, go to Projects, click + Add new, and choose SQL file.

Paste and run the following to create the database all converted notebooks will use:

All three SAS libraries referenced in this lab (LDEMO, LDEMODI, LMART) map to LDEMO.PUBLIC — a single Snowflake database and schema is all you need.

Register at app.getalchemist.io to get access to the portal. The trial signup is free and gives you the credentials needed to download the installer.

This guide requires Alchemist version 2025.2 or later. Sign in to the portal, open the Downloads page, and pick the installer for your operating system (macOS Apple Silicon or Windows).

Run the installer.

On macOS, add Alchemist to your PATH after installing:

To make this permanent, add the line above to your ~/.zshrc or ~/.bash_profile.

On Windows, the installer should register the Alchemist install directory on your Path automatically — open a fresh Command Prompt or PowerShell window after installation to pick up the change.

Verify the installation:

If alchemist --version returns "command not found" in a fresh shell, add the install directory to your user Path manually:

The line above is for the current session only. To make it permanent, add C:\Program Files (x86)\Alchemist\Alchemist to your user Path via System Properties → Environment Variables, then restart the shell.

The license file you'll need is bundled with the lab assets you'll clone in the next sections — license activation is covered at the end of step 5.

This tutorial uses three sample SAS artifacts that represent a typical insurance analytics estate.

Pull just this guide's lab folder using a shallow, sparse clone. The same commands work on macOS, Linux, Git Bash on Windows, and Windows PowerShell — git is identical across platforms.

You are now in the lab directory. It contains everything you need:

A few pieces of scaffolding worth understanding:

The lab uses a source-locked demo license: the source_lock.json file in the lab folder is paired with a license key listed on the Alchemist Portal Get Started page. Scroll to the bottom of that page, find the SAS to Snowflake Migration — Self-Service Quickstart Demo section, and copy the Demo-locked license required key.

From the lab directory you cd'd into above, run:

When prompted, paste the license key you copied from the portal.

Before converting, run Alchemist Analyzer to get a breakdown of your SAS estate: file count, token volume, workload type mix, and the constructs that will need attention during conversion.

In this video, you can see the capabilities of the Analyzer

A single analyzer run covers all three sample asset types at once:

The dashboard opens automatically at http://127.0.0.1:10888. The Main tab shows a project overview across every source file — token count, executable-type breakdown (SAS Programs, EG Flows, DI Jobs), and workload-type mix:

The EG, DI, and Code tabs alongside Main provide more details about each specific asset type.

The converted code reads from three tables — INSURANCE_CUSTOMERS, INSURANCE_CLAIMS, and INSURANCE_ADJUSTERS. Load them into Snowflake using the Add Data UI before running the converted code.

The CSV files are in the data/ folder of the cloned repository.

Repeat the following steps for each of the three CSV files:

1. From your lab directory, run:

2. Open the converted notebook. Alchemist preserves the source-tree shape under your output directory, so the full path is:

   output/base_code/insurance_review_converted_files/Programs/sas_artifacts/base_code/quarterly_portfolio_review.ipynb
   

   Subsequent steps abbreviate this as output/base_code/.../quarterly_portfolio_review.ipynb.

3. Any construct Alchemist couldn't convert automatically is marked # TODO:. Reporting-only statements (ODS, TITLE, FOOTNOTE) can be deleted — they have no notebook equivalent.

1. In Snowsight go to Projects → Workspaces.

2. Click + Add new → Notebook → Upload .ipynb file. Select output/base_code/.../quarterly_portfolio_review.ipynb, name it, and click Create.

3. Click Connect → + Create new service, set Idle timeout to 30 minutes, and click Create and connect.

4. Add a SQL cell at the very top of the notebook:

5. In cell "1. Imports", swap SparkSession for snowpark_connect:

6. In cell "2. Setup Spark session", replace the builder:

7. Click Run all.

The sample project contains 15 task nodes across two parallel branches — a customer risk path and a claims enrichment path — that merge into a final reporting stage:

1. Run the converter:

2. Open output/enterprise_guide/.../insurance_Process_Flow.ipynb. Delete the four cosmetic TODOs:

3. The Adjuster report task ends with # TODO: Unsupported procedure PRINT. Replace it with:

The three source tables are already loaded in LDEMO.PUBLIC from step 7.1 — no additional data loading needed.

1. In Snowsight go to Projects → Workspaces → Add new → Notebook → Upload .ipynb file. Select output/enterprise_guide/.../insurance_Process_Flow.ipynb, name it insurance_eg_flow, and click Create.

2. Click Connect and select the compute service you already created.

3. Add a SQL cell at the very top:

4. In cell "1. Imports", swap SparkSession for snowpark_connect:

5. In cell "2. Setup Spark session", replace the builder:

6. Click Run all.

The sample DI job (LOAD_CUSTOMER_RISK_MART) reads three operational tables and produces the CLAIMS_ENRICHED mart through six numbered transforms — Extract, Data Validation, two Joins, Sort, and a Table Loader:

1. Run the converter:

2. Open output/data_integration/.../LOAD_CUSTOMER_RISK_MART.ipynb. Find the step2_Data_Validation stub and replace it — the notebook will fail without this:

1. In Snowsight go to Projects → Workspaces → + Add new → Notebook → Upload .ipynb file. Select output/data_integration/.../LOAD_CUSTOMER_RISK_MART.ipynb, name it load_customer_risk_mart, and click Create.

2. Click Connect and select the compute service you already created.

3. Add a SQL cell at the very top:

4. In cell "1. Imports", swap SparkSession for snowpark_connect:

5. In cell "2. Setup Spark session", replace the builder:

6. Click Run all. Once complete, open Database Explorer → LDEMO → PUBLIC → CLAIMS_ENRICHED to verify the output.

Congratulations — you have completed the full SAS-to-Snowflake migration workflow using T1A Alchemist.

Validation of the converted code outside the scope of this guide, but T1A provides a dedicated validation framework that automatically compares the output of original SAS code against the converted Snowflake code — row by row, column by column — to confirm logical equivalence. This is especially important for regulated workloads where manual spot-checking is not sufficient.

Manual SAS migration is slow, error-prone, and expensive. Alchemist shortens the cycle significantly by automating the parts that can be automated — syntax translation, dependency mapping, and notebook scaffolding — while keeping a human in the loop for the logic that requires review. Key advantages: