# Automated SAS to PySpark migration

> Alchemist analyzes your SAS estate and converts SAS code to PySpark deterministically for Databricks, Snowflake and Microsoft Fabric. Free trial.

- Canonical page: https://getalchemist.io/
- Updated: 2026-11-05
- Site index: https://getalchemist.io/llms.txt

A SAS-to-cloud migration acceleration toolkit: analysis, deterministic conversion to PySpark, and automation, built for PySpark-compatible platforms: Databricks, Snowflake and Microsoft Fabric.

Free trial account. Code Analyzer on any number of SAS assets, no time limit, no usage limit.

- Get Started: https://app.getalchemist.io/
- Talk to us: https://getalchemist.io/#contact

Listed on: [Microsoft Marketplace](https://marketplace.microsoft.com/en-us/product/tieroneanalyticsinc1760439825228.alchemist), [Databricks Marketplace](https://marketplace.databricks.com/details/f032172b-2e90-4972-953c-2f48f32d624d/_T1A-Alchemist), [Databricks Partner Solutions](https://www.databricks.com/company/partners/consulting-and-si/partner-solutions/t1a-alchemist)

**Conversion.** A macro-driven DATA step becomes a Python function: a DataFrame chain and a temp view.

`key_orders.sas` (SAS):

```sas
%macro key_orders(min);
  data work.key_orders;
    set sales.orders;
    where amt > &min;
    keep id region amt;
  run;
%mend;
%key_orders(1000);
```

`key_orders.py` (PySpark):

```python
# Converted by Alchemist
def key_orders(min):
    df_key_orders = (
        spark.table("SALES.ORDERS")
        .filter(f"AMT > {min}")
        .select(
            "ID",
            "REGION",
            "AMT",
        )
    )
    df_key_orders.createOrReplaceTempView(
        "KEY_ORDERS"
    )

key_orders("1000")
```

**Analysis.** Code Analyzer dashboard of a SAS estate: tokens, programs, macros, workload types and complexity.

**Automation.** E2E Migration Framework dashboard: each conversion runs analyze, samples, convert, upload, tests and publish.

## Why do teams migrate with Alchemist?

Because the output, the kits and the cost are settled before the project starts.

### Deterministic conversion, not guesswork

The Converter parses SAS into a syntax tree (a structured map of the code) and rewrites it with pattern matching and templates. Same input, same output, every run. An LLM pass is optional, never the engine.

### Toolkit, not a monolith

Analyzers, the Converter and the migration frameworks are separate tools. Run only the ones your migration needs, in your own infrastructure and your own process, and add the next one when the project calls for it.

### Self-serve, with the cost known first

The Assessment and Conversion Kits need no services: sign up, download, run your first analysis in minutes. The free Code Analyzer counts the tokens in your estate (Alchemist's own code metric, not an LLM token), and the calculator in your account turns that count into a figure before you talk to anyone.

## End-to-end migration with the Alchemist toolkit

Analyze the estate, convert data pipelines and macro-driven workflows to PySpark, automate the rest.

### Analysis

Four analyzers read the code, the logs and the metadata server, so the scope is measured before anything is rewritten.

[How a SAS code assessment works](https://getalchemist.io/sas-assessment/)

#### Free Dashboard

Included in: Trial

Runs on the trial license against any number of SAS assets and reports the token count, executable types and workload mix of the whole scope.

![Alchemist trial dashboard for a sample SAS estate: token count, source files, executable types (SAS programs, EG flows, DI jobs) and workload types](https://getalchemist.io/images/analyzer-dashboard-2240.webp)

#### Code Analyzer

Included in: Assessment · Conversion · Full

Parses SAS programs, EG projects and DI jobs into a syntax tree and returns statement-level statistics, complexity scores, and data and macro lineage as a dependency graph.

![Alchemist Code Analyzer network graph of a SAS estate: clusters of jobs and datasets linked by lineage, with the filter panel open and a Graph view / Lineage switch](https://getalchemist.io/images/analyzer-network-graph-2240.webp)

#### Log Analyzer

Included in: Assessment · Full

Reads SAS server and audit logs and lists who ran what: every dataset opened, read, written or updated, by user, library and engine, with runtimes and failures.

![Alchemist Log Analyzer, Data Access view: counts of open operations by mode, status and engine, and a table of dataset accesses with library, user and source log file](https://getalchemist.io/images/log-analyzer-data-access-2240.webp)

#### Metascope

Included in: Assessment · Full

Extracts the SAS metadata server data: users, groups, libraries, tables, jobs, stored processes, EG projects, deployed flows and servers, exportable as CSV.

![Metascope by T1A: inventory of stored processes from the SAS metadata server with folder, usage version, description and ID; the sidebar lists users, groups, libraries, jobs, tables, servers and EG projects](https://getalchemist.io/images/metascope-stored-processes-2240.webp)

### Conversion

Deterministic SAS-to-PySpark conversion you configure.

[How SAS code maps to PySpark](https://getalchemist.io/sas-to-pyspark/)

#### Code Converter

Included in: Conversion · Full

Deterministic SAS-to-PySpark conversion: pattern matching on the syntax tree, templates and data-object remapping, tuned for readable output.

- EG projects and DI jobs converted from their metadata, not only their code
- Configurable output format, code structure and formatting
- Templates adapt the conversion to your architecture decisions
- An optional LLM fallback for unsupported constructs, off by default
- No standalone runtime framework required

1. **Source:** SAS code and metadata (.sas · .egp · .spk)
2. **Parse:** Code to syntax tree (a full, detailed representation of the code)
3. **Generate:** Templates and configuration (table mapping, readability, optimisation)
4. **Optional:** LLM fallback (for unsupported constructs, off by default)
5. **Output:** PySpark (Databricks native · Snowflake and Fabric compatible)

Same input, same output: the tree decides, not a model.

Inputs: SAS Base (`.sas`), Enterprise Guide (`.egp`), DI jobs (`.spk`)

Output: PySpark, for three platforms:

- Databricks: Native. Native output, including Databricks notebooks.
- Snowflake: Compatible. Fully compatible. [Snowflake guide](https://www.snowflake.com/en/developers/guides/sas-to-snowflake-migration/)
- Microsoft Fabric: Compatible. Fully compatible.

Prophecy is also a target, for SAS DI sources.

### Automation

Frameworks and tools that turn conversion into migration.

Included in: Full Kit

- **End-to-end:** Agentic migration workflow (analyze → convert → validate → deploy, at scale)
- **Self-service:** Your teams convert on demand (in the Alchemist UI or via API)
- **In your platform:** Your cloud platform's AI assistant (Skills and the API: Databricks Genie, Snowflake Cortex, Microsoft Copilot, or a coding agent; MCP in development)

**One core:** Alchemist Analyzer + Converter (deterministic, the same in every route)

Three routes in, one deterministic conversion out.

#### E2E Migration Framework

Analysis, conversion, validation and deployment as one repeatable, agent-orchestrated workflow.

- One-click conversion at scale (100+ assets)
- Validation and troubleshooting inside the workflow
- Configured for your environment with T1A.

#### Change Management Platform

Tracks and governs migration changes across teams and phases: versioning, approvals, rollout control.

- Work items, UAT records, progress dashboards
- Gantt timelines and incident management in one place

#### Self-Service Conversion

Teams convert their own scope on demand: in the Alchemist UI, via the API, or from a coding agent.

- Skills and the API today in your local agents or your cloud platform's agent (Genie, Cortex, Copilot); an MCP server is in development
- Deployed locally or in-platform, code never leaves your workspace

[How coding agents drive Alchemist](https://getalchemist.io/agents/)

#### SAS–Cloud Coexistence Framework

SAS and Databricks run side by side during the migration, so each team cuts over at its own pace.

- Shared data access between SAS and Databricks
- Phased cutover, no big-bang switch

## How is Alchemist priced?

One-time license per kit, plus pay-as-you-go on the tokens you convert. No subscription, no renewal. Only the Converter is metered; the analyzers run without usage limits. Licenses start at $10K.

- [Calculate your price](https://app.getalchemist.io/)
- [How pricing works](https://getalchemist.io/pricing/)

The calculator is in your Alchemist account: run the free Code Analyzer on your own SAS assets, and it turns the token count into a figure.

| Kit | Price | Tools | Metering | T1A services | Best for |
| --- | --- | --- | --- | --- | --- |
| Trial | Free | Code Analyzer dashboard | Any number of assets, no time limit, no usage limit | Not required | Measure your token count before you buy anything |
| Assessment Kit | One-time license | Code Analyzer; Log Analyzer; Metascope | Any number of assets, no time limit, no usage limit | Not required | Full discovery of a SAS environment before a large migration |
| Conversion Kit | One-time license + pay-as-you-go | Code Converter; Code Analyzer | Token allowance included, then pay-as-you-go | Not required | Known scope, under about 100 assets |
| Full Kit | One-time license + pay-as-you-go, with T1A enablement | Assessment and Conversion Kits; E2E Migration Framework; Other tools on request | Token allowance included, then pay-as-you-go at a discount | T1A migration enablement, scoped separately | Hundreds of assets; comes with T1A services |

T1A services: migration enablement delivered by T1A, the company behind Alchemist, as your migration partner. Other migration partners are listed on the [partners page](https://getalchemist.io/partners/).

### License cost references

Four real cases, license cost only. Full Kit engagements add T1A services, scoped separately.

| Case | SAS scope | Tokens | Kit | License cost | Use |
| --- | --- | --- | --- | --- | --- |
| Healthcare | 150 SAS EGP | 200K tokens | Full Kit | $37,500 | BU self-service analytics |
| Health insurance (reporting) | 3 SAS EGP | 67K tokens | Conversion Kit | $13,600 | Analytical reporting |
| Health insurance (DWH) | 1,600 SAS DI | 997K tokens | Full Kit | $89,000 | EDWH: ODS + DM |
| Health insurance (data mart) | 75 SAS scripts | 101K tokens | Full Kit | $35,000 | Data mart |

License cost only, as of 2026. Full Kit engagements add T1A services, scoped separately.

## What has Alchemist already migrated?

Production SAS migrations since 2022, delivered on Databricks: a company-wide data warehouse, reporting estates, predictive models.

### Public references

- [Migrating legacy SAS code to the Databricks Lakehouse: what we learned along the way](https://www.youtube.com/watch?v=jcBja4PPLQc): Matt Adams, PacificSource, and Dmitriy Alergant, T1A. Data + AI Summit, June 2025. 54-step monthly workflow: about four days in SAS, under three hours on Databricks in final testing.
  PacificSource moved a monthly SAS data-mart process to Databricks with T1A. Alchemist converted most of the code automatically, before it had an LLM step, and engineers finished the few unsupported steps. In final testing at the time of the talk, the 54-step workflow ran in under three hours instead of about four days.
- [Building complex workflows with Databricks and SAS at Raiffeisen Bank International](https://www.youtube.com/watch?v=q4PFWE9mdT0&t=2575s): Stefan Baumann, Raiffeisen Bank International, with Databricks and T1A. Webinar, June 2022. SAS and Databricks side by side: shared data, one workflow, SAS models in Databricks MLOps.
  Raiffeisen Bank International moved SAS workloads to Databricks with T1A and Alchemist. In this webinar its retail-risk team runs SAS, Python and R together in Databricks notebooks, and T1A shows how SAS and Databricks share data and run in one workflow while the migration is under way.
- [CZ on the completed SAS migration, crediting T1A as the delivery partner](https://www.linkedin.com/posts/eelkozegers_touchdown-afgelopen-maand-hebben-we-een-share-7472906525527818240-fqoC/): Project lead at CZ, Dutch health insurer. SAS to Databricks migration with T1A and Alchemist, completed and live.
  CZ, a Dutch health insurer, completed its SAS to Databricks migration with T1A and Alchemist and has it live. The team kept delivering on SAS throughout, while building the new platform alongside it.
- [Alchemist: from Brickbuilder to a Databricks Marketplace App](https://www.databricks.com/blog/alchemist-brickbuilder-databricks-marketplace-app): Dmitriy Alkhimov, Aaron Zavora and Mark Lee. Databricks blog, January 2026. Analyzer, converter, validation and four customer projects; benchmark from release 2024.2.
  Co-authored by T1A, this Databricks blog post covers the Analyzer, the syntax-tree converter, how converted code is validated, and four customer projects. Its 30-flow benchmark was measured on release 2024.2 with templates, configuration and the LLM step all turned off.
- [SAS-to-Snowflake migration guide](https://www.snowflake.com/en/developers/guides/sas-to-snowflake-migration/): Ivan Baranov, T1A. Snowflake developer guides. Hands-on lab: Alchemist's PySpark output running on Snowflake through Snowpark Connect.
  A hands-on lab that runs the Analyzer and Converter on three sample SAS assets (Base, EG and DI), then runs the PySpark output on Snowflake through Snowpark Connect.
- [Alchemist demos and walkthroughs](https://www.youtube.com/@Alchemist_Migration): The Alchemist channel. Analyzer and Converter walkthroughs, a DI job demo, and a coding agent converting SAS with Alchemist Skills.
  Product videos by T1A: walkthroughs of the Analyzer and the Converter, a demo that converts a SAS DI job to Spark, the Live Analyzer in a browser, and a SAS to PySpark conversion run by a coding agent with Alchemist Skills.

### Delivered migrations

| Customer | Years | Source | Scale | Migration |
| --- | --- | --- | --- | --- |
| Major insurer, Benelux | 2022–2025 | SAS DI jobs | 1,600 ETL jobs | Company-wide enterprise data warehouse (EDWH) to Databricks, with SAS and Databricks running side by side during conversion. Delivered as a multi-phase program of several projects. 80% automation; 70% less UAT time |
| Healthcare company, Japan | 2024 | SAS Enterprise Guide | 180+ EG assets | Reporting from on-prem SAS Enterprise Guide to Azure Databricks, with a data mart and user training. 40% scope reduction in discovery |
| Luxury fashion house, Italy | 2026 | SAS Base scripts | 100+ scripts | End-to-end migration of predictive models to Databricks: business-driven model validation, automated validation agents, scalable pipelines. 50% faster runtime |
| Telco, Canada | 2022–present | SAS platform | Per business unit | Technical audit and business discovery of the SAS platform, then cloud enablement and workload migration, one business unit at a time. |

- 50+ SAS environments analyzed
- 3,000+ SAS DI jobs migrated
- 400+ Enterprise Guide flows migrated
- 2M+ lines of SAS code migrated

as of 2026

## Start with your own SAS code

Sign up, run the free Code Analyzer on your own assets and see the size of the job today. Or try the live demo without an account.

- Get Started: https://app.getalchemist.io/
- Try Live: https://live.getalchemist.io/

Sign-up to first analysis takes minutes; the [quick start](https://docs.getalchemist.io/quick_start.html) walks through it.

## What do evaluators ask before a trial?

Short answers to what evaluators ask first. For licensing details, see the [pricing page](https://getalchemist.io/pricing/#faq); for everything else, the [documentation](https://docs.getalchemist.io/).

### Which platforms does Alchemist support?

Inputs are SAS Base scripts (.sas), Enterprise Guide projects (.egp) and DI jobs (.spk). The output is PySpark for every PySpark-compatible platform: native on Databricks, including Databricks notebooks, and fully compatible with Snowflake and Microsoft Fabric. A SAS-to-Snowflake guide was written by T1A and published by Snowflake. Prophecy is a target for SAS DI sources.

### How long does a SAS migration with Alchemist take?

Even with automation, an end-to-end migration takes at least 2–3 months, mostly organizational time in larger companies. Mid-size projects have taken 4–6 months to move the business process fully. The conversion step alone ran up to 20× faster with 10× less manual effort than LLM-only conversion. That figure covers the conversion step, not the whole migration.

### Does my code leave my environment?

Not if you run Alchemist locally or inside your cloud platform. The Code Analyzer and Converter are binaries that run on your SAS assets where they sit. The full stack can also be deployed air-gapped, as a Docker image inside your own network. A hosted service also exists for teams that prefer nothing to install; connection details are in your Alchemist account.

### Do I need T1A services to use Alchemist?

No. The Assessment Kit and the Conversion Kit work standalone: sign up, download, activate the license and run, without talking to anyone. The Full Kit adds the end-to-end framework and other tooling that needs enablement and configuration for your environment. It therefore comes with T1A services, and is recommended only for migrations with hundreds of assets.

### Can coding agents drive Alchemist?

Yes, through Alchemist Skills and the API today, in your local coding agents or in your cloud platform's agent; an MCP server is in development. The agent delegates SAS-to-PySpark conversion to Alchemist and receives the converted code back. The conversion itself stays deterministic; the agent only orchestrates it. Setup is described on the Agents page and in the documentation.
