Systems nominal 7.5M executions / 30d
Usage export · enterprise identity · now live

Turn Excel Models into Production APIs

Upload your existing pricing or rating workbook, validate it against trusted results, and deploy a secure, observable API—without rebuilding the model in application code.

  • Desktop Excel-verified
  • OIDC + SCIM
  • SaaS or your AWS
  • Regression-tested
  • Usage-export ready
rating_v4.xlsx Live
Sheet: pricing130 in · 112 out
A
B
C
1
Input
Value
2
age
34
num
3
region
EE
str
4
cover
50000
num
5
6
Output
7
premium
287.42
=RATE*…
8
tier
STANDARD
=IF(…)
Compiled v4 · engine v11signed · cached
~/eucalyptus — zsh
$ curl -X POST api.eucalyptus.sh/v1/exec/rating_v4
  -H "Authorization: Bearer ctkn_9af2…"
  -d '{"age":34,"region":"EE","cover":50000}'

→ POST /v1/exec/rating_v4
 200 OK  · x-eucalyptus-lat: 1ms · hot

{
  "result": {
    "premium": 287.42,
    "tier": "STANDARD",
    "currency": "EUR"
  },
  "meta": { "version": "v4", "engine": "v11" }
}

The platform

A complete platform for turning Excel pricing models into governed production services.

01 — 04
01Upload

Upload Excel. Get a production-ready model.

Upload .xlsx or .xlsb workbooks, automatically discover inputs and outputs, preview the model, and resolve clear validation errors before compilation.

  • Automatic I/O discovery
  • Preview before you compile
  • Actionable cell-level validation
  • Reliable async compilation · any size
pricing_2026.xlsb · 4.2 MBCompiling
Parsed 83 sheets · 2.47M formulas
I/O discovered 130 inputs · 112 outputs · typed automatically
Preview sample execution matches sheet values
Validation 1 issue · actionable

Rates!B14 · WEBSERVICE() unsupported → replace with an input range.

Compile asynchronous · webhook on completion
Queue pos 1 · ETA ~40sNotify · webhook + email
rating · modelLive

Business · Excel

rating_v5.xlsxuploaded by maria.k · 09:14Validating

Engineering · API

POST /v1/exec/ratingpinned v4avg 2ms · hot 1ms
v5validation suite running18 / 24 cases
v4serving production trafficDefault
Promote when green · rollback is one clickPromote v5
02Source of truth

Keep Excel as the source of truth.

Business teams continue working in Excel while engineering receives a versioned, low-latency API. Update the workbook, run the same validation suite, and promote when ready.

  • Analysts never leave Excel
  • Versioned API · pin, promote, roll back
  • 1ms hot · 2ms average
  • Same validation suite every upload
03Regression tests

Prove every model change.

Save approved quotes and calculations as regression tests. Run them against every workbook version and see exactly which outputs changed before production promotion.

  • Approved quotes become test cases
  • Runs against every new version
  • Field-level output diffs
  • Gate promotion on a green run
regression · rating v5 ← v41 changed
24 saved cases · captured from approved quotes
Q-2198motor · EEall outputs unchanged
Q-2201motor · LVall outputs unchanged
Q-2214property · EEpremium changed

premium: 287.42 291.10 (+1.28%)

Q-2230motor · LTall outputs unchanged
23 unchanged · 1 changed · promotion heldReview diff

usage · last 30 days

Selected scope
Scope filters applied
Executions
7.5M
7,447,516 successful
Success rate
99.6%
Across selected scope
Avg latency
2ms
1ms hot / 89ms cold
Cold start
0.5%
38,145 cold executions
Cache hit
98.7%
7,383,059 mem / 0 disk
04Observe & allocate

Understand and allocate every execution.

Track execution volume, errors, success rates, and latency across models, customers, products, and deployment destinations.

Apply business labels to models and allocate usage by customer, product, brand, or business unit.
  • Volume, errors, success, latency
  • Slice by model, customer, destination
  • Business labels
  • Allocation-ready aggregates

Usage export · S3 + Parquet

Export usage into your data estate.

Deliver hourly or daily aggregate usage reports as versioned Parquet files directly into your S3 bucket. Query them with Athena and integrate them with billing, reporting, or cost-allocation workflows.

Usage exports contain allocation data—not calculation inputs, outputs, tokens, IP addresses, or pricing information.
s3://your-data/eucalyptus/hourly · versioned
date=2026-08-10/hour=13/usage_a41f.v3.parquet  12.4 KB
date=2026-08-10/hour=14/usage_a41f.v3.parquet  11.9 KB
date=2026-08-10/usage_daily.v3.parquet         88.1 KB
athena> SELECT label_customer, SUM(executions)
        FROM eucalyptus_usage GROUP BY 1;

  acme_re     541,204
  nordica     372,118
  baltic_re   164,332

Enterprise controls

Runs where you run.
Signs in how you sign in.

Deploy on your terms

Start with managed SaaS or run the execution plane inside your own AWS account and region. Keep calculation traffic, workbooks, compiled models, tokens, and execution history under your control.

Managed SaaSapi.eucalyptus.sh · eu-central-1Live
Your AWSproduction · private execution planePrivate

Stays in your account · traffic · workbooks · compiled models · tokens · history

Enterprise identity included

Connect your identity provider with OIDC SSO, map groups to application roles, and automate user provisioning and deprovisioning through SCIM 2.0.

OIDC SSOissuer verifiedConnected

idp:underwriting → role: model-editor

idp:platform-eng → role: admin

SCIM 2.0provisioning and deprovisioningAuto

Okta · Entra ID · any OIDC-compliant IdP

Signed bytecode

Content-addressed. No runtime formula injection.

Token isolation

Scoped to model, version, and environment.

Full audit log

Every execution, deployment, and rotation is traceable.

Security review

Penetration-test reports available under NDA.

Automation · MCP

Built for people and AI agents.

Manage the model lifecycle from the dashboard, API, CI pipeline, or an OAuth-secured MCP agent. Agents can upload workbooks, discover I/O, compile versions, run tests, and inspect logs and metrics through a least-privilege toolset.

DashboardREST APICI pipelineOAuth-secured MCP
Explore agent automation
agent ⇄ eucalyptus-mcp
▸ tool upload_workbook("rating_v6.xlsb")
  ← version v6 · compile queued

▸ tool discover_io("rating", "v6")
  ← 130 inputs / 112 outputs · typed

▸ tool run_tests("rating", "v6")
  ✓ 23 unchanged   △ 1 changed · premium +1.28%

▸ tool read_metrics("rating", { window: "30d" })
  ← execs 7.5M · success 99.6% · avg 2ms

Upload · Validate · Deploy · Observe · Allocate · Automate

Eucalyptus is the governed runtime for Excel-based pricing and calculation models—without translating the workbook into another programming language.

FAQ

Questions engineers and buyers actually ask.

Still stuck? [email protected] — real humans, 24h response target.

01Do I need to change my spreadsheets?+

No. Eucalyptus parses .xlsx and .xlsb workbooks as-is, discovers inputs and outputs automatically, and flags unsupported formulas at the exact cell before compilation.

02What do I get back when I call the API?+

Typed JSON. Discovered inputs and outputs become the I/O schema, and the response includes named outputs plus execution metadata.

03What happens when an analyst pushes a new workbook?+

It compiles to a new version under the same model ID. Regression tests run against it, you review changed outputs, and you promote when ready. Existing tokens remain pinned until you change the default.

04What is inside a usage export?+

Hourly or daily aggregate usage as versioned Parquet files delivered into your S3 bucket. Exports contain allocation data, not calculation inputs, outputs, tokens, IP addresses, or pricing information.

05Can the execution plane run in our AWS account?+

Yes. Run it in your own account and region so calculation traffic, workbooks, compiled models, tokens, and execution history remain under your control.

06Do you support SSO and SCIM?+

Yes. Connect an OIDC identity provider, map groups to application roles, and automate provisioning and deprovisioning through SCIM 2.0.

07Can approved AI agents work with my models?+

Yes. An OAuth-secured MCP surface lets approved agents upload workbooks, discover I/O, compile versions, run tests, and inspect logs and metrics through a least-privilege toolset.

08How is this different from a headless Excel host?+

Eucalyptus compiles the workbook to signed bytecode and executes it on a purpose-built runtime instead of opening Excel for every request. That keeps hot paths fast and deployment artifacts deterministic.

09How do you bill?+

Per compiled model and API call — never per seat. Evaluation starts on a free pilot; production runs on an Enterprise agreement sized to your call volume and deployment model. Talk to sales for a quote.

[ READY WHEN YOU ARE ]

Ship your first workbook in under an hour.

Start with a free pilot on your own workbook. Keep it as the source of truth — we'll take care of the rest.