# Epicor Eclipse AI Assistant Sizing and Value Calculator

> Source: https://netray.co/tools/epicor-eclipse-ai-assistant-sizing-and-value-calculator
> Publisher: Netray (netray.co), AI and integration engineering for ERP and manufacturing systems.
> Section: Free Tools

## Short answer

Estimate GPUs, index size, hours saved and payback for a private AI assistant over Epicor Eclipse extract data.

## Category

epicor

## Erp

Epicor Eclipse

## Industry

Wholesale Distribution

## Audience

- CIO
- IT manager
- Finance controller
- ERP administrator

## Tool

**Kind:** calculator

**Currency:** $

**Inputs:**

#### Named users

**Key:** users

**Type:** number

**Default:** 90

**Min:** 1

**Step:** 10

**Unit:** users

**Help:** Users licensed to use the assistant.

#### Daily active share

**Key:** active_pct

**Type:** slider

**Default:** 50

**Min:** 5

**Max:** 100

**Step:** 5

**Unit:** %

**Help:** Share of named users who ask at least one question on a given day.

#### Questions per active user per day

**Key:** q_per_user

**Type:** number

**Default:** 12

**Min:** 1

**Max:** 200

**Step:** 1

**Unit:** questions

**Help:** Pilot logs give the real figure; 5 to 15 is common once habit forms.

#### Prompt plus answer tokens per question

**Key:** tokens_per_q

**Type:** number

**Default:** 3000

**Min:** 500

**Max:** 32000

**Step:** 250

**Unit:** tokens

**Help:** Question, rows from the reporting store and document chunks, plus the answer. Short branch answers keep this near 2,500.

#### Documents to index

**Key:** docs

**Type:** number

**Default:** 70000

**Min:** 0

**Step:** 1000

**Unit:** documents

**Help:** Price lists, product data sheets, contracts and branch procedures.

#### Average pages per document

**Key:** pages_per_doc

**Type:** number

**Default:** 4

**Min:** 1

**Max:** 500

**Step:** 1

**Unit:** pages

**Help:** Average across the document set.

#### Minutes saved per question

**Key:** minutes_saved

**Type:** number

**Default:** 3.5

**Min:** 0

**Max:** 120

**Step:** 0.5

**Unit:** minutes

**Help:** Branch lookups save a few minutes each; use the time to find stock or history across screens.

#### Loaded hourly cost

**Key:** hourly_cost

**Type:** number

**Default:** 36

**Min:** 0

**Step:** 1

**Unit:** per hour

**Help:** Salary plus benefits and overhead for the people asking.

#### GPU monthly cost

**Key:** gpu_monthly

**Type:** number

**Default:** 1100

**Min:** 0

**Step:** 100

**Unit:** per GPU per month

**Help:** Hardware amortised over 36 months plus power, cooling, support and rack space.

#### Throughput per GPU

**Key:** tps_per_gpu

**Type:** number

**Default:** 1100

**Min:** 100

**Max:** 10000

**Step:** 100

**Unit:** tokens/s

**Help:** Aggregate tokens per second under batching for your model and quantisation. Measure it; this is the most sensitive assumption.

#### One-time implementation cost

**Key:** impl_cost

**Type:** number

**Default:** 55000

**Min:** 0

**Step:** 5000

**Unit:** one-time

**Help:** Integration, document pipeline, security review and pilot.

**Outputs:**

#### Questions per day

**Key:** q_day

**Expr:** users * active_pct / 100 * q_per_user

**Format:** integer

**Description:** Named users x active share x questions per active user.

#### Tokens per day

**Key:** tokens_day

**Expr:** q_day * tokens_per_q

**Format:** number

**Description:** Questions per day x tokens per question.

#### Peak concurrent requests

**Key:** peak_conc

**Expr:** max(1, ceil(q_day * 0.25 / 3600 * 15))

**Format:** integer

**Description:** Assumes 25 percent of daily questions land in the busiest hour and each request holds a slot for about 15 seconds.

#### GPUs needed (before redundancy)

**Key:** gpus

**Expr:** max(1, ceil(q_day * 0.25 / 3600 * tokens_per_q / tps_per_gpu))

**Format:** integer

**Primary:** true

**Description:** Peak-hour tokens per second divided by throughput per GPU, rounded up. Add a spare GPU if you need failover.

#### Vector index size

**Key:** index_gb

**Expr:** docs * pages_per_doc * 2 * 1024 * 4 * 1.5 / 1000000000

**Format:** decimal

**Description:** Gigabytes, assuming 2 chunks per page, 1024-dimension float32 embeddings and 50 percent index overhead. Original files are extra.

#### Hours saved per month

**Key:** hours_saved

**Expr:** q_day * minutes_saved / 60 * 21

**Format:** hours

**Description:** Questions per day x minutes saved, over 21 working days.

#### Monthly value

**Key:** value_month

**Expr:** hours_saved * hourly_cost

**Format:** currency

**Primary:** true

**Description:** Hours saved x loaded hourly cost. Treat as an upper bound until a pilot confirms the time saved.

#### Simple payback

**Key:** payback

**Expr:** if(gt(value_month - gpus * gpu_monthly, 0), impl_cost / (value_month - gpus * gpu_monthly), 0)

**Format:** months

**Description:** One-time cost divided by monthly value minus GPU cost. Shows 0 when value does not exceed GPU cost.

**Chart:**

#### Monthly value vs daily active share

**Type:** line

**X:** active_pct

**From:** 5

**To:** 100

**Steps:** 20

**Series:**

- value_month

**Disclaimer:** Planning estimate only. Tokens per second, time saved per question and peak share vary widely; replace the defaults with measurements from your own pilot before committing to hardware.

## Content

**Intro:** This calculator sizes a private AI assistant over Epicor Eclipse extract data. It estimates GPUs for peak load, vector index size, hours saved and payback from assumptions you can edit.

**Sections:**

#### How to use it

**Body:** Enter named users, the daily active share and questions per user. Eclipse sites with many branches often see sharp morning peaks, so use measured busiest-hour counts when you have them. One-time cost should include the extract design and reporting store, which is the bulk of the work on this platform.

**Bullets:**

- Include extract and loader build in one-time cost
- Use realistic branch lookup times
- Keep tokens per question modest
- Sweep the active share to see adoption effects

#### The method

**Body:** Questions per day come from users, active share and questions per active user. The busiest hour carries 25 percent. GPUs = peak tokens per second / throughput, rounded up. Index = documents x pages x 2 chunks x 1024 x 4 bytes x 1.5 overhead. Hours saved = questions x minutes / 60 x 21; value is hours x hourly cost; payback = one-time cost / (value - GPU cost).

#### Reading the sweep

**Body:** The line shows value rising with adoption. The steepest gains come early as more people use the assistant; beyond about two thirds active, value flattens as heavy users dominate. Use the shape to set an adoption target for the pilot rather than assuming every named user will be active.

#### Common mistakes

**Body:** Frequent errors are assuming real-time data from an extract, forgetting extract and loader costs, and counting questions that Eclipse already answers in one screen. Be explicit about the as-of time so time saved is not offset by rechecks.

**Cta:** If you want the numbers checked against your branch usage and extract options, Netray can help. The calculator needs no contact.

**Faq:**

#### Why is one-time cost significant for Eclipse?

**Answer:** Because the data path is extract based. Designing, scheduling and validating extracts and a reporting store takes real effort, and every later answer depends on that foundation, so the cost deserves an honest figure.

#### How many GPUs does an eight-branch distributor need?

**Answer:** Often one 24 GB GPU for around 90 named users with a 7B to 14B model. A sharp morning peak or long catalogue context may require a second, so measure the busiest hour.

#### What does the value figure ignore?

**Answer:** It ignores fewer errors, faster customer responses and better buying, which can matter more than time saved. It also ignores the risk of stale data. Treat the figure as one input, not the whole case.

#### Can I reuse the hardware for other work?

**Answer:** Yes. A GPU server that runs an assistant for Eclipse can also serve other internal tools later. If you plan this, spread GPU cost across uses in the monthly cost input.

## Search terms

Epicor Eclipse AI sizing, Eclipse ERP AI ROI, distribution AI GPU sizing, Eclipse AI assistant cost, wholesale AI payback calculator, private LLM sizing calculator

---
Talk to an engineer: https://netray.co/contact?intent=erp_ai | Machine catalog: https://netray.co/agent/catalog.json