# Epicor ERP 10 AI Assistant Sizing and Value Calculator

> Source: https://netray.co/tools/epicor-erp-10-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 on Epicor ERP 10 on-premises.

## Category

epicor

## Erp

Epicor ERP 10

## Industry

Discrete Manufacturing

## Audience

- CIO
- IT manager
- Finance controller
- ERP administrator

## Tool

**Kind:** calculator

**Currency:** $

**Inputs:**

#### Named users

**Key:** users

**Type:** number

**Default:** 120

**Min:** 1

**Step:** 10

**Unit:** users

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

#### Daily active share

**Key:** active_pct

**Type:** slider

**Default:** 40

**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:** 6

**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:** 3500

**Min:** 500

**Max:** 32000

**Step:** 250

**Unit:** tokens

**Help:** Question, retrieved BAQ rows and document chunks, plus the answer. Smaller models with shorter context keep this near 3,000.

#### Documents to index

**Key:** docs

**Type:** number

**Default:** 60000

**Min:** 0

**Step:** 1000

**Unit:** documents

**Help:** Job packets, drawings, certs and SOPs on the shares Epicor references.

#### 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:** 5

**Min:** 0

**Max:** 120

**Step:** 0.5

**Unit:** minutes

**Help:** Time saved versus asking a planner or running a BAQ yourself.

#### Loaded hourly cost

**Key:** hourly_cost

**Type:** number

**Default:** 48

**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:** 1200

**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:** 1000

**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:** 45000

**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:**

#### Simple payback vs minutes saved per question

**Type:** line

**X:** minutes_saved

**From:** 1

**To:** 15

**Steps:** 29

**Series:**

- payback

**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 for Epicor ERP 10 on-prem. It estimates peak GPU need, index size, time saved and payback, using assumptions you can change.

**Sections:**

#### How to use it

**Body:** Enter named users, the share active on a typical day and questions per active user. Set tokens per question to include retrieved BAQ rows and document chunks. Enter your one-time implementation cost to see payback, and the sweep shows how sensitive payback is to the minutes saved per question.

**Bullets:**

- Smaller plants often fit on one 24 GB GPU
- Do not count licence holders who rarely log in as active
- Include integration and security review in one-time cost
- Compare against simply building more BAQs

#### The method

**Body:** Peak load is modelled as 25 percent of daily questions in the busiest hour, converted to tokens per second and divided by GPU throughput. Concurrency assumes each request occupies a slot for about 15 seconds. Index size multiplies documents, pages, chunks per page and a float32 embedding with overhead. Value is hours saved over 21 working days times loaded cost.

#### Reading the sweep

**Body:** The payback line falls sharply as minutes saved rise, then flattens. If payback is long at your realistic minutes-saved figure, the case may rest on fewer errors or faster quotes rather than time alone. A value near GPU cost means a smaller model or shared hardware may be the better choice.

#### Common mistakes

**Body:** The usual mistakes are applying the busiest-hour throughput from a benchmark to mixed real prompts, ignoring retrieval context in tokens, and treating ERP 10 as static. If an upgrade is coming, keep hardware reusable and BAQ names stable.

**Cta:** If you want the numbers checked against your ERP 10 usage, Netray can review them with you. The calculator needs no contact.

**Faq:**

#### Is one GPU enough for an ERP 10 plant?

**Answer:** Often yes for 50 to 250 named users with a 7B to 14B model, because peak concurrency is low. Confirm with measured throughput and your busiest-hour question volume before buying.

#### Why is payback zero in some cases?

**Answer:** Payback shows zero when monthly value does not exceed monthly GPU cost, because the project would not recover its one-time cost. Increase realistic minutes saved or reduce cost, for example with a smaller model or shared hardware.

#### Does the calculator include licensing?

**Answer:** No. Open-weight models have no per-user licence, but your own server, database and operating system licences are outside this estimate. Add them to one-time or monthly cost if they apply.

#### How should I estimate minutes saved?

**Answer:** Time a handful of real questions the old way: finding a job status or shortage across screens and BAQs. Use the average of those timings, then discount for questions where the assistant is wrong or incomplete.

## Search terms

Epicor 10 AI sizing, Epicor ERP 10 AI ROI, private LLM GPU sizing Epicor, Epicor 10 chatbot cost, on-prem AI payback calculator, ERP AI GPU calculator

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