MCP App Store
by KeenEthics
Nteli Quoting icon

Nteli Quoting MCP Server for ChatGPT

By Nteli

Overview

Nteli Quoting is AI-native product and pricing configuration (configure-price-quote, CPQ) for manufacturing verticals. It covers engineered, configure-to-order equipment: data center pods (white space, power train and cooling), industrial control panels and switchgear, commercial HVAC and cooling towers, packaged pumps and skids, hazardous-area enclosures and panels, and electronic and cable assemblies. Estimators, inside sales and application engineers use it to turn an RFQ, a spec or an email into a priced, standards-checked quote. Paste a customer's request into ChatGPT. The engine works out which configurations are valid and asks only the questions that change the answer, one at a time, each with the price it leads to. Panels and data center pods are built from their parts: list the drives, starters and I/O, or the racks and kW per rack, and the engine sizes the enclosure, breakers, UPS, generators, busway and cooling. When requirements clash it says which rule is in the way and offers the ways out; jobs outside the standard range go to a Specials team instead of getting a guessed price. Values filled in that the customer didn't state are marked as assumptions and listed to confirm. The quote lists every part, what to confirm and the standards it was checked against (NEC, NFPA, UL, ASHRAE), with good / better / best options side by side; panels come with a preliminary submittal sheet. It all happens in an interactive card: tap an answer, assume a default, switch options, undo, submit for approval. Every price and compatibility answer comes from Nteli's rules engine, not the model. Sign in with a work email: the first person at a company creates its workspace and picks its product line.

How to connect

  1. ChatGPT: Open the Nteli Quoting listing in the ChatGPT app directory and choose Connect. ChatGPT listing Nteli Quoting exposes 73 tools here.

Tools

add_custom_line

ChatGPT
Add a part that isn't in the catalog yet (a special paint, a vendor's breaker, a larger window kit) at the sell price the estimator sets, per unit. cost is optional: until purchasing confirms it, the quote shows the line as subject to confirmation and a sales manager approves it on submit. sell_price and cost are the estimator's own figures: without one, ask for it; never suggest or estimate a price.

approve_submittal

ChatGPT
An engineer (or the company's admin or owner) approves the latest submittal revision: the title block takes their name and date, and the sheets read ISSUED FOR CUSTOMER APPROVAL instead of PRELIMINARY. A revision the assembly has changed since can't be approved: make a new one first.

assembly_add

ChatGPT
Add a group of identical devices to the panel, or under parent. type: starter, drive, breaker, io, relay or operator (door devices: e-stop, pushbutton, selector, pilot light, pot); its questions and value ids come from assembly_questions. stated: the question ids of the answers the customer or the person said; the rest are your choice, listed for the customer to confirm.

assembly_alternate

ChatGPT
An alternate priced side by side with the assembly in the proposal, e.g. soft starters instead of drives, or NEMA 4X instead of 4. changes: [{"path": "panel", "answers": {"nema": "n4x"}}] or a group as another type: [{"path": "panel/pumps", "type": "starter", "answers": {"start": "soft", "hp": "h10"}}]. It's re-applied to the assembly as it stands each time, so later changes carry into it. remove: an alternate's letter.

assembly_custom_line

ChatGPT
A line the catalog doesn't have, at the estimator's sell price: per panel (qty each) or once for the order (per_panel false: crating, a site visit). cost is optional until purchasing confirms it (an unconfirmed cost goes to approval). group: the group it belongs with (e.g. panel/pumps), else its own Custom group. remove: a line's id. sell_price and cost are the estimator's own figures: without one, ask for it; never suggest or estimate a price.

assembly_deal

ChatGPT
How many identical panels this quote is for (materials and shop labour are per panel; engineering and drawings once), the customer's channel and a project discount. Prices come from the company's price book.

assembly_fix

ChatGPT
Apply one of the fixes the assembly suggests (index from its fixes list).

assembly_options

ChatGPT
Every answer to one question of one node, tried before choosing it: whether the whole assembly still fits, the price difference, the problems it clears, and the new problems it would add (with where).

assembly_proposal

ChatGPT
What to send the customer for a panel assembly, ready to paste: price per panel and for the order, what each panel holds, the site facts (as the customer gave them, or as the estimator set them), the sizes chosen to fit the contents, notes and exceptions, the bill of materials by group, and the alternates priced side by side. Nothing is saved.

assembly_qty

ChatGPT
Set how many identical devices a group has.

assembly_questions

ChatGPT
The questions of the given node types (e.g. ["panel", "drive", "operator"]; for model data_center ["hall", "row", "rack"]) with each value's id and label: call it before assembly_start or assembly_add with only the types the request needs, then send these ids exactly.

assembly_remove

ChatGPT
Remove a group (and anything inside it) from the assembly.

assembly_set

ChatGPT
Set answers on one node of the assembly (path: "panel" or "panel/<group>"), e.g. {"cooling": "fan"}. A value of null takes the answer back (the engine picks again). stated (required): true when these answers were said, in the request, a document or by the person in this chat; false when they're your own choice (shown as such and listed for the customer to confirm). A site fact (fault current, supply voltage, environment) is only set when stated.

assembly_start

ChatGPT
Start a nested assembly (an industrial control panel holding groups of devices, or a data center pod). Data center questions (rack density, floor rating, cooling, redundancy, a utility letter giving fault current or voltage) are answered by building the pod here from the engine's sized power train and cooling. answers: the panel's answers; anything left out takes the engine's pick (the engine sizes enclosure, main, PLC, power supply and cooling to the contents). groups: device groups, each {"type": <type>, "qty": n, "answers": {...}, "name": "conveyor starters", "stated": [...]}. stated (required) lists the question ids of the answers that were said: written in the request, RFQ, spec or email, or said by the person in this chat ("35kaic" states fault_ka; "n4x ss" states nema). Send [] when none were. Every other answer you send is your choice: shown as such and listed for the customer to confirm. A site fact (fault current, supply voltage, environment) that isn't stated is held: the quote waits for it. Door devices (e-stops, pushbuttons, selectors, pilot lights, pots) are the operator type, one group per kind. model: control_panel (above) or data_center, a pod: answers are the hall's (redundancy and site facts; the power train and cooling are sized), groups are its rows ({"type": "row", "name": "row a", "groups": [racks...]}) and each row's groups are racks ({"type": "rack", "qty": 6, "answers": {"kind": "gpu", "kw": "k40"}, "name": "gpu"}). Panel types: panel (Industrial control panel (UL 508A)), starter (Motor starter), drive (Variable frequency drive), breaker (Branch breaker), io (PLC I/O module), relay (Interposing relay), operator (Door operator or pilot device). Pod types: hall (the root), row, rack. First call assembly_questions with the model and the types the request needs (always the root: "panel", or "hall" for a pod) and send their value ids exactly (e.g. fault_ka "k65", not "65"). A panel's supply voltage (supply_v: 208, 240, 480 or 600 V three-phase; 480 V is the 460 V motor class) is a site fact like fault current: send it only when stated. Motor loads come from NEC 430.250 at that voltage; never re-size the main yourself. source: the document the request came from (e.g. "RFQ-4471.pdf"), when the quote starts from an RFQ or spec; the answers taken from it count as the customer's. The card draws the backplate, the capacity gauges (DIN rail, heat, 24 V, PLC slots, main load), the problems and the fixes.

assembly_state

ChatGPT
The assembly as it stands (the id from assembly_start; without one, the open assemblies to pick from): tree, drawing, capacity, problems, fixes, price. The reply leads with anything the estimator changed in the quote card since you last looked.

assembly_step

ChatGPT
The assembly as a guided flow, one decision per step (site, motors, starting, controller, operator, extras, review): each question's answers as tiles with plain words, what each adds to the order and why one won't fit. step: one step's id, or "all". price: the steps whose answers get their price differences ("first" for the first step); by default every step's. The quote card prices the step on screen and the rest as the person moves; change answers with assembly_set / assembly_qty.

assembly_submit

ChatGPT
Submit a panel assembly as a quote (reps). The whole tree is checked again first: a panel with site facts missing, a group that can't be built as set, or anything that doesn't fit is refused with what's left to do. The quote carries the itemised BOM through the price book, the deal, the alternates, the site facts as given and the sizes the engine chose, stamped with the assembly model's version; policies decide who approves. revises: a quote this replaces (with reason).

assembly_submittal

ChatGPT
The submittal a customer's engineer approves before the PO: four tabloid sheets (cover and UL 508A nameplate ratings; front elevation and dimensioned backplate layout; one-line and calculations; bill of materials with catalog numbers from the price book, and notes), as a PDF. Each change to the assembly makes the next revision letter; a revision is PRELIMINARY until an engineer approves it (approve_submittal). Control panels only.

assembly_undo

ChatGPT
Undo the last change to the assembly.

choice_prices

ChatGPT
What each answer to one question would do to this quote, before choosing it: the total, the difference from now, and whether it conflicts with other answers or goes to Specials (with the reason). For a quote session (session) or a preview (model, facts, assumed, inferred, qty). Nothing is saved.

choose_option

ChatGPT
Switch the quote to one of the good / better / best builds offered with it: good is the basic valid build, best the most fully optioned, better in between. The customer's answers stay as they are.

close_session

ChatGPT
Close a session (default: your current one). Closed sessions can't be changed; submitted quotes are unaffected.

company_controls

ChatGPT
How strict approvals are for this company. With no arguments, shows the current controls (anyone may look). Admin only to change. self_approve_quotes / self_approve_rule_changes let a person who holds the approving role approve their own quote or rule change (for small teams where one person wears several hats; every self-approval is flagged in the audit trail). approvers picks the role that approves each policy, e.g. {"P4": "sales_manager"}: P1 Specials route, P2 discount above the person's authority, P4 relies on unverified rules, P5 large order, P7 margin below the floor, P8 custom cost not confirmed. large_order_usd sets the P5 threshold. estimators_see_margin lets estimators see cost and margin (managers and pricing always do).

configurator

ChatGPT
The whole configurator for a quote session or a preview: every question of the product line, grouped into sections, with every answer and what it would do (total, difference from now, or conflict / Specials with the reason). Nothing is saved.

confirm_custom_cost

ChatGPT
Record the real cost of a submitted quote's custom line once purchasing has it (and a new sell price if it changes). When every custom cost is confirmed the cost approval is settled; a margin below the floor then needs a sales manager.

customer_proposal

ChatGPT
What to send the customer for this quote, ready to paste into an email: the recommended build and its price, each difference from what they asked for and why (assumptions stated as such), priced alternatives, and the as-specified option (an engineered quote) when that differs. Prices are this quote's deal. Nothing is saved.

decide_quote

ChatGPT
Approve or reject a quote for the roles this person holds. The submitter can approve their own only when company controls allow it.

decide_rule_change

ChatGPT
Approve or reject a proposed change. Approval publishes a new pack version immediately; the proposer cannot approve their own change.

diagnose

ChatGPT
Every minimal set of the session's hard decisions whose removal makes the rest consistent, with the alternative values that would work for each. Use it to offer the customer all the ways out of a conflict.

enumerate_options

ChatGPT
Distinct valid combinations of the given attributes (default: all product attributes) within the session's decisions, cheapest first, each with its list price. Use it to lay options side by side.

explain

ChatGPT
Why is var=value available or not? If not: the smallest set of current decisions that rules it out, the constraints with sources, and which other answers would make it available.

fee_statement

ChatGPT
What this company owes Nteli: nothing for using it; a small share of each order its customers accepted online (month="2026-10", default all). Owners, admins and pricing only.

find_past_quotes

ChatGPT
Search quote history: by account, model sold, estimator, outcome, or words in the customer's email (e.g. text="Midland Div 2"). Newest first.

get_quote

ChatGPT
The priced bill of materials for the recommended configuration, with ANSI site checks.

get_state

ChatGPT
Current status, decisions, recommendation, quote, conflicts with repairs, ranked questions, routes, assumptions and warnings. detail="full" adds every attribute's allowed values and the reason each removed value was removed.

governance_report

ChatGPT
One page on how quoting was governed (managers, engineers, pricing, admins): quotes and totals, who had to sign off and why, time to approval, discounts beyond the usual band in dollars, quotes relying on assumptions, unmet safety standards, unverified rules in use, rule changes, self-approvals, security alerts, and whether the audit trail is intact.

history_report

ChatGPT
The go-live gate for a manufacturer: their past quotes replayed on a vertical pack (default: live version), the match rate, and each disagreement theme (with its id) with its question and the engineer's verdict, if recorded. Ready for go-live when no theme is open.

history_stats

ChatGPT
Quote history at a glance (all accounts or one): count, win rate, median hours to quote by hand, revisions, discount, most-sold models.

import_quote_history

ChatGPT
Load a manufacturer's past quotes for a vertical pack (engineers, admins). Each quote: {"id", "date"?, "account"?, "request": {"facts": {fact: value}, "qty"?}, "sold": {"configuration": {attribute: value}, "skus"?, "total"?}, "outcome"?}. For a spreadsheet, use the CLI: nteli.py import-history FILE.csv --pack.

invite_person

ChatGPT
Workspace owners: invite someone to this company's workspace by email (a contractor's address works too). roles default to ["rep"]; an owner can give rep, sales_manager, specials, pricing and owner. They get an email with a link that works for 7 days. People and join requests are managed at /admin/people in the web app.

keep_as_asked

ChatGPT
Put back what the customer asked for on one answer the engine recommended changing, and keep it that way. The engine recommends again around it; if nothing standard fits, the job goes to the Specials team as asked.

list_changes

ChatGPT
Rule and price changes with who proposed, who decided and their impact summary. pack names the product line each change applies to.

list_models

ChatGPT
Product lines this workspace can quote: each flat line with the customer facts it uses (fact ids, questions and allowed values), and each assembly line (kind "assembly": control panels, data center pods) with the tools that start it. Requests about these lines, technical ones included, are answered from the engine's rules.

list_people

ChatGPT
The people in this workspace, their roles, and what each role may do.

list_sessions

ChatGPT
Your open configuration sessions, newest first. Tools act on the current (newest) one unless you pass session.

mark_won

ChatGPT
The customer's order came in another way (a PO by email, a procurement system, a call): mark the approved quote won with its PO number. Same as the customer accepting online: the quote is won, and for an order from a new channel (online configurator or a dealer) Nteli's fee applies however the PO arrives. The person who quoted it, or an owner or sales manager. With assembly, returns that assembly's view (the quote card).

my_queue

ChatGPT
Quotes and rule/price changes waiting for this person's decision, with the reason each one needs them.

my_quotes

ChatGPT
Quotes you submitted, newest first: customer, total, status, who it is waiting on and why, and any decisions with comments. Use it for 'where are my quotes?' or 'did the Permian quote get approved?'. Managers, engineers, pricing and Specials can pass everyone=true to see all quotes.

new_session

ChatGPT
Start a quote on one of the workspace's product lines (model: its id; default: the workspace's first line). facts: what the customer said, e.g. {"area_class": "c1d2"}, using that line's fact ids and allowed values; inferred: answers the request implies without saying them (a pump area at a terminal is outdoors), listed on the quote as clarifications. The engine fills every other gap with the usual answer for a priced draft and marks as the person's call any gap where a wrong guess would change the price materially or mean a Specials job. When the answers can't be built as stated or would go to Specials, the engine quotes its recommended closest build instead (recommend=True): each change from the customer's spec is listed with why, the as-asked version stays one call away (keep_as_asked), and customer_proposal drafts what to send. as_asked: answers never to change. assumed: answers the estimator assumed, to confirm with the customer. decisions: the long form, [{"var": ..., "values": [...], "strength": "hard"|"soft"}]. Returns the session state. Quoting is free.

next_questions

ChatGPT
Open questions ranked by information gain over the recommended configuration. relevant=false means the answer cannot change the outcome. include_products also ranks product attributes nobody has picked.

optimize

ChatGPT
The best complete configuration for an objective (default: lowest list price) within the session's decisions, found by branch and bound, with the objective value.

override_price

ChatGPT
Set a price by hand, with the reason it goes on the record: total sets the whole quote's price (the difference shows as a price adjustment), or line plus unit_price sets one line's unit price (a part id or CUSTOM-n). Above the price book needs no approval; below it counts as a discount and the company's authority bands decide who approves; the margin floor still applies. clear=True removes the total override (or the line's, when line is given).

pack_versions

ChatGPT
Every published version of a product line's rules (model: the product line) with its fingerprint, and which one is live; matches_pack_file says whether a version is the one shipped with the product line.

price_book

ChatGPT
The company's price book: channels and their multipliers, quantity breaks, freight and surcharge, the margin floor, discount authority by role, and how many parts have their own list price or cost. Costs and margins show only to people who see them.

queue_summary

ChatGPT
Counts of everything waiting on you or your quotes: drafts in progress, quotes to approve, custom costs to confirm, quotes waiting on someone else, Specials, approved; and the five most urgent items. Managers and pricing see the team.

quote_preview

ChatGPT
Check a customer request without starting a quote: nothing is saved. Technical requests on a line this workspace quotes are answered here from the engine's cited rules: a pasted specification, a rating or code question (NEMA type, Class/Division or Zone, gas group, T-code, fault current), "what do we need for ...". Takes the same model, facts, inferred and assumed answers as new_session. The draft fills every other gap with the usual answer: it returns the draft price, its main parts and the standards it was checked against, the person's calls (gaps where a wrong guess would change the price materially or mean a Specials job, each answer priced), the clarifications the quote will carry, or a conflict with its ways out, or a Specials route. new_session with the same answers returns the itemised quote with each part's price and the good / better / best options. Where the chat shows apps, the quote card shows it and offers each answer as a tap.

recall_quote

ChatGPT
Take back a quote that is still waiting for approval, to change it: its open approvals are withdrawn and it shows as in revision. Revise it next (revise_quote); resubmitting gives a new number that links back. The person who submitted it or a sales manager.

recommend_fix

ChatGPT
When the current answers conflict or would go to Specials: apply the engine's recommended way out of each trade-off until the job can be built and priced, keeping the answers the person chose to keep as asked.

remove_custom_line

ChatGPT
Take a custom line (CUSTOM-n) off the quote.

remove_decision

ChatGPT
Withdraw the decision on one variable.

requote

ChatGPT
Start a new quote from a past one, with changes (e.g. {"qty": 6} or {"site_voltage": 220}). Runs it through today's rules, so anything that changed since is caught. Returns the new session's state.

review_packet

ChatGPT
Everything a reviewer needs: why they are reviewing, the configuration and lines, proof that it replays identically on the pack version it was built with, unverified rules and assumptions it relies on, and how it compares with past quotes.

revise_quote

ChatGPT
Open a submitted quote (approved, rejected or recalled) to change it: a new quote session starts from its answers, channel, discount, custom lines and overrides. Submitting it gives a new quote number that links back, and the old quote is marked superseded. Pricing guardrails (discount authority, margin floor, custom costs, large order) are checked again; other approvals carry over when the same reason still applies.

security_alerts

ChatGPT
Admin only: accounts paused for suspected rule extraction, with the pattern that triggered it.

set_decision

ChatGPT
Record an answer the person gave: a fact the customer stated or a choice the estimator made. Answers come from the person, not from inference; when they aren't sure, strength='assume' records the suggested default as something to confirm with the customer. Where the chat shows apps, the quote card offers the same answers as taps. Returns the new state and 'changed': values this answer ruled out elsewhere. reason: why an answer the customer gave is being changed (e.g. "customer confirmed Class 2"); it goes on the quote's exceptions to the customer's spec.

set_pricing

ChatGPT
Price this quote for a customer channel from the company's price book (e.g. end_user, contractor, distributor, oem) and/or with a project discount in percent off net. A discount above the person's authority goes to approval when the quote is submitted. The configuration is unchanged.

similar_past_quotes

ChatGPT
Past quotes most like the current one (same account, exposure, area class, power, inlet, options), with what was sold, the discount and the outcome. Use it for 'what did we do last time?'.

submit_quote

ChatGPT
Submit a configured quote for approval (reps): the session from new_session (without one, the open quotes to pick from), or a request. Prices come from the company's price book with the session's deal (channel, project discount, custom lines); discount_pct here sets the project discount. Policies decide who must approve: Specials routes, a discount above the person's authority, margin below the floor, custom costs not yet confirmed, unverified rules, large orders. Returns the quote with what it is waiting for. The quote lists its own exceptions to the customer's spec (answers changed from what the customer asked for); a note should describe what is quoted, not the original request.

undo

ChatGPT
Undo the last change to the session's decisions.

update_price_book

ChatGPT
Pricing or admin: change the price book. changes may set channels ({"distributor": 0.68}), default_channel, qty_breaks ([{"min": 5, "off_pct": 3}]), freight_pct, surcharge_pct, margin_floor_pct, cost_factor, discount_authority ({"rep": 10, "sales_manager": 30}), and per-part prices: {"items": {"<part>": {"list": 1200, "cost": 640}}} (null removes a part's own price). Every change is a new version in the audit trail.

usage_report

ChatGPT
Admin only: one month of usage (e.g. month="2026-10"; default this month): active people, configuration sessions, quotes submitted and approved, and engine calls, per person.

what_if

ChatGPT
Preview a decision without making it: which values it would remove (and why), which attributes it would fix, and the change in list price. If infeasible: what it clashes with.

whoami

ChatGPT
Who this person is, their roles, what they may do, and what is waiting for them. Call this first. actor can be omitted: the signed-in person is used. For a workspace owner it includes onboarding: what is set up (own prices, past quotes, go-live, colleagues, approvers) and the next step, with the web page to do it on.

Capabilities

WritesInteractive

Example Prompts

Click any prompt to copy it.

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/nteli-quoting"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/nteli-quoting)