Slow quotes can cost you orders. Pricing mistakes can cost you profit.

A customer sends a PDF with window dimensions, an Excel file listing the items and an email with a few changes. An employee reads the documents, copies the data into a pricing spreadsheet, finds the current prices and chases the missing details. Only then can they prepare a quote. When the customer changes the colour or quantity, some of that work has to be done again.

A business can work this way for years, until more enquiries arrive and prices start changing more often. Not every rule is necessarily written down, either. If only one person knows which product needs an extra manufacturing step, everyone else may have to wait for their answer. The quote is delayed, and every line still needs a careful check before it goes out.

We can build a system that reads customer documents, helps clarify missing information and calculates a price using your company's rules. It is worth considering in manufacturing and construction when orders involve similar products but their dimensions, materials, specifications or scope of work vary.

While one company is still preparing its quote, the customer may already be discussing a price with another supplier. But a rushed estimate can also be expensive if it leaves out work or materials. The challenge is to prepare quotes faster while keeping calculations verifiable and exceptions subject to approval.

Customers need a quote they can make a decision on

An email acknowledgement is not enough. The customer needs a price and a clear scope so they can compare options, agree the purchase with colleagues or plan the project. The less time an employee spends searching for information and copying it between files, the sooner they can prepare a quote and discuss it with the customer.

A faster calculation is useful early in the conversation, too, when the customer has not decided what to order. It helps establish whether the idea fits their budget and which choices have the biggest effect on the price. If the initial solution costs too much, the employee can propose a technically suitable alternative. The conversation can continue without immediately offering a discount.

Responding promptly to requested changes matters just as much. A window buyer chooses different glazing, a furniture customer wants different cabinet fronts, or the quantity changes on a CNC order. If the system retains approved data and pricing rules, the employee does not have to start again. They can recalculate the changed items and anything affected by them, then explain the price difference.

The customer can compare options, while the business sees how each one affects costs. A fast response alone does not guarantee an order. Customers also consider price, quality, reliability and delivery time. But a quote delivered promptly can be considered while the decision is still open.

A customer portal can accept enquiries outside office hours and immediately flag missing information. It should automatically prepare and send a complete quote only when the request fits tested calculation rules and the company's approval process. If specialist judgement is needed, the customer can be told what still needs checking and when the company expects to respond.

Questions about missing information can be collected against the relevant items in one place. The customer does not have to answer a separate email for every unclear field, and specialists can spend more time on unusual orders.

Before automating, it is worth mapping the whole workflow and showing how information passes from one employee to the next. NIST recommends this approach to identify unnecessary steps and opportunities to reduce lead time. In estimating, it can reveal whether the main delay is data entry, clarification or waiting for approval.

The method is described in NIST's guide to mapping work and information flows.

Faster calculations leave more time to check a bid

A manufacturer or construction company may receive several requests for quotations at once, each with its own deadline. If copying data consumes much of the available time, the team has to choose which bids it can prepare. Automating that work could make room to assess more suitable opportunities and examine complex calculations more carefully.

When a customer sends revisions shortly before the deadline, it helps to compare document versions, identify changed quantities and recalculate the affected items. Time saved can go into supplier questions, internal approvals and checking the submission. A finished estimate is not a complete tender response. The other required documents, conditions and submission deadline still matter.

The system can show quotes in progress, deadlines and responsible employees in one place. Everyone can see where a customer clarification, supplier price or manager's approval is still missing. This also exposes delays that faster arithmetic cannot solve.

There is little value in sending more quotes simply to increase the count. Each order still needs an assessment of whether the company can deliver it and whether the price makes commercial sense. If production is already full, additional orders first need a place in the schedule. An estimator's freed-up time can also be used to select orders more carefully and discuss them with customers.

Public procurement requires compliance with deadlines and documentation rules. Latvia's Procurement Monitoring Bureau, IUB, reminds suppliers that bids are assessed against predetermined criteria. Submitting earlier does not, by itself, mean winning. Faster calculations leave more time to prepare and check a complete bid.

The requirements are explained in IUB's procurement guide for suppliers.

Why a small estimating error can be expensive

Consider an illustrative example. A quote totals €10,000 excluding VAT, while the variable costs of fulfilling the order are €9,000. The remaining €1,000 is the order's contribution towards fixed costs and profit. The company's fixed costs still have to be paid from that amount, so it is not profit after all costs and taxes.

If the estimate leaves out a task costing €100, the error is just 1% of the sales value. Yet the order's planned contribution falls by 10%. The workshop is just as busy and the order still has to be completed, but less money remains to cover the company's other expenses.

Counting how many quotes the team produces is therefore not enough. Check how often omitted items appear after approval, which products were priced using outdated figures and which extra tasks regularly fall outside the estimate. This helps identify where automation could first prevent errors.

It is also important to distinguish markup from margin. If costs are €100 and you add a 20% markup, the selling price is €120. The difference is €20, but that represents a margin of approximately 16.7% of the selling price. The percentages use different starting amounts. If an employee enters a markup but the system interprets the number as a target margin, it will calculate a different price.

AI reads the document. Software calculates the price

AI can help find dimensions, quantities, materials and customer preferences in a specification. It can turn a free-text request into structured fields and identify missing information. The price itself is then calculated using the company's formulas and price tables.

Width is multiplied by height using approved dimensions. Material quantities receive the applicable unit price, working hours use the agreed rate, and discounts follow the approval rules and the employee's permissions. These calculations can be tested against examples with known correct results. If the inputs and rule version remain unchanged, the result should be the same.

AI helps collect and organise the information needed for the calculation. The software then applies formulas, tables and conditions that can be checked. It does not have to guess what the order might cost.

Generative AI, which produces text and other content, can present incorrect or invented information convincingly. NIST describes this risk in its generative AI risk management profile. An estimating system needs a practical response to it. If a material or labour price is unavailable, the item should be flagged and sent to an employee for review.

The risk of confidently presented false content is covered in NIST's generative AI risk management profile.

From customer documents to a quote

An employee can enter a request into the system, or the customer can submit it through a portal. In either case, retain the origin of the data and the record of when it was approved.

1. Save the request against the customer or project. The user attaches a PDF or Excel document in a supported format. Keep the original so its contents can be checked later.
2. Prepare the estimate's line items. AI extracts the data needed for the calculation. A dimension or specification can be linked to the document page or table location it came from.
3. Clarify what is uncertain. If the material, opening type or another essential choice is missing, ask a specific question. Answers can be selected from the options the company offers.
4. Review the inputs. The customer or employee confirms dimensions, quantities and choices. Make clear what this approval covers and where a specialist's check is still needed.
5. Calculate the price using the rules. The system applies the relevant price-list version, labour costs and agreed commercial terms. Items it cannot price go to an employee for review.
6. Approve and save the quote. The responsible employee reviews it under the company's procedure. The customer receives a PDF or access to a particular quote version in the portal.

Retain changes to the quote as well. If a customer chooses a different glazing unit after receiving the price, the previous approval cannot automatically cover the new specification. The system should show what changed, recalculate the affected items and request fresh approval where required.

Both the employee and customer should be able to tell whether the quote is complete. If some items are still unpriced, the calculation is unfinished. A provisional quote can be sent, but it must clearly state what is included and what remains to be calculated. Otherwise, the customer may misunderstand the total price.

Even this initial setup can reduce retyping and make checks easier. Once the calculations have been verified, the system can be extended with a customer portal, order planning and connections to other software.

What an estimate needs to account for in each industry

Reading the document is only the beginning. Window makers, roofers and metalworking companies use different materials, operations and conditions to arrive at a price. Build the system around a particular product or work category. The examples below show possible applications in different industries. They are not descriptions of specific CreateIT client projects.

Windows and doors

Two windows of the same size can have different prices because their specifications differ. The estimator needs to know the profile system, glazing, number of sashes, opening type, hardware, colour and accessories. Establish separately whether measurement, delivery, removal of old units and installation are included.

In an illustrative scenario, a customer submits a specification for twelve windows but mentions the colour only in the covering email. AI can help connect that choice to the relevant items. Before calculating, however, someone must check whether it applies to every window and to both sides. One clarification could prevent an error across all twelve items.

Specialist estimating software already exists in this industry. Logikal's documentation, for example, describes rules for profile, hardware and glazing compatibility, supplier prices and labour time. Importing tender items depends on the document format and workflow. AI document reading can be designed as an addition to the existing software, after checking the integration options.

These capabilities are described in the Logikal FAQ. A particular integration must be assessed separately.

Roofing materials and installation

Total area alone is not enough to quote for roofing materials and installation. You need to know exactly what was measured, the roof shape and which elements are included. The area to be covered does not, by itself, tell you how much material must be ordered or how much installation work is involved.

The calculation rules can include overlaps, offcuts, fixings, ridges, flashings and other items required by the chosen roofing system. The installation estimate may also need to include site access, removal of old roofing, waste disposal and lifting materials into position. The customer should see which tasks the quoted price covers.

Roofr, for example, uses measurements to prepare material and labour calculations. The user sets prices, waste allowances and markup or margin. It illustrates how measurements feed into a price; it does not establish an integration with Latvian suppliers.

This division of work is shown in Roofr's proposal tool description.

A materials estimate does not replace a structural calculation. A dimension extracted from a document does not prove that a structure can withstand the required loads or is suitable for its intended use. Where a designer or another specialist must make that decision, obtain it separately.

Construction work and bills of quantities

In construction, start with a defined group of tasks, such as partitions, finishing work or installation of a particular product. The system collects line items, finds the applicable unit prices and helps spot missing information. It cannot reliably establish the full scope of a project from incomplete documents.

Items counted automatically on a drawing still need checking. Procore's instructions require the scale to be set, catalogue and pricing matches to be checked, and detected and possible matches to be reviewed. Counting symbols is not the same as preparing a complete construction quote.

The checks are described in Procore's automatic counting instructions.

The quote should separately identify the known scope, assumptions, exclusions and allowances for conditions that have not yet been established. If the condition of the substrate has not been assessed, for example, tell the customer. A precisely written total cannot replace missing information about the site.

CNC machining of metal and wood

For computer-controlled machining, or CNC, an estimate needs to account for material, part geometry, batch size, machine setup, machining operations and tooling. Dimensional tolerances, surface finish and quality control requirements also affect the price. Setup costs are allocated differently across one part and a batch of one hundred identical parts.

AI can help read drawing notes, material designations and order conditions. Paperless Parts describes document processing that links extracted information to its source. The estimator decides whether to accept, correct or reject a suggested value. The company also states that its AI does not set a particular shop's prices.

These boundaries are explained in Paperless Parts' AI principles.

A digital part model created in computer-aided design software, or CAD, does not by itself tell you how long production will take. Computer-aided manufacturing software, or CAM, plans the machine's cutting operations. Autodesk Fusion estimates time using calculated toolpaths, feed and rapid movement parameters, and tool-change time. Geometry alone is therefore insufficient.

The calculation components are explained in Autodesk Fusion's machining time documentation.

First agree which part families can be priced using verified tables and historical data, and which need a manufacturing specialist or programmer. When batch size changes, recalculate setup cost per part. That does not mean the work that must be repeated for each part becomes smaller.

Furniture, kitchens and other bespoke products

Even made-to-order furniture uses recurring cabinet types, front materials, fittings and installation tasks. Dimensions, combinations and customer choices change. Repeated calculations can follow defined rules, while unusual structures go to a specialist.

In an illustrative example, a customer chooses a different material for the cabinet fronts after receiving the first quote. The system should recalculate the relevant items and associated work while preserving the earlier version for comparison. The employee can see what changed and explain the price difference. Any effect on manufacturing feasibility or lead time needs separate approval.

The same approach could support quotes for stairs, fitted wardrobes and other customised products. First define which choices are compatible and when a separate calculation is required. Written rules also help an employee who handles that kind of enquiry less often.

Manufacturing management software uses these rules too. Epicor's furniture solution describes rule-based choices of materials, dimensions and finishes, connecting the order to bills of materials and production.

An example appears on Epicor's furniture manufacturing solution page.

Plywood, veneered panels and made-to-order components

A panel manufacturer may receive a customer spreadsheet containing dimensions and quantities. Pricing may also require the material code, thickness, surface grade, coating and processing requirements. Grain direction can matter for decorative components. If the panels will be processed further, dimensional tolerances must be agreed.

The same thickness does not mean the same product. The European Panel Federation describes differences in plywood according to wood composition, bonding and surface. Use an approved product code and its corresponding data in the calculation. The system must not independently substitute a cheaper material simply because the dimensions match.

Material differences are explained in the European Panel Federation's plywood overview.

When components are cut from sheets, their layout on the sheet affects the price. Two orders can have the same total component area but require different numbers of sheets. The calculation may need to account for saw-cut width, permitted rotation of parts, usable offcuts and the company's material consumption rules. Include drilling, milling, edge or surface treatment, packaging and delivery costs too.

Calculate the cutting layout with specialist software rather than relying on AI to guess a waste percentage. Panel-processing software documentation shows how the cutting plan and machining steps affect consumption. Edge-banding material and the relevant machine's consumption settings are considered separately. These results can also support purchasing and production planning.

An example is provided in HOMAG's documentation on material consumption and grain direction. It describes software calculations, not an AI accuracy guarantee.

In an illustrative scenario, the customer first requests 200 components, then asks to compare a batch of 350 and a different coating. The system can retain the original specification and calculate both options using approved rules. The employee sees material consumption, setup costs and the contribution remaining after variable costs in each case. The customer can compare prices sooner, and production can later receive the exact version they approved.

Corrugated boxes and customised packaging

A customer may ask a packaging manufacturer to calculate the unit price of a box at different order quantities. In another case, the size, printing or split across deliveries changes after the first quote. The original specification and calculation need to remain available to assess these revisions.

A request can specify the box design, internal dimensions, board type, quantity, printing requirements and delivery terms. The design may be identified by a FEFCO code. The code alone, however, does not determine the required material or establish whether the box will adequately protect a particular product.

Design references are available in the FEFCO packaging design catalogue.

AI can help extract these details from documents and identify contradictions. The pricing software then applies the company's rules for materials, machine setup, tooling, printing, die-cutting, folding, gluing and delivery. If new cutting tooling or print preparation is needed, the quote must include those costs.

FEFCO's manufacturing practice document describes the main processing stages, specification review and traceability. A system can transfer the approved quote's data into a production job while retaining the correct specification and artwork versions. Integration with the company's existing software needs a separate assessment.

The process is described in FEFCO's Good Manufacturing Practice document.

In an illustrative example, a customer wants to compare orders for 1,000 and 5,000 identical boxes. The system can prepare both prices using the company's approved calculation logic. The salesperson can explain the difference, and the customer can choose a suitable quantity. A larger order does not automatically guarantee a lower unit production cost, so materials, batch arrangements and delivery feasibility still need checking.

Packaging strength and product protection must be assessed separately. If packaging is intended for food contact, its suitability for that use must also be checked. Neither a design code nor an automatically generated price establishes these properties.

A large factory can start with one product group

A larger business may already have enterprise resource planning software, or ERP, a product catalogue and an established pricing process. An AI tool can help at the stage where employees still collect and copy information manually. Salespeople may, for example, read customer emails and spreadsheets, select the required order details and enter them into the company's software.

A pilot can start with one sales team, one product group and a few supported document types. The system turns a request into a list of items, matches them to the company's catalogue and flags discrepancies. Prices, minimum batch sizes and customer-specific terms come from an approved price list or database. Delivery dates are confirmed after checking material availability and production capacity.

The approved quote's data can be used to prepare a bill of materials and a draft production job. Once the order is completed, planned material use and labour time are compared with actual figures. If milling a particular part family regularly takes longer than estimated, a manufacturing specialist can investigate the reason and approve changes to future calculations.

This makes it possible to measure time saved in the selected task and how often calculations need correction. Technical suitability, load-bearing capacity and required certifications must still be checked under the company's procedures. A correct price does not establish that a product is fit for its intended use.

When data is missing, ask for clarification

If a dimension has no unit, the system must not guess. If one document uses millimetres and another centimetres, the user should see how values were converted. A colour code absent from the company catalogue cannot be accepted without checking either. Ask the customer to clarify the choice or refer it to an employee.

Place the question against the item with missing data. For example, ask for the glazing choice on the particular window. That is clearer than a general warning that the data is incomplete. The employee should also see the relevant part of the original document, so checking the answer does not require rereading the entire request.

Some document-processing tools provide a confidence score for extracted data. That score concerns a particular field, not the accuracy of the entire estimate. Microsoft's documentation explains how it can help determine when a person should review a result. A correctly read dimension, however, does not establish that every item in the document was found.

The distinction is explained in Microsoft's documentation on accuracy and confidence scores.

Check that every page was processed and no items were omitted. Quantities should match the document, and the calculation should use the latest revision. When something is unclear, the tool should allow work to pause, retain what has been done and resume once the answer arrives.

Check new prices before putting them into use

Replacing a file in a shared folder does not ensure that everyone uses the right prices. The system can import an Excel or CSV price list and show changed prices, new items and unrecognised rows before it is approved.

Check units, currency, pack quantities, discounts and the effective date too. A price per metre is not a price per blank. A supplier's catalogue price may also differ from what the company actually pays. The import must account for those distinctions.

The company decides which calculations use the approved prices and from what date. Retain an already issued quote with the price-list version used to prepare it. Recalculating should create a new version so the previously approved document remains unchanged.

Material prices may rise while the customer is still deciding. The system can flag quotes for review, but the next step depends on their validity period and the company's terms. A new price list does not, by itself, justify changing an existing agreement without consent.

After a price change, rerun the saved test examples. This verifies that results change only where they should. If the price of a particular glazing type is updated, for example, the change should appear in calculations that use it. Unaffected items should stay the same.

If only prices change while the data structure and calculation rules remain the same, those price changes do not require retraining the AI model. Update the designated price list or database through the approval process. A new product, specification format or set of conditions may also require changes to extraction and calculation logic.

You may be able to keep your existing production software

If the company already uses specialist manufacturing or accounting software, its calculations can be retained. The new tool can assemble a customer's request and pass verified data to that software. Depending on the workflow, it may instead use prices exported from the existing program or a calculation produced there.

Choose an integration based on what is actually possible. Options may include the program's data export or an API, an interface that lets software exchange data. Check the specific version, licences, permissions and vendor support. A product name alone is not enough to promise a direct connection to any Klaes installation or other manufacturing system.

Before connecting programs, decide which one holds current prices and which assigns order numbers. Define what happens when a transfer fails. Retrying must not silently create another copy of the order. The employee should see whether the transfer is complete, will be retried or needs their attention.

A verifiable data export may be enough initially. Automatic two-way exchange can follow later. First remove repeated data entry while retaining the ability to trace each record to its source.

What an agreed 1% tolerance could mean

Before agreeing a tolerance, define what the system's result will be compared with. The reference can be an independently checked calculation approved by both parties. Both calculations must use the same approved inputs, price-list version and rules. This tests agreement between calculations. Comparing an estimate with actual manufacturing or construction costs incurred later is a different task.

If a review of the data and technical feasibility supports it, we can agree a maximum 1% deviation as a project acceptance criterion. Each estimate's total in a predefined test set would then need to stay within that limit relative to the reference calculation. The condition would apply only to agreed product groups, documents and the approved workflow. It is a target that still has to be demonstrated. This article does not claim that such performance has already been achieved or that it can be guaranteed in every situation.

For a positive reference total, calculate percentage deviation by dividing the absolute difference between the totals by the reference total, then multiplying by 100. VAT, discounts, markup and rounding must be applied consistently in both calculations. If the reference value is zero, define a tolerance in a specific amount or quantity instead. Percentage calculations cannot divide by zero.

A correct total is not enough. The deliberately constructed example below shows two errors cancelling each other out.

The total deviation is zero even though individual prices are wrong. If the customer later orders only part of the work, the error may also affect the total. Check not just the sum but whether all items are present and each has the correct quantity, specification and price.

Agree acceptable tolerances for individual items before testing. An incomplete estimate must not count as a successfully completed full estimate. Evaluate initially extracted data separately from the result after human corrections. An error fixed by an employee must not be reported as work the AI performed correctly.

Acceptance testing should also include examples not used to configure the system. Record the price-list and system versions tested, which calculations were produced automatically, which were referred to a person and what needed correcting. A few successful examples do not establish 99% accuracy for every future request.

This agreement tells both parties what must be checked before accepting the system. It does not guarantee that site conditions will remain unchanged, suppliers will hold their prices or every order will proceed according to the original plan.

From an approved quote to an order

An approved quote already contains information needed for purchasing, manufacturing, installation and customer communication. Entering it again in another program takes time and creates another opportunity for mistakes.

We can build a shared view of enquiries, unanswered questions, quote versions and the colleagues responsible for them. In a portal, customers can access their own company's projects and documents and approve selected dimensions, materials and specifications. Access follows each user's responsibilities. Customers should see only their own company's data, and employees only what their work requires.

Create the order from the specific approved quote version. The system should show who approved it and what changed afterwards. Special discounts, unusual specifications or exceptions can require additional approval.

Before promising a delivery or completion date, check material availability and production capacity. Confirm the date through the company's procedure. If the estimating tool does not yet receive these data, an employee performs the check and confirms the deadline.

How much time could you save, and how would you measure it?

Before introducing the tool, measure the tasks you intend to automate. The following numbers are fictional and illustrate the calculation. They are not CreateIT client results or a promise of savings.

Suppose a company prepares 120 quotes a month. An employee spends an average of 35 minutes working on each, excluding time spent waiting for answers. If a comparable quote took an average of 12 working minutes with the new tool, the monthly saving would be 2,760 minutes, or 46 hours. Those 12 minutes must not be measured using only the easiest cases. The average must include review, clarification and requests the system cannot process on its own.

At a value of €25 per employee hour, that time would be worth €1,150 a month. Suppose monthly system costs were €150. Subtracting them leaves €1,000 in the calculation. Implementation costs are not included in this example.

This does not mean an extra €1,000 will remain in the bank account every month. The employee may use the time for other enquiries, more careful checking or different duties. Cash is saved only when expenditure actually falls. The value of additional orders, meanwhile, depends not just on sales but on the contribution remaining after their variable costs.

Measure active employee time separately from the time the customer waits. Twelve minutes of work does not mean an answer in twelve minutes. The request may sit in a queue because the estimator is busy, a supplier price is missing or a manager has not approved it. Record when the initial request arrives, when all necessary inputs are available and when the first complete quote is sent. This also shows how long customer clarifications took.

During a pilot, compare the wait for the first complete quote, the time needed to revise it and the proportion submitted by the deadline. Record corrections, cases referred to staff and requests that could not be processed. An incomplete calculation must not be counted as a finished quote to improve a speed metric.

For sales, examine how many quotes customers accept and the contribution from completed orders. Compare similar product and customer groups, accounting for price changes, seasonality and differences in incoming requests. Only then can you assess whether the company is winning more worthwhile work and whether additional discounts or errors erode the benefit.

First establish what is worth automating

For the first pilot, choose a frequent task with a result that can be checked. It might be one window product group with known configuration options, a particular roofing system or a family of similar parts.

A tool can be especially useful where the business receives many similar requests, maintains a structured price list and has estimators spending substantial time retyping known information. If each order needs fresh investigation, collecting documents and preparing clarification questions may be the better starting point. Automatically calculating a final price may be premature.

If the pricing process is not written down, document it with the employees who use it. Development can turn their experience into testable calculation rules. The business still decides markups and customer-specific exceptions. Agree who is responsible for those decisions.

Collect both ordinary and more difficult requests for the pilot. Record what the estimator clarifies, which pricing rules they use and what they exclude. Prepare an estimate for the same request using the new tool and the existing method. Compare the results before allowing the system to handle customer-facing quotes more independently.

Completed orders provide useful evidence too. If planned working time regularly differs from actual time, investigate the cause and review the assumptions. GAO's cost estimation guidance emphasises defining scope and assumptions, assessing risk and updating estimates using actual costs. These principles also help review a company's estimating process.

More on establishing and maintaining an estimate is available in GAO's Cost Estimating and Assessment Guide.

Once the first product group has been tested, you can see which calculations can be reused and where different rules are needed. Decide on the next product group or software integration using what the tests have shown.

What to bring to the first conversation

Documents from everyday work are useful for the first discussion. You do not need a perfect requirements specification. A straightforward request and a difficult one, accompanied by an explanation of how your company handles them, can reveal much more.

• A few representative source documents with checked quotes and an explanation of any corrections.
• Current price lists and details of units, discounts, markup or margin, VAT and rounding.
• An explanation of labour, material allowances and configuration calculations, including extra items the estimator adds from experience.
• Examples with missing information, poor scans or contradictory revisions.
• The names and versions of existing software, available export files and information about possible data exchange.
• Enquiry volumes, active quote preparation time, customer waiting time and submission deadlines. Identify the employees who decide on exceptions.

One example containing all correspondence and documents from the first enquiry to the approved order is particularly valuable. It reveals not just the calculation but also clarification, waiting and repeated data entry.

You can collect this information in the estimating system preparation worksheet. It is a blank template to complete with your company's examples and questions.

An estimating system built around the way you work

When building business systems, CreateIT uses AI to work with documents and free-text requests. We implement company rules, access permissions and important calculations in software so they can be verified.

We can improve one stage of estimating first, then connect enquiry intake, pricing, a customer portal and order processing. Start with what most often delays a quote or causes expensive mistakes. We will agree the target preparation time, the calculations to test and how results will be assessed. The aim is to help your team deliver timely quotes for work the company can fulfil and that can generate profit.