The shop works. The intake does not.

AI reviewed the website, email, phone records and shared drive and found 10 places where requests or time are lost before production starts. Each item has a monetary value and parameters you can adjust to your numbers.

total loss calculation
lost requests91,9 items×% to order×margin $+manual hours210 h×$/h
losses, $/month
$ 21 764
$/year
$ 261 173
covered by AI in stage 1
$ 11 592
Monthly inquiry flow
website96calls55email49messengers27reached calculation135lost92orders51did not buy84

Band width shows the number of inquiries. Red marks what is lost before calculation; insight parameters recalculate the flow live.

Your rating against the Smart Workshop standard
0 из 8 · частично 3 · зрелость 16 / 100

Smart Workshop is a set of eight capabilities that distinguish a workshop with a managed intake from one that depends on individual people. Below is your position against this standard. The insights that follow break down the losses behind each gap.

capabilityyour companySmart Workshop
Inquiries are accepted 24/7PartlyYes
Emails are processed automaticallyNoYes
Customers and inquiries are connectedPartlyYes
Files are classified automaticallyNoYes
The team works from a single workspaceNoYes
Routine work is automated2 processes8+ processes
Control is available through chatNoYes
Solution usage is measuredNoYes
insights · 10
insightparameterslosses
01No AI assistant on the websitestage 1
Visitors need an answer now: price, lead time and feasibility. Without one, they go to whoever replied.
visits/month% lost requests
$ 5 267/month
26,4 inquiries/month
metricvaluesource / calculation
visits/month2 400website counter, June
% lost requests1,1difference from assistant-enabled site conversion (2.4% → 3.5%)
lost inquiries26,42 400 × 1,1%
orders among them10,0conversion 38%
total losses$ 5 267/month26,4 × 38% × $ 525
what AI does:The assistant answers from your catalogue and price list, clarifies material and volume, captures the contact and creates a parametrized request.2 weeks
02No 24/7 LLM telephonystage 1
17% of calls arrive after 18:00 and on weekends. The voicemail does not capture a request and not everyone calls back.
out-of-hours calls/month% without callback
$ 6 061/month
30,4 inquiries/month
03Incoming correspondence is handled manuallystage 1
One inbox for four employees. Everyone reads the mail, the first available person handles it and some messages are read twice.
emails/monthminutes per email
$ 264/month
24 h/month
04Repeated searches across systems
Drive, email, the foreman's notebook and the manager's sheet. The same answer is searched for again every time.
searches/monthminutes per search
$ 198/month
18 h/month
05No one tracks unanswered emails
They are noticed when the client calls again. Median first response is 5 h 40 min; 68% arrive the next day.
unanswered requests/month% lost
$ 4 577/month
22,9 inquiries/month
06Messenger requests are not recorded
WhatsApp and Telegram live on managers' personal phones. When someone is away or leaves, the conversation leaves with the phone.
requests/month% lost
$ 2 424/month
12,2 inquiries/month
07Quotes from drawings are calculated manually
Dimensions are read by eye, coefficients from the technologist's memory, Excel is from 2019 and a Word quote is assembled for each client.
quotes/monthhours per quote
$ 1 547/month
141 h/month
08No file versioning
'final', 'final_2' and 'final_corr' sit in one folder. Recency is inferred from the modification date.
revisions/monthhours per revision
$ 264/month
24 h/month
09Only the shop knows the order status
The client calls, the manager asks the foreman, and the foreman checks the section log.
'where is the order' calls/monthminutes to clarify
$ 37/month
3 h/month
10Schedule slippage appears on delivery day
Section load is not calculated anywhere. The client learns of the delay on the day the order was due.
delays/monthdiscount/penalty $
$ 1 126/month
direct costs

If it does not work

What if AI brings no measurable benefit?

The first 30 days exist precisely to test that. We take one real process, measure the result, and expand only what is confirmed. If it does not work, you lose no more than the fixed $ 1 000 assessment budget and do not enter an expensive implementation blindly.

What if implementation takes too much time?

We do not replace your systems or require long meetings and employee training. The first stage is limited to one process and takes no more than 60–90 minutes of team time.

What remains if we decide to stop?

You keep a validated audit, process map, collected knowledge, test results, documented constraints, and an implementation roadmap. Even if you stop, you know exactly where AI works in your company, where it does not, and what makes sense to do next.

start assessment · $ 1 000fixed budget · 30 days · one process · 60–90 minutes of team time

method: email — 412 messages over 9 days; phone system — 288 June records; drive — 1,240 files; website — counter and form logs. 1C is not connected and cost price is excluded. Parameter values are assumptions agreed during the initial review; the rest is calculated from the sources.