The verdict in three sentences
A Berlin distribution company handling 120 price requests a day spends the equivalent of five full-time staff writing quotes. An AI quote drafting project connected to the ERP costs EUR 25,000 to 60,000 in Berlin in 2026, with a marginal API cost of EUR 0.01 to 0.05 per quote. Budget and success depend mostly on the quality of the ERP connector (price lists, stock, customer discounts) and human approval before every quote goes out.
What the automation actually does
The typical flow has five steps. The inbound email (body, PDF or Excel attachment) is read by an LLM that extracts part numbers, quantities, requested lead times and the customer identity. The system then queries the ERP (SAP Business One, Microsoft Dynamics, Odoo, proALPHA) for the customer's net price, discounts by product family and stock availability. A draft quote is generated in the ERP in the usual format, then shown to the sales rep in an approval queue. The rep corrects if needed and sends in one click. Unrecognised part numbers are flagged in red, never invented.
| Step | Manual method | With AI + ERP | Time saved |
|---|---|---|---|
| Reading and sorting the email | 3 min | 5 s | 98% |
| Identifying part numbers | 7 min | 15 s (catalogue matching) | 96% |
| Retrieving prices and discounts | 5 min | 2 s (ERP API) | 99% |
| Formatting the quote | 4 min | automatic | 100% |
| Review and sending | 1 min | 2 to 3 min | negative |
| Total per quote | 20 min | 3 to 4 min | 80 to 85% |
Review time goes up slightly, on purpose. The sales rep moves from typing to checking. Across 120 quotes a day, effective work drops from 40 hours to about 7 hours.
What a project like this costs in Berlin in 2026
The ranges below are 2026 estimates from integrators and agencies in Germany. Berlin rates run 30 to 50% above a regional French city like Grenoble (EUR 15,000 to 40,000 there). The main cost drivers are the ERP and how clean the catalogue is.
| Item | 2026 range (excl. VAT) | Note |
|---|---|---|
| Scoping and email audit (200 samples) | EUR 3,000 to 7,000 | Measures share of usable requests |
| ERP connector (read prices, write quotes) | EUR 8,000 to 22,000 | More on legacy ERPs without an API |
| LLM pipeline (extraction, catalogue matching) | EUR 7,000 to 18,000 | Includes test datasets |
| Sales approval interface | EUR 5,000 to 9,000 | Web, integrated with Outlook or Gmail |
| Testing, training, go-live | EUR 2,000 to 4,000 | 2 to 3 weeks of parallel run |
| Total project | EUR 25,000 to 60,000 | 8 to 14 weeks |
| LLM API in production | EUR 0.01 to 0.05 per quote | EUR 30 to 150 a month at 120 quotes/day |
| Maintenance and hosting | EUR 400 to 1,200 a month | Monitoring, prompt updates |
A generic "email assistant" SaaS at EUR 30 to 60 per user per month is not enough: it knows neither your net prices nor your customer discounts. The ERP connector is where the value sits.
Response time and conversion rate
In B2B distribution, the first supplier to answer often wins the deal. B2B sales studies published in recent years agree on one point: the probability of closing falls sharply after the first few hours. Going from a 24-hour to a 2-hour average response changes your competitive position.
| Response time | Estimated conversion rate | Comment |
|---|---|---|
| Under 2 h | 30 to 35% | Customer has not yet chased competitors |
| 2 to 8 h | 22 to 28% | Comparison under way |
| 8 to 24 h | 15 to 20% | Current situation for many SMEs |
| Over 48 h | 8 to 12% | Quote often arrives after ordering elsewhere |
These rates are 2026 estimates to be checked against your own CRM data, but the order of magnitude holds across sectors.
Mini case study
Need a professional website?
Kolonell builds websites that attract clients, optimized for the Sénégalese market. Free quote in 2 minutes.
Jonas, sales director of an industrial supplies distributor in Berlin, runs 6 inside sales reps and 120 requests a day, with an average order of EUR 1,800 and a 28% gross margin. His project is priced at EUR 48,000, plus EUR 800 a month for maintenance and EUR 90 for the API.
Time saved: 33 hours a day freed up, the equivalent of 4 people moved to follow-ups and prospecting. Commercial gain: if conversion rises from 18% to 24%, that is 7 more orders a day, about EUR 12,600 in revenue and EUR 3,500 in daily gross margin. Over 220 working days, even dividing that gain by four for prudence, extra margin exceeds EUR 190,000 a year. The project pays back in under three months.
FAQ
Can the AI send quotes without human approval?
Technically yes, but we advise against it at the start. A price error on a EUR 15,000 quote costs more than 2 minutes of review. After 3 months, automatic sending can be enabled for simple requests (under 5 lines, known customer).
What happens if the requested part number is not in the catalogue?
The system proposes the 3 closest references with a confidence score and flags them to the rep. On a clean catalogue, 85 to 92% of lines are matched automatically.
Does our customer data go to OpenAI or Anthropic?
You can use API plans with a no-training commitment and EU hosting, or an open source model hosted with a German cloud provider (extra EUR 200 to 500 a month). The choice is made during scoping based on your GDPR policy.
Do we need to change ERP?
No. SAP Business One, Dynamics, Odoo or proALPHA expose APIs or connectors. For a legacy ERP without an API, we go through the database or imports, which adds EUR 4,000 to 10,000 to the budget.
How long until go-live?
Allow 8 to 14 weeks, including 2 to 3 weeks of parallel run where the AI drafts and the rep compares with the usual method.
Let's scope your project. Send us an anonymised sample of your requests and your ERP name: we will price the connector, the AI pipeline and the go-live timeline. Detailed quote within 48 h. WhatsApp +221 77 596 93 33.
Mohamed Bah
Fondateur, Kolonell
Passionate about digital and entrepreneurship in Africa, Mohamed has been helping Sénégalese businesses with their digital transformation since 2020. Founder of Kolonell, he believes every SME deserves a professional and accessible online présence.
