Tacto in detail: what AI in procurement makes possible today
As an investor in Tacto, I follow the company's development with particular interest. For my network, I would like to put some of the latest developments into context: what concretely changes in procurement, and what benefit can companies draw from it?
Industrial procurement rarely lacks data. What is often missing is the connection between the data: a price sits in the ERP, the specification in a drawing, the reason for an award in an email. Tacto's current development starts exactly at this connection.
What interests me most is how scattered information turns into a verifiable alert, a comparable request for quotation or a better basis for negotiation. These workflows show concretely where the economic benefit is created.
1. From data overview to the next action
Tacto is built for industrial procurement. The platform is organised into Supplier Intelligence, Spend & Cost Intelligence, Sourcing Intelligence and Compliance Intelligence. Behind these are supplier data, cost analyses, requests for quotation and compliance documents. Specialised AI agents connect analysis and execution.
In my view, this is an important shift: a dashboard shows that a price has gone up. A supported workflow then helps to prioritise the case, gather the information and prepare an action. Value is only created when the signal turns into a decision that is actually implemented.
2. New possibilities for supplier management and requests for quotation
The redesigned supplier profile: context instead of searching for files
The fundamentally redesigned supplier profile brings together an AI summary, key figures such as purchasing volume and price development, and documents, contacts and activities. History and notes make changes and past communication traceable. The profile thus becomes the central starting point for working with a supplier.
The practical value lies in a shared starting point. When a buyer takes over a supplier at short notice, they first need to understand: what do we buy? Which issues are open? Which commitments have been made? A consolidated profile can make handover and preparation easier. This is particularly relevant for teams working across sites or with frequently changing responsibilities.
A summary, however, is only as reliable as its sources. For important decisions it should remain clear which document a statement comes from and which period a key figure refers to. "The supplier delivers reliably" is of little help if that assessment concerns a different material group or a different plant.
The RFQ module: from savings alert to comparable request
The new RFQ module connects identified savings potential with the request process. RFQ stands for "request for quotation". Items, templates and suppliers are prefilled; a request can be created directly from an identified potential. Automatic reminders and the extraction of quotation PDFs make the process easier. The quote comparison takes price components and award scenarios into account. The final decision remains with procurement.
This changes the economics of smaller requests. When preparing and evaluating a tender takes a lot of effort, often only the large items are reviewed regularly. If this effort decreases, smaller categories can also be put out to competition more often. That does not necessarily lead to lower prices, but it creates a better information base.
Supplier A charges €9.40 per part, supplier B €9.10. With B, however, tooling costs, freight and a higher minimum order quantity come on top. Only a common costing basis shows which option is more economical at the expected demand.
Currency, price unit, quantity breaks, delivery terms, quality requirements and one-off costs belong in the comparison. An automatically extracted unit price alone is not enough for an award decision.
Before using it at scale, it is therefore advisable to test a sample of difficult quotations: scanned PDFs, amendments, mixed currencies or prices per 100 pieces. What matters is not only whether the AI can read a document, but whether it makes uncertainty visible and treats missing information as missing.
3. Technical drawings become procurement knowledge
Drawing analysis is particularly interesting for manufacturing companies. Tacto extracts technical features into structured item profiles – for example material, weight, geometry, surface treatment and fits. Combined with transaction and market data, this creates a basis for making parts comparable and deriving target prices. Recognising similar or identical parts also creates a link to engineering.
Why does this matter? Two items can have similar names in the ERP and be completely different in terms of manufacturing. Conversely, two parts with different names can share the same main cost drivers. Only structured technical features allow more meaningful comparisons than an item number and a short text.
A target price is a well-founded orientation, not proof of the "right" supplier price. Batch size, set-up effort, scrap, capacity utilisation, inspection requirements or special delivery commitments can justify price differences. A good analysis makes its assumptions visible so that procurement and engineering can decide together whether a comparison holds.
For a pilot project, part families with sufficiently similar material, manufacturing process and volume structure are a good choice. First, check whether the features were extracted correctly. Only then does it make sense to assess the quality of the target prices. Otherwise an error in interpreting the drawing will be mistaken for a surprising savings potential.
4. Further use cases for AI agents
The agent portfolio includes specialised helpers for market deviations, volume growth, quantity-break pricing, maverick buying, payment terms and negotiation preparation. In addition, there are drawing and target-price agents as well as material compliance. Companies can also build their own agents. For use in a company, clear tasks, suitable permissions and defined approvals are decisive.
The benefit of this division of labour can be explained with three typical situations:
- Volumes grow beyond the previous calculation. A higher annual volume can be a reason to negotiate a new price scale. First, it has to be clarified whether the increase is permanent or caused by a one-off project.
- Orders bypass a framework agreement. A maverick-buying alert only helps if technically suitable contract items are available. Otherwise the deviation may well be justified.
- A material index falls while the item price stays the same. This raises a question to review and negotiate. The expected effect depends on the material share and the agreed adjustment mechanisms.
Order confirmations: review by exception
Another use case is reading order confirmations and matching them with purchase orders. Deviations are escalated to procurement based on thresholds the company sets itself. A savings tracker supports the follow-up of savings; managing compliance documents makes working with supplier and item documents easier.
In order matching, the benefit comes from fewer routine checks. Non-critical cases can pass through more quickly; attention shifts to deviations. This requires differentiated rules: a two-day delay can be uncritical for stock items but significant for a production-critical component. Partial quantities, changed units and substitute items also need business rules.
Compliance: simplify document work, keep responsibility
For compliance documents, the link between document, supplier, item and validity period is what matters most. A filed declaration is of little help if it refers to a different product version. Companies should therefore define their review requirements in advance and agree them with the responsible experts. Which documents are required and how they are assessed remains a question for each company and its products.
5. Business value: three different measures
The benefits of a procurement platform are easily mixed up. An identified potential, a negotiated price reduction and a saving that is actually booked are three different results. The same applies to time: shorter processing time creates capacity, but does not automatically reduce personnel costs.
| Benefit | Meaningful measure | Typical fallacy |
|---|---|---|
| Lower purchasing costs | Price actually paid compared at the same specification and adjusted quantity | Booking every identified potential as savings |
| Less processing effort | Working time per request or document, including rework | Measuring only the automated sub-task |
| Better coverage | Share of reviewed items, suppliers or order confirmations | Equating more alerts with more value |
| More robust decisions | Traceability of sources, error rate, time to resolution | Accepting precise-looking AI output as evidence |
A transparent sample calculation
Suppose a company handles 300 requests for quotation per year. If the average effort drops from four to two and a half hours, 450 hours are freed up on paper. At an assumed internal full-cost rate of €60, this corresponds to a capacity value of €27,000 per year. This is not a guaranteed Tacto effect and, at first, not a cash saving.
How large could the annual capacity effect be?
Scenario calculation, not a forecast. Licence, integration, training and operation have not yet been deducted.
In addition, with a truly addressable volume of five million euros, procurement could achieve a realised price improvement of one percent: €50,000 gross before additional costs. These figures, too, are freely chosen assumptions. For an investment decision, capacity value and cash savings should be shown separately. Overlapping measures must not be counted twice.
A robust calculation takes the total costs into account: implementation, data preparation, interfaces, licences, training and ongoing quality assurance. Implementation capacity also matters. A system can find more potential than the team can negotiate and qualify. Then priorities have to be set by expected impact, effort and supply risk.
6. What makes a good implementation
The most sensible starting point is a clearly limited process with a measurable baseline. A pilot over a few weeks can bring more insight than a broad rollout without a basis for comparison. I would proceed in six steps:
- Choose one problem. For example, quote evaluation in one category or order confirmations from selected suppliers. Record the goal, scope and owners in writing.
- Measure the baseline. Record processing time, rework, errors and lead time for representative cases. Particularly difficult documents explicitly belong in the sample.
- Clarify data and responsibilities. Item numbers, supplier identities, units of measure and drawing revisions must match. Define which system prevails in case of contradictions.
- Define approvals and limits. Who may send requests, change prices or accept confirmations? Which deviations must always be reviewed? Permissions must fit the task.
- Cross-check the results. Spot-check source references, extracted fields and economic statements. Also record the number of false alerts and missed deviations.
- Expand only after the evaluation. Scale up once benefit and quality, including rework, are confirmed. Translate unclear cases into rules or training needs.
Questions that belong in a product demo
A demo should use your own sample data. Can the system correctly assign an ambiguous price? Does it recognise an outdated drawing revision? Does it show why it recommends renegotiation? What happens when information is missing? Questions like these test everyday reality better than a purely prepared presentation.
For the technical assessment, interfaces and update frequencies are also important. It must be clear whether results are only displayed or written back into another system, and how errors are corrected in the process. For technical drawings and contracts, access rights, retention, use of data for models and subcontractors involved need to be clarified concretely. These points should be settled early in the specific project.
When the benefit can be limited
With few suppliers, infrequent requests or processes that are already highly automated, the additional effect can be small. With highly fragmented data, on the other hand, the potential benefit is high, but so is the implementation effort. And a company looking only for short-term price cuts may overlook the bigger effect of better collaboration between procurement, quality and engineering.
Transferability should also be checked. Tacto is focused on industrial procurement. Which functions are suitable for other areas of purchasing cannot be derived solely from a successful use with drawing-based parts.
7. What remains from the current developments
The decisive progress is the connection of the work steps.
The redesigned supplier profile makes context easier to access. The RFQ update connects identified potential with requests and quote evaluation. Drawing analysis unlocks information that procurement has so far found hard to use systematically. Its effect depends on how reliably data, expert review and implementation work together.
For companies, the best evaluation question is therefore: which specific process becomes faster, more traceable or more economical with our data – and can we prove it? Those who answer this question in a pilot can judge the value of the new possibilities on a sounder basis than general promises of AI or savings.
A personal look at Tacto from an investor's perspective · October 2026. More on Tacto and other vendors in our market landscape.