Valerdat
Valerdat is a Spanish B2B SaaS assistant for purchasing and replenishment teams. It forecasts demand, recalculates stock policies daily and drafts optimised purchase orders that buyers approve in two clicks, plugging into an existing ERP.
What is Valerdat?
Valerdat is a Spanish B2B SaaS platform that sits on top of a company's existing ERP and carries purchasing teams from demand forecast through to a signed purchase order. Its publisher describes it as a digital purchasing assistant and as a Next Gen MRP powered by AI, aimed at consumer goods and industrial companies juggling hundreds or thousands of references across many suppliers.
The product is organised around three blocks. Demand planning rests on a forecasting engine that no longer tries one model at a time. Five families of models, market-standard algorithms alongside in-house ones, generate dozens of configurations, run in parallel and compete, with the best performer selected automatically for each individual reference and re-evaluated at every cycle, weekly, fortnightly or monthly. Historical data is combined with exogenous variables such as public holidays, market conditions and weather, and each day the previous forecast is measured against actual sales so the gap feeds back into the model. Where history is thin, at a launch or after an abrupt shift, the customer can enter their own plan and the system blends it with its predictive logic.
Dynamic stock policies recompute optimal stock levels, coverage, safety stock and reorder points for every product and warehouse pair, against more than fifty variables and constraints including warehouse capacity, minimum order quantities and shelf life. Purchase order generation then produces orders grouped by supplier, warehouse, date and logistics and optimised for cost, while suppliers are continuously assessed on lead time, price, price breaks and service level. The vendor states that customers approve 95% of proposed orders without changes.
Two more recent additions widen the audience. A conversational LLM is embedded in the platform and presented as trained on the customer's own database, covering products, warehouses, suppliers, sales and purchase history, stock policies, orders and metrics, with suggested questions to get users started. A collaborative layer adds in-app chat and comments attached to individual SKUs or product families, so Purchasing, Supply Chain, Sales, Marketing and Finance can settle a forecast in context rather than by email.
Valerdat connects to any ERP through API or SFTP, or failing that through CSV or Excel files, with named connectors for Sage X3, Sage 200, SAP S4HANA, SAP B1, Microsoft 365, NetSuite, Odoo, A3ERP and Cegid.
What it does
- Forecast future sales per reference using proprietary probabilistic Machine and Deep Learning models
- Recalculate optimal stock levels, coverage, safety stock and reorder points daily for every product and warehouse
- Generate purchase orders automatically, grouped by supplier, warehouse, date and logistics, and optimised for cost
- Score suppliers continuously on lead time, price, price breaks and service level
- Answer questions about your own purchasing and supply chain data in plain language through an embedded LLM
- Let several departments comment, justify and validate forecast changes inside the platform, at SKU or product-family level
- Push approved orders back into the ERP for execution
When to use Valerdat / When not to
A quick filter to help you decide if Valerdat is the right fit.
When to use Valerdat
- Purchasing and replenishment teams handling hundreds to thousands of references across multiple suppliers
- Companies still steering their buying from spreadsheets or from the plain reorder points of an MRP
- Distribution and fast-moving consumer goods businesses, the profile of the food distributor featured in the vendor's published case study
- Manufacturers already running Sage, SAP, Microsoft, NetSuite, Odoo, A3ERP or Cegid who want to augment that ERP rather than replace it
- Organisations where Purchasing, Supply Chain, Sales, Marketing and Finance have to agree on a single demand plan
When not to use Valerdat
- Buyers who want to sign up and see a price without talking to a salesperson: there is no public rate card, no free plan and no self-service trial
- Teams needing something running this week, since the vendor quotes a paid setup phase and roughly eight weeks after the connector is available
- Companies with no ERP or structured sales, stock and supplier history to connect through API, SFTP, CSV or Excel
- Buyers with strict, documented hosting or security requirements, as the site publishes no hosting location, no subprocessor list, no DPA and no security certification
- Anyone looking for sales, production or finance planning: the scope is strictly purchasing, replenishment and stock
How to use Valerdat
A typical end-to-end flow, from setup to results.
- Book a demo through the scheduling link published on every page; there is no self-service sign-up and no public price
- Agree the scope with the vendor, then go through the chargeable initial configuration phase described in the FAQ
- Set up the connector to your ERP, using an API, SFTP, or failing that CSV or Excel files
- Supply the historical demand, stock and supplier data the models need
- Allow roughly eight weeks after the connector is available before the first proposed purchase orders can be approved
- Log in to app.valerdat.com, where the interface is available in Spanish, English and Catalan
- Review the demand forecasts and the stock policies recalculated daily for each product and warehouse
- Question your own data in plain language through the embedded assistant, or start from its suggested questions
- Discuss and justify forecast changes with other departments in the in-app chat and SKU-level comments
- Approve the proposed purchase orders, which then return to your ERP for execution
Pros & Cons
Pros
- Covers the whole chain from forecast to purchase order, where many tools stop at the forecast
- The forecasting engine selects its own model per reference at every cycle, so no one has to choose or maintain a model
- Adds to the existing ERP instead of replacing it, and the vendor states the data stays in that ERP
- Named connectors for nine ERPs plus generic API, SFTP, CSV and Excel routes
- The embedded assistant makes purchasing data queryable in plain language, and the collaborative layer records why a forecast was changed
- A named customer case study is published, with an attributed quote from the buyer concerned
- Interface available in Spanish, English and Catalan
Cons
- No public pricing at all: no amount, no tier and no pricing page anywhere on the site, and a chargeable setup phase on top of the monthly subscription
- No free trial and no free plan, so the only way in is a sales demo
- Nothing published about where data is hosted, in which country or region or with which provider
- No subprocessor list, no data processing agreement and no security certification claimed
- No opt-out from training is documented, although the embedded model is said to be trained on the customer's own database
- No public API documentation, and no mobile application
- All performance figures are the vendor's own, and the published customer case study gives two sets of numbers that do not agree
Pricing & Plans
There is no free plan and no free trial, and no price is published anywhere on the site: the 128 URLs of the sitemap contain no pricing page, no tier and no worked example. The FAQ describes the commercial model without quantifying it, stating that starting the assistant requires an initial configuration process and that, once the connector is working correctly, a monthly subscription must be taken out in order to access the automated purchase orders. Because no figure is public, the starting price, currency and billing unit are deliberately left empty rather than filled with a misleading number. The only cost claim the vendor makes is comparative and unquantified: an implementation cost about three times lower than the sector average. Access is obtained by booking a demo.
Data, GDPR & hosting
A consolidated view of how Valerdat handles your data.
GDPR overview
Valerdat is established in Spain and its terms place the site under Spanish and EU law, with the courts of Barcelona having jurisdiction. Its privacy policy is written under Regulation (EU) 2016/679, which the cookie policy cites by name. Data subjects are told they may exercise access, rectification, objection, erasure, portability and restriction rights, either by writing to the registered office or by emailing valerdat@valerdat.com with the reference "Ref. Proteccion de Datos" and a copy of an identity document. The Spanish supervisory authority, the Agencia Espanola de Proteccion de Datos, is named as the route for complaints. Beyond that the record is thin: no DPO is appointed, no effective date is published, and there is no information on transfers outside the EEA, no subprocessor list and no DPA. Notably, the site never claims GDPR compliance in so many words; it applies the regulation without asserting conformity.
Who owns the data?
Valerdat, S.L., NIF B16688467, registered at Carrer de la Llacuna 162, 08018 Barcelona, Spain, is named as the data controller. Its privacy policy covers two populations only: people who fill in the website forms, and the legal representatives or contacts of customers who subscribe to the SaaS through app.valerdat.com. It says nothing about who owns the business data pushed into the platform, such as stock levels, sales history or supplier records. The nearest statement is operational rather than contractual: the FAQ asserts that the information stays in the customer's ERP, Valerdat extracting it, modelling it and returning purchase orders that flow back once approved. No subprocessor list is published and no data processing agreement is offered or mentioned.
Reuse rights
The published documents do not address whether the customer may freely reuse the data or the outputs the platform produces, so no permission regime is set out either way. What the site does state is operational: the FAQ presents Valerdat as a plugin on top of the ERP, extracting data, modelling it with the vendor's algorithms and returning purchase orders which, once approved, return to the ERP for execution, with the customer's information remaining in that ERP throughout. On Valerdat's own side, the privacy policy limits use of personal data to the listed purposes, which are answering form enquiries, giving access to resources such as webinars, white papers and ebooks, handling demo requests, running recruitment, and managing the SaaS contractual relationship. It states that no transfers to third parties are envisaged except where required to carry out those purposes or by legal obligation. Separately, the vendor writes that the embedded conversational model has been trained on the customer's own database; nothing is said about whether customer data feeds a model shared between clients, and no opt-out is described.
Data retention & training
Hosting summary
Valerdat publishes no information about where customer data is hosted. Searches across every page collected returned no mention of a hosting country, region, provider or data centre, no subprocessor list and no data processing agreement. The only jurisdictional anchor is the publisher itself: Valerdat, S.L. is established in Barcelona, its terms state that the site and its content are intended to comply with the laws of Spain and the European Union, and disputes are submitted to the courts of Barcelona. Its privacy policy is written under Regulation (EU) 2016/679 and names the Spanish data protection authority as the route for complaints, but says nothing about transfers outside the EEA. One technical detail should not be mistaken for an answer: the marketing site is published on Framer and resolves to an anycast AWS address in Amsterdam. That describes where the brochure site is served, not where the platform stores customer data. Anyone with hosting or residency requirements will have to obtain them from the vendor directly.
Things to keep in mind
Risks and trade-offs to weigh before adopting Valerdat.
- No price is published anywhere, and a chargeable configuration phase sits on top of the monthly subscription: budget only after a written quotation, never from the site.
- The vendor states its embedded model is trained on your own database, yet no opt-out is documented and nothing says whether your data feeds a model shared with other customers. Settle this contractually before connecting your ERP.
- Nothing is published on hosting location, subprocessors, data processing agreements or security certification, which is a real gap for a system given read and write access to your ERP.
- Every improvement figure is the vendor's own and unaudited, and the published customer case study states 24%, 37% and 1% in its text while its tiles show 19%, -30% and -22%. Treat the numbers as marketing until you see your own pilot.
- Handing reorder decisions to an engine that selects its own model can quietly erode a team's feel for its own demand patterns; the vendor's own answer, that 95% of orders are approved unchanged, is as much a warning as a selling point.
- The site advertises API connectivity but publishes no API documentation, so integration terms, limits and long-term portability have to be established in the contract rather than read.
- The registered office in the commercial register differs from the address published on the site, and the demo link points to a third-party scheduling calendar tied to one named employee: check current contact details before relying on them.
Setup & Integrations
Technical difficulty
Moderate, and a project rather than a sign-up. There is no self-service onboarding: the route in is a demo, followed by a chargeable initial configuration phase. A connector must then be established to the ERP, through an API or SFTP, or failing that CSV or Excel files, and historical demand, stock and supplier data supplied. Prebuilt connectors exist for nine named ERPs, and the vendor claims integration up to ten times faster than the sector average. Expect roughly eight weeks after the connector is available before the first proposed purchase orders can be approved, with Customer Success support throughout.
Deployment
Integrations
Supported languages
Behind Valerdat
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What does Valerdat actually do?
How much does it cost?
Is there a free trial or a free plan?
Which systems does it connect to?
How long does implementation take?
Does Valerdat train its AI on my data?
Where is my data hosted?
Is there a public API or a mobile app?
What happens to my data if I stop using Valerdat?
Who is behind Valerdat?
Should you pick Valerdat?
Valerdat is a coherent vertical tool rather than a general-purpose assistant, and that is its strength. It covers the whole chain from demand forecast to stock policy to purchase order, sitting above the ERP instead of trying to displace it, which keeps the integration argument credible and the customer's data where it already lives. The most substantial technical claim is the forecasting engine: five families of models competing in parallel, the winner chosen automatically for each reference and re-tested at every cycle, with the forecast measured against actual sales every day. That removes the model-selection problem that usually falls to a consultant, and it is a real design decision rather than a slogan. The recent additions, a conversational assistant working on the customer's own data and a collaborative layer that records why a forecast was changed, sensibly widen the tool beyond the buyer's desk to the Sales and Finance people who argue about the numbers.
The reservations are commercial and documentary rather than functional. Nothing about price is public: no amount, no tier, no trial, and a chargeable setup phase before the subscription even starts, so the only way to evaluate the cost is to enter a sales process. More importantly for a tool wired into an ERP, the site publishes nothing on where data is hosted, names no subprocessors, offers no data processing agreement and claims no security certification, while stating that its embedded model is trained on the customer's own database with no opt-out described. Every performance figure comes from the vendor, and the one published customer case study gives two sets of numbers that do not match each other. The publisher itself is verifiable: a Spanish company registered since September 2021 whose registered purpose matches the product exactly.
- Choosing a selection results in a full page refresh.
- Opens in a new window.