Public website, contacts and the data lifecycle.

Last updated: 15 July 2026. This is a service transparency summary; it does not replace the privacy notice, contract or data processing agreement applicable to a specific deployment.

1. Parties and contact

The public website is operated by Ercoli Consulting srl. For privacy and TalkToMyExcel questions, write to ttme@ercoliconsulting.eu. In a demo or deployment, controller, processor and subprocessor roles depend on the scenario and must be defined before personal data is processed.

2. Public website and technical logs

The website consists of static HTML pages. It does not include external analytics, advertising pixels, web forms or profiling scripts and does not set its own application cookies. The hosting provider may still produce technical logs, such as IP address, user agent, timestamp and requested resource, according to its configuration and retention periods.

3. Email contact

If you write voluntarily, we process your address, signature, message content and any professional details needed to reply, manage the request or take pre-contractual steps. Messages are accessible to authorized personnel and technical email providers and are kept as long as needed for the request and applicable obligations.

Do not send attachments containing personal data, special-category data, trade secrets or business datasets in the first email. Describe the use case and format; we will agree on a suitable channel if a later exchange is needed.

4. Demos

We prefer synthetic or anonymized data for the first demo. If real data is necessary, before transfer we define purpose, roles, authorized people, AI provider, processing location, retention, deletion and a secure channel. The public website does not receive uploads or directly process datasets.

5. Data stored by the application

Actual behavior depends on how the application is deployed and configured. In the current implementation, uploaded files and prepared versions are saved in the workspace staging area; imported data is stored in DuckDB and, when semantic fields are selected, in a vector index. Questions and a limited conversation history are stored on the deployment's storage.

Removing a dataset deletes its tables, index entries and staging files managed by the application flow. Backups, logs, exports or files kept outside the application follow the rules set by the deployment operator. Before production, the operator must therefore define a retention policy, a verifiable deletion procedure and a backup and recovery plan.

6. Data sent to AI providers

During indexing, text from selected descriptive fields is sent to the configured embedding provider; a question used for semantic retrieval may also be converted into an embedding. To generate an answer, the language provider receives the question, recent context and evidence selected by the engine. The whole raw file is not sent as one prompt, but the evidence may contain data from selected rows.

Location, retention, training use and subprocessors depend on the chosen provider and plan. The recommended Regolo.ai configuration is oriented toward European inference and zero data retention; these guarantees must be checked against the effective contractual terms. A local endpoint can reduce external transfers but requires dedicated infrastructure management.

7. Security, rights and requests

Access controls, encryption, backups, logging, updates and incident response also depend on the deployment operator. Data subjects can contact the controller named in the applicable privacy notice to exercise their statutory rights and, where applicable, lodge a complaint with the competent supervisory authority. To identify the correct party or ask about the service, use the address above.