With DIGIT-TM III now moving into execution, one question is coming up more and more often: can AI be used when preparing CVs and responding to Service Requests?
The tender itself answers this in Section 1.11, which sets out exactly what a tenderer or contractor must disclose about AI use. This post walks through what that section requires, and where AI does and does not sit inside the Sprint CV workflow built to operate under it.
- Section 1.11 of the DIGIT-TM III tender requires tenderers and contractors to disclose any general-purpose AI models used in tender preparation or in service provisioning.
- For each model, the tenderer must provide links to the provider’s public training-data summary, following the Commission’s required template.
- For general-purpose models placed on the market after 2 August 2025, the tenderer must confirm the provider adheres to the General-Purpose AI Code of Practice.
- This requirement ties directly to Regulation (EU) 2024/1689, the EU AI Act, and gaps in the disclosure can be grounds for the contracting authority to reject an offer.
- Most of the Sprint CV workflow does not use AI at all: template conversion, candidate matching, questionnaires, experience calculations, technology overlap calculations, and email automation are all rule-based.
- AI is used specifically for the optional AI Review and translations, and this step can simply be disabled.
- You can use Sprint CV for CV management, conversion, matching, and candidate workflows in full without ever activating AI Review.
- Where AI is used, Sprint CV discloses the models involved and documents the approach in its AI Transparency Statement, shareable with legal, security, and procurement teams.
What Section 1.11 of the DIGIT-TM III Tender Requires
Section 1.11, titled “Artificial intelligence (AI) products, services and deliverables,” sets a disclosure requirement rather than a general restriction on AI use. It applies to two moments: preparing the tender itself, and provisioning the service afterward.
“Tenderer must inform in the tender submission to the contracting authority: 1. If any AI systems have been used for the preparation of its tender, and if this AI system incorporates a general-purpose AI model; 2. If, for the provisioning of the service, the contractor will use any general-purpose AI models or if the contractor uses or will use an AI system that integrates or will integrate a general-purpose AI model.”DIGIT-TM III tender specification, Section 1.11
The section then specifies exactly what that disclosure must contain for each general-purpose AI model involved.
What information must a tenderer actually provide under Section 1.11?
Links to each model provider’s public training-data summary, and, for newer models, confirmation the provider follows the General-Purpose AI Code of Practice.
For each general-purpose AI model used, the tenderer must give links to the provider’s website or another publicly accessible platform where the training-data summary is published, following the Commission’s required template. For any general-purpose model placed on the market after 2 August 2025, the tenderer must also confirm the provider adheres to the General-Purpose AI Code of Practice. If it does not, the tender specification is explicit: that model cannot be used for tender preparation or service provision until the provider demonstrates compliance.
The obligation does not end at contract signature. Any addition or change to the general-purpose models used in the AI systems must be communicated to the contracting authority using the same disclosure format, and the same applies again at the level of each specific contract if it has not already been communicated to the main contracting authority.
The tender specification is also direct about why this matters. A separate note attached to Section 1.11 ties the requirement to the EU AI Act.
“The information requested from the tenderer under section 1.11 is essential for the contracting authority to conduct mandatory risk assessments and fulfil its obligations. If the information provided in this section contain clear gaps or omissions, incongruities or contradictions, or other apparent deficiencies, that would result in contracting authority’s non-compliance with mandatory legal requirements, the contracting authority reserves the right to reject the offer for failing to meet the requirements of Regulation (EU) 2024/1689 (AI Act).”DIGIT-TM III tender specification, note to Section 1.11
In practice, this means the quality and completeness of the disclosure matters as much as the decision to use AI in the first place. A vendor that can name its models, link to the required training-data summaries, and confirm Code of Practice adherence has already done the work this section asks for.
Where AI Is Not Involved in a CV Workflow at All
One way to reduce disclosure surface is straightforward: use AI only where it adds real value, and keep the rest of the process fact-based. That is how Sprint CV’s CV management for consulting companies is structured. Most of the workflow a CV moves through inside Sprint CV does not involve a general-purpose AI model at all.
| Workflow step | Uses AI? | What it actually does |
|---|---|---|
| CV template conversion | No | Maps sections from a source CV into the target template using configured field rules |
| Candidate matching | No | Compares required skills and experience against what is already in the profile |
| Candidate questionnaires | No | Generates a list of unmatched requirements for the candidate to confirm directly |
| Experience calculations | No | Calculates months and years of experience per skill from dated project history |
| Technology overlap calculations | No | Accounts for overlapping project timelines so experience is not double counted |
| Automated email workflows | No | Handles the back-and-forth around CV submission, matching, and candidate follow-up automatically |
| AI Review and translation | Yes, optional | Rewrites or tailors CV language to a specific Service Request, and translates content |
The CV parsing step that reads a source document and places its content in the right fields is extraction, not generation. It locates a section and copies it into the matching block without a general-purpose model rewriting anything, which is exactly why it falls outside the scope of Section 1.11.
Where AI Is Actually Used, and Why It Is Optional
The one place a general-purpose model enters the picture is AI Review. This step can rewrite a CV summary, restructure content to better align with a specific job description or Service Request, and translate a profile into another language on request.
AI Review is not a default a client has to enable it. It is a setting, and it can be turned off.
“If you want, you can just disable the AI Review, and then we keep the CV as it is. You don’t need to use it. It is configurable, and it can be disabled by default.”Marco Pincho, Founder & CEO, Sprint CV
If we disable AI Review, do we lose the rest of the CV management functionality?
No. The other features (template conversion, matching, questionnaires, calculated fields…) keep working exactly as before.
Turning AI Review off removes one optional step: the translation and the rewritten, job-aligned version of a CV. Everything else in the pipeline, from onboarding a consultant to generating a formatted CV and running it against a Service Request’s requirements, is unaffected because none of it depends on a general-purpose model to begin with.
How Sprint CV Meets the Disclosure Requirement
Where AI Review is used, Sprint CV discloses the models involved and confirms they meet the conditions Section 1.11 sets out. That approach is documented in writing rather than left as a verbal assurance. Sprint CV’s AI Transparency Statement covers which models are used, that they are accessed through paid, professional APIs, and that the data sent to those models is not used to train them. This is also written into the contract.“We are using paid APIs, professional ones. We do not use that data to train models. Every call we make is not training data, and we don’t keep it.”Marco Pincho, Founder & CEO, Sprint CV
For a bid or tender manager, the practical value is that this statement can go directly to a client’s legal, security, or procurement team as supporting evidence for the disclosure a Service Request may require, in the same format the tender specification expects.
Security and Compliance Beyond AI
AI disclosure is one part of a broader compliance picture. A vendor handling CV data for a European institution contract should also be able to speak to its general security posture, since that is usually the next question a client’s security team raises.
That includes certifications in progress, such as ISO 27001, and independent penetration testing carried out on the Sprint CV platform. These are reasonable things to ask for before a contract is signed, and a vendor doing the work should be able to share results as they come in.
- ✓Ask which general-purpose AI models are used, if any, and request the vendor’s AI Transparency Statement and Section 1.11 disclosure details
- ✓Confirm the AI step can be disabled entirely and check what functionality remains active without it
- ✓Ask whether data sent to any AI model is used for training and how it is retained
- ✓Check where certification work such as ISO 27001 stands and whether penetration test results can be shared
- ✓Confirm your data stays in your own environment, whether shared or dedicated, and who can access it
What This Means for Your Next Service Request
For a team preparing its first wave of DIGIT-TM III Service Requests, the practical takeaway is that AI use is a choice, not a precondition. Template conversion, matching, questionnaires, and the calculated experience fields run in Sprint CV without AI involved, so onboarding candidates and responding to a Service Request can start immediately.
Can we start using Sprint CV before our legal team has reviewed the AI disclosure?
Yes. The non-AI parts of the workflow are available immediately and do not depend on that review.
AI Review is the only step that touches Section 1.11 disclosure, and it is off by default in a new setup. A team can onboard candidates, run matching, and generate formatted CVs in Sprint CV while the AI Transparency Statement and contract terms are still working through legal, then turn AI Review on later if it is approved.
Our approach is simple: use AI where it brings value, keep the process fact-based, give the client control, and be transparent about where AI is actually being used.
If your team wants to see how the AI Review setting behaves before committing to anything, a short walkthrough is a faster way to answer that than reading a specification a second time.
Frequently Asked Questions
What does Section 1.11 of the DIGIT-TM III tender require on AI?
It requires tenderers and contractors to disclose any general-purpose AI models used in tender preparation or service provisioning, including links to each provider’s public training-data summary and, for models placed on the market after 2 August 2025, confirmation the provider follows the General-Purpose AI Code of Practice.
Which parts of Sprint CV’s workflow use AI?
Only AI Review, which rewrites or tailors a CV to a specific job description or Service Request, and the related translation feature. Template conversion, candidate matching, questionnaires, experience calculations, technology overlap calculations, and email automation do not use AI.
Can AI Review be turned off?
Yes. It is configurable per client and can be disabled by default. Disabling it does not affect any other part of the CV management workflow.
What is an AI Transparency Statement and why does it matter for a tender?
It is a written document describing which AI models a vendor uses, how they are accessed, and how data sent to them is handled, including whether it is used for training. It supports the disclosure a Service Request may require under Section 1.11.
Is candidate or CV data used to train AI models?
No, as described in Sprint CV’s AI Transparency Statement. Data sent for a specific call through the paid, professional APIs in use is not used to train the underlying model.
What happens if a tender’s AI disclosure has gaps or omissions?
Under Section 1.11, the contracting authority reserves the right to reject an offer if gaps, omissions, incongruities, or other apparent deficiencies in the AI disclosure would leave it unable to meet its obligations under Regulation (EU) 2024/1689, the EU AI Act.
Final Thought
For us, using AI is not about putting AI everywhere. It is about knowing exactly where it is needed, where it is not, and giving the customer the choice.
DIGIT-TM III asks tenderers to be specific about where general-purpose AI models are used. A workflow built so that most of it does not need one, with the one AI-dependent step disclosed and switchable, is a straightforward way to meet that requirement.
About the Author
Seven years of watching recruiters format the same CV three different ways for three different clients led Marco Pincho to build Sprint CV, an enterprise CV manager used by companies including VASS, Fujitsu, Sword Group, and Expleo to standardize CV creation and submission. He holds an Executive Master in IT Management from Solvay Brussels School and spends most weeks on calls with recruitment leaders and HR directors, working through documentation problems like the one covered in this article. Connect with Marco on LinkedIn
