Answers whether you have done this before for a company like theirs, drawn from your actual project record rather than the nine case studies marketing happened to write up.
An AI agent for reference requests handles the experience question that decides most consulting evaluations: have you done this, in this sector, at this scale, recently. It establishes what the prospect is actually comparing against, searches your project record rather than your published case studies, describes relevant work at whatever level of detail your confidentiality allows, and routes a formal reference request to a person. Case study pages are a poor instrument for this. They cover the projects marketing had time to write up and a client agreed to name, which is a small and unrepresentative slice of what a consultancy has actually delivered.
Searches what you have done, not what you published.
Establishes the sector, scale and problem the prospect is comparing against
Searches the project record rather than the published case studies
Describes relevant work within your confidentiality constraints
Says plainly when you have no comparable experience
Routes formal reference requests to a person to arrange
Records which comparisons prospects keep asking for
A consultancy's strongest asset in an evaluation is relevant delivery history, and most of it is invisible. It lives in project records, in the heads of the people who ran it, and under confidentiality terms that prevent naming the client — none of which is the same as being unable to describe the shape of the work. A prospect asking whether you have done something similar is not asking for a logo, they are asking whether you will be learning on their money. Being able to answer that from the real record, anonymized where necessary, is a different conversation from pointing at a case studies page and hoping one of them is close enough.
A three-step flow from an experience question to a grounded answer.
The agent asks what the prospect is actually matching against — sector, scale, technology, problem shape — rather than accepting a vague ask for case studies.
It looks at delivered projects rather than published ones, and describes relevant work at the level your confidentiality terms permit.
Gaps are stated plainly. Requests for a named reference or an introduction go to a person, because those need client consent.
A prospect asking whether you have worked in their sector.
Scenario: a consultancy with nine published case studies was being asked for sector experience it had but had never written up. A prospect in a regulated manufacturing sector asks whether the firm has done a similar migration. The published case studies include none from that sector. The project record contains three broadly comparable migrations, two of which are under confidentiality terms that prevent naming the client. The agent describes all three by shape — sector, approximate scale, the technical problem and how long it took — without naming anybody, and notes that one client has previously agreed to act as a reference and could be approached. That request goes to a person, because arranging it needs consent. The prospect gets a substantive answer to the question that was actually asked, and the firm learns that its sector experience is invisible on its own website.
Anybody whose credibility rests on work they cannot easily show.
Delivery history is your strongest asset and mostly invisible.
Sector experience decides shortlists before price does.
You know the relevant projects; the website does not.
Reference selection is manual and starts from memory.
Repeated comparison requests tell you what to write up next.
Unnamed work can still be described in shape.
Searches the project record, respects confidentiality, routes consent.
Holds the project record with sector, scale, technology and constraints.
Stores delivery documentation the descriptions are drawn from.
Records which comparisons each prospect asked for.
Sends the written summary of comparable work.
Routes named-reference requests to the person who can seek consent.
Reports which sectors and problems prospects ask about most.
The experience questions a case study page cannot answer.
Questions about proving experience you cannot publish.
An AI agent for reference requests answers whether you have done comparable work: it establishes what the prospect is matching against, searches your project record rather than published case studies, describes relevant work within confidentiality limits, and routes formal reference requests to a person.
By describing shape rather than identity — sector, scale, problem, duration — which is usually permitted where naming is not. The constraints have to be recorded per project, because they differ, and treating them as uniform will either overshare or say nothing.
No. Approaching a past client to act as a reference requires consent and judgment about the relationship, and it is the sort of request a delivery lead should make personally rather than have automated on their behalf.
Say so. Consultancies are hired for judgment, and a firm that claims experience it lacks will be found out during delivery. An honest gap paired with adjacent experience is a credible answer; a stretched claim is not.
Your delivery documentation, and assembling it is the real work. Most firms discover no structured record exists — projects are described in proposals and closure reports that nobody has indexed by sector or scale.
It is the clearest signal marketing can get about which case studies to write next. Most firms choose what to publish based on which client agreed, rather than on what prospects keep asking to see.
Arguably better, because a small firm's delivery history is entirely in a few people's heads and therefore completely invisible online. The record is smaller and easier to assemble, and the credibility gain is larger.
Answers whether you have done this before for a company like theirs, drawn from your actual project record rather than the nine case studies marketing happened to write up.