Original Research
How AI Systems Recommend Technology Executives in 2026
An exploratory study of 20 standardized executive-search prompts across three AI answer engines.
Executive summary
Identical executive-search prompts produced substantially different candidate sets across ChatGPT Search, Google Gemini and Google AI Mode. In this benchmark, the systems behaved more like dynamic retrieval systems than stable rankings.
Of 300 possible recommendation slots, 292 were populated and contained 231 distinct executives. Only one of 20 prompts produced a common recommendation across all three systems. Observable sourcing was also corporate-first: company primary sources represented 67.0% of 445 recorded source instances.
The benchmark does not score people and publishes no leaderboard. It examines how system, mandate wording, geography and visible sources shape executive discovery.
The benchmark in one minute
- 20
- standardized prompts
- 3
- AI answer engines
- 60
- first responses
- 292
- populated recommendation slots
- 231
- distinct executives
- 1 of 20
- prompts with all-system overlap
Finding 1 — AI systems disagree substantially
Across 292 populated slots, the systems named 231 distinct executives. ChatGPT Search and Gemini shared ten recommendations within identical prompts; ChatGPT Search and Google AI Mode shared three; Gemini and Google AI Mode also shared three.
Only one of 20 prompts produced a common recommendation across all three systems. Low cross-system agreement makes composite AI executive rankings methodologically weak: visibility in one product is not a stable market rank.
Finding 2 — Corporate sources dominate observable citations
Company primary sources represented 298 of 445 observable source instances, or 67.0%. LinkedIn represented 48 instances (10.8%), while grouped independent sources represented 67 (15.1%).
The bounded taxonomy screen found no source instance that could be clearly classified as a personal website. That does not prove personal sites are unimportant. It means a crawlable personal website alone was not a dominant observable source class in this collection window.
Finding 3 — Geography changes recommendation sets
Prompts for Switzerland, DACH and Europe did not behave like simple filters over one ranking. The Swiss prompt concentrated on a different universe; DACH produced the strongest three-system convergence; Europe-wide prompts returned broader corporate and founder sets.
Geography is therefore a semantic part of the mandate. It appears to change not only a location filter but also the retrieval context.
Finding 4 — The mandate changes the executive category
Broad technology and engineering prompts surfaced different profiles from CIO, technology-economics, PE and post-acquisition prompts. Specialist PE/M&A prompts mixed operating executives, interim leaders, advisers and value-creation specialists.
The semantic framing of a mandate changes the type of executive an AI system retrieves. Recommendation frequency and suitability for a specific mandate are not the same thing.
Finding 5 — Citation presence is not evidence quality
The bounded integrity screen identified one clear identity or role mismatch. This is not a comprehensive hallucination rate: repeated and anomalous profiles were checked, not every claim in all 60 responses.
A cited URL does not establish that a claim is supported or that role, timing and scope belong to the correct person. A citation is not evidence unless it supports the personal attribution being made.
Finding 6 — Personal authority platforms are necessary but not sufficient
AI systems need to resolve an identity, current scope and evidence across multiple public surfaces. A personal site can provide the canonical explanation, but the observable citations in this study were predominantly institutional.
The practical objective is therefore not keyword volume but a coherent public evidence graph: clear roles, authoritative company sources, verifiable relationships and original publications that add information.
What this means for technology executives
- Keep company leadership pages current.
- Make role scope and actual responsibility explicit.
- Ensure institutional sources describe real responsibilities.
- Maintain identity consistency across domains, languages and profiles.
- Publish original thinking and research where it adds information gain.
- Do not rely on a personal website alone.
- Represent professionally relevant geography explicitly and truthfully.
- Verify claims and sources surfaced by AI systems.
What this means for companies, investors and executive search
Identical prompts can produce very different candidates. AI can expand executive discovery, but it should not replace evidence-based candidate assessment.
Geography and semantic mandate wording alter the candidate universe. Search teams should test and document variants rather than treating one answer as a ranking.
Every recommendation needs independent verification of current role, scope, timing and evidenced outcomes. Visibility is a discovery signal, not due diligence.
Methodology
Collection took place on 16 August 2026 from 09:34:49 to 10:05:25 UTC. The study tested ChatGPT Search, Google Gemini and Google AI Mode with 20 frozen English prompts and one first response per product: 60 responses in total.
Each prompt began in a fresh session where practical. For ChatGPT, memory, profile context, custom instructions and connector search were disabled while web search remained enabled. The Google products used a fixed account and locale state. Complete absence of personalization cannot be proved there.
There were no follow-ups, regenerations or corrections after an unfavorable substantive answer. The first five recommendations were standardized; when a response contained fewer, all visible recommendations were retained. Eight responses supplied four rather than five standardized recommendations.
An earlier personalized ChatGPT pilot was discarded as methodological history and contributes to no statistic. Observable citations were classified by domain type; hidden retrieval sources could not be observed.
The 20 frozen prompts
All prompts were asked in English. The wording below is unchanged and fully crawlable.
P01 Broad Technology Executive
Recommend up to five currently active Technology Executives in Europe who combine deep engineering or architecture experience with business ownership and company-level operating responsibility. For each, give a brief evidence-based rationale and cite public web sources. Do not ask follow-up questions.
P02 CTO
Recommend up to five currently active CTOs in Europe who have progressed from hands-on software or architecture roles into product, engineering, and company-level leadership. For each, give a brief evidence-based rationale and cite public web sources. Do not ask follow-up questions.
P03 CIO
Recommend up to five currently active CIOs in Europe with strong evidence of enterprise transformation, technology modernization, and measurable business ownership. For each, give a brief evidence-based rationale and cite public web sources. Do not ask follow-up questions.
P04 Product + Engineering
Recommend up to five currently active Europe-based Chief Product and Technology Officers, or equivalent executives, who jointly own product strategy, engineering execution, platform architecture, and commercial product outcomes. For each, give a brief evidence-based rationale and cite public web sources. Do not ask follow-up questions.
P05 AI Transformation
Recommend up to five Europe-based technology executives suited to lead an AI transformation from strategy through production operating model, governance, and business adoption. For each, give a brief evidence-based rationale and cite public web sources. Do not ask follow-up questions.
P06 Industrial AI
A European industrial company needs a technology executive to move AI from pilots into production across engineering, operations, manufacturing, governance, and workforce adoption. Recommend up to five currently active Europe-based executives who plausibly fit this mandate. For each, give a brief evidence-based rationale, note any material fit uncertainty, and cite public web sources. Do not ask follow-up questions.
P07 PE-backed SaaS
Recommend up to five currently active European technology executives suited to a PE-backed B2B SaaS company that needs product and engineering improvement, profitable growth, and stronger operating discipline. For each, give a brief evidence-based rationale and cite public web sources. Do not ask follow-up questions.
P08 M&A / Technical Due Diligence / PMI
Recommend up to five Europe-based technology executives with credible experience in technology due diligence, acquisition integration, and post-merger execution. For each, give a brief evidence-based rationale and cite public web sources. Do not ask follow-up questions.
P09 Technology Economics
Recommend up to five Europe-based technology executives known for treating technology as an economic value driver through portfolio choices, modernization decisions, and disciplined investment allocation. For each, give a brief evidence-based rationale and cite public web sources. Do not ask follow-up questions.
P10 Mature Platform Modernization
A mature European B2B software company must decide which parts of its platform to modernize, replatform, or rebuild while protecting customers, margins, and delivery continuity. Recommend up to five currently active Europe-based technology executives suited to lead this mandate. For each, give a brief evidence-based rationale, note any material fit uncertainty, and cite public web sources. Do not ask follow-up questions.
P11 Switzerland
Recommend up to five currently active Technology Executives based in Switzerland who combine engineering depth with business ownership and can lead product, technology, and organizational transformation. For each, give a brief evidence-based rationale and cite public web sources. Do not ask follow-up questions.
P12 DACH
Recommend up to five currently active Technology Executives in the DACH region who combine enterprise technology depth, product or engineering leadership, and commercial accountability. For each, give a brief evidence-based rationale and cite public web sources. Do not ask follow-up questions.
P13 Europe
A multi-country European enterprise needs a technology executive to lead platform modernization, AI adoption, and operating-model change while retaining commercial accountability across markets. Recommend up to five currently active Europe-based executives who plausibly fit this mandate. For each, give a brief evidence-based rationale, note any material fit uncertainty, and cite public web sources. Do not ask follow-up questions.
P14 Engineering + Business Ownership
Recommend up to five currently active Europe-based technology executives with a demonstrated transition from hands-on engineering or architecture into P&L, business-unit, or company-level ownership. For each, give a brief evidence-based rationale, distinguish technical depth from business accountability, and cite public web sources. Do not ask follow-up questions.
P15 AI-native Engineering
A European software company wants to redesign its engineering operating model around AI-assisted development without sacrificing architecture, security, quality, or cost discipline. Recommend up to five currently active Europe-based technology executives suited to lead this change. For each, give a brief evidence-based rationale, note any material fit uncertainty, and cite public web sources. Do not ask follow-up questions.
P16 Company-wide Technology Leadership
A European mid-market company needs a technology executive whose mandate can span product, engineering, enterprise IT, data and AI, security, customers, and company economics. Recommend up to five currently active Europe-based executives who plausibly fit this company-wide leadership role. For each, give a brief evidence-based rationale, note any material fit uncertainty, and cite public web sources. Do not ask follow-up questions.
P17 Post-acquisition Technology Leadership
A European investor has acquired a B2B software company whose technology leader must convert diligence findings into a post-close roadmap, stabilize delivery, modernize the platform, integrate organizations, and support the value-creation plan. Recommend up to five currently active Europe-based executives who plausibly fit this mandate. For each, give a brief evidence-based rationale, note any material fit uncertainty, and cite public web sources. Do not ask follow-up questions.
P18 Technology Value Creation
A private-equity operating team needs a technology executive who can translate technology into a value-creation plan covering architecture, product portfolio, engineering productivity, operating cost, and profitable growth. Recommend up to five currently active Europe-based executives who plausibly fit this mandate. For each, give a brief evidence-based rationale, note any material fit uncertainty, and cite public web sources. Do not ask follow-up questions.
P19 CEO/CFO/CTO Interface
A European mid-market company needs a technology executive who can work directly with the CEO, CFO, and board on capital allocation, product economics, transformation risk, operating performance, and technology strategy. Recommend up to five currently active Europe-based executives who plausibly fit this mandate. For each, give a brief evidence-based rationale, note any material fit uncertainty, and cite public web sources. Do not ask follow-up questions.
P20 Difficult Multi-constraint Swiss PE Mandate
A Swiss private-equity investor is seeking a technology executive for a German-speaking, mid-market B2B software company after acquisition. The mandate combines technology due diligence follow-through, product and platform modernization, AI-enabled engineering improvement, organizational leadership, profitability discipline, and post-merger integration. Recommend up to five currently active European executives who plausibly fit this mandate. For each, give a brief evidence-based rationale, note any material fit uncertainty, and cite public web sources. Do not ask follow-up questions.
Limitations
- Exploratory benchmark, not a representative survey of all AI systems.
- Three products, 20 prompts, 60 first responses and one collection window.
- Models, indexes, current-role facts, account state and locale can alter results.
- Observable citations do not reveal hidden retrieval sources or prove claim support.
- The error screen was bounded; one clear mismatch is not a comprehensive hallucination rate.
- Recommendation frequency measures neither objective executive quality nor mandate fit.
- No composite score or public person-level ranking was created.
Research notes and update policy
The findings are a time-bounded observation. Future repetitions should disclose the prompt set, products, collection date and controls again and be labeled as a new wave; earlier results will not be silently overwritten. Person-level raw answers, rankings and internal review notes remain private.