The answer lives in this podcast
What approach does Ana-Maria Thijssen recommend for prompting AI to find reliable external research and case studies?
Ask AI to include a reference alongside every idea it generates — not in a single bibliography at the end, but paired with each paragraph individually. Specifically request academic research, naming sources like peer-reviewed journals or Harvard Business Review. Then manually click every cited link to confirm it leads to a real, substantive source.
References inline, not at the end — the detail that changes everything
The most common mistake when asking AI to support research is requesting sources as an afterthought: a list tacked onto the end of a response. Thijssen's method is more demanding — and far more useful. She insists that each reference must appear directly alongside the paragraph it supports, so she can evaluate the source in context rather than cross-referencing a detached list after the fact.
This matters because it forces accountability at the paragraph level. If a claim cannot be paired with a verifiable source, it becomes visible immediately — rather than being buried in a credible-looking but unchecked bibliography at the end. This approach, detailed in Unprompted: Real AI in IC, reflects a disciplined workflow rather than a casual use of AI as a search engine.
Academic sources over general web results — and why the distinction matters
Thijssen does not ask AI to surface "research" generically. She explicitly prompts it to find academic research: peer-reviewed journals and established publications like Harvard Business Review, which carry editorial standards that general web content does not. For a change communications professional working to build credible stakeholder materials on a large SAP transformation, that distinction is not a nicety — it is a professional requirement.
The final step is the one most practitioners skip: manually verifying every link. AI models — including tools like Microsoft Copilot — are known to generate plausible-sounding citations that do not exist, a phenomenon often called "hallucination." As Thijssen explains in this episode of Unprompted, clicking through to check each source is non-negotiable, not optional.
AI models sometimes generate references — complete with author names, journal titles, and publication dates — that appear legitimate but do not correspond to any real document. The source looks credible in the output; the link either does not exist or leads to something unrelated. Thijssen's inline verification method is a direct, practical defence against this risk.
"AI became my virtual colleague. It's not a person that does my job. It's more like an entity that helps me bounce ideas and not feel so alone."
Ana-Maria Thijssen — Change Communications Lead, OCM/CMO team, large global corporation (SAP IT transformation programme)
Thijssen joined her current programme approximately a year and a half after it had already launched — with no formal induction and no time to ramp up gradually. As the sole communications professional embedded within an Organizational Change Management team, she operates across a global, multi-wave programme spanning dozens of locations worldwide. The next wave alone will cover 30 locations, compared to just 3 in the earlier phases. Her colleagues have been involved since January 2015, making institutional knowledge scarce and AI tools a critical substitute for the onboarding she never received. Her rigorous approach to source verification — detailed in Unprompted: Real AI in IC — reflects the professional standards she holds herself to when working without a comms team around her.
That framing — AI as a virtual colleague rather than an autonomous researcher — is precisely why the verification step is built into Thijssen's method. She treats AI as a capable but fallible collaborator: one that surfaces ideas and drafts at speed, but whose outputs require a human eye before they become credible, citable evidence. You can hear the full reasoning behind this approach in the episode on Unprompted.
See also
How did Ana-Maria Thijssen use Microsoft Copilot to develop audience personas for a global change communications programme?
Thijssen's team fed documents, org charts, communication models, and examples into Microsoft Copilot and asked it to identify the most distinct personas across the programme's global population — rather than building profiles manually from scratch.
How did Ana-Maria Thijssen use AI to get up to speed quickly when joining a mid-flight SAP transformation programme?
Thijssen joined the programme about a year and a half after it had started, with no time for proper onboarding, so she used AI to rapidly decode technical language and programme history that her colleagues had spent years accumulating.
How can AI act as a 'wingman' for internal communications professionals working alone on large change programmes?
Ana-Maria Thijssen describes AI as her 'virtual colleague' rather than a tool that does her job: it allows her to bounce ideas, check for best practice, and avoid the isolation of being the sole comms professional embedded in a large global programme.
Key takeaways
- Pair references with each paragraph as you prompt — never request a single source list at the end of a response.
- Specify academic sources explicitly: peer-reviewed journals and publications like Harvard Business Review, not general web content.
- Manually click every link AI generates — hallucinated citations are common, and a plausible-looking reference is not a verified one.
- Treating AI as a fallible colleague — not an autonomous researcher — is what makes this verification habit both necessary and sustainable, as Thijssen explains across the episode.
Hear Ana-Maria Thijssen walk through her full AI research method — including how she prompts, verifies, and applies external sources in a real global change programme.
Listen to the episode on Listenly