AI strategy consulting
Where AI helps your institution — and where it doesn’t, yet.
Every week someone is pitching your institution an AI initiative, and every pitch arrives with the same certainty. Strategy is the opposite of that: deciding where AI genuinely helps your work, where it does not yet, and what your record can actually support — decided before anything is bought, by people whose recommendation is allowed to be “not this, not now.”
The engagement
Four questions, answered honestly.
An AI strategy engagement exists to answer the questions a vendor pitch is structured to skip.
- Where does AI genuinely help?
- Which of your institution’s actual workloads — not the industry’s imagined ones — AI meaningfully improves, and which it merely decorates.
- Where doesn’t it, yet?
- The initiatives to decline or defer, named plainly. “Not yet” is a strategy conclusion, not a failure of imagination — and it is frequently the right one.
- What can your record support?
- AI is only as useful as the record underneath it. If the student record is fragmented across systems, that is the first project — whatever the pitch deck said.
- Where should the models run?
- For some institutions the answer is their own infrastructure rather than a third party’s. Self-hosted LLM deployment is something we do, so the recommendation comes with the ability to carry it out.
We sell deployment work too — Agentforce, our own products — which is exactly why the strategy engagement has to be able to conclude against all of it. Advice that always ends in a purchase order is a pitch with an invoice for the pitch.
The prerequisite
The unglamorous first question.
Salesforce has carried machine learning natively since the 2010s; models are the industry’s oldest new thing. What separates institutions where AI works from institutions where it disappoints is almost never the model — it is whether there is one unified student record for the AI to read, or six systems telling six versions of the truth. That is an unglamorous conclusion for a strategy page, which is a reasonable sign it is an honest one.
Insights & events
Related reading.
Article · May 14, 2026
Automated Transfer Credit Evaluation in Minutes: What We Showed at AACRAO Annual
What Tondro demonstrated at AACRAO Annual — extraction, 40+ fraud signals, and equivalency matching in about two minutes — plus the registrar Q&A.
Article · Apr 15, 2026
The AI Visibility Gap in Higher Education: 46% of Students Are Searching With AI. Only 30% of Schools Are Ready to Be Found.
Half of prospective students research colleges with AI. Only 30% of institutions have a strategy for appearing in the answers.
Article · Apr 15, 2026
Your Website Is Optimized for LLMs, But LLMs Aren't Using It
An LLM brand audit extracted over 100 facts about one institution. None of them came from the institution's own website.
Start here
Bring the pitch you’re being given.
The fastest test of whether we are useful: bring the AI initiative currently on your desk — the vendor deck, the board question, the pilot proposal — and we will tell you candidly what we would do in your place. Sometimes that costs a project its momentum. That is the service.