The Educational Technology Framework: Assessing Your 2026 Roadmap

I wrote the first version of this framework in 2010 because I kept seeing the same problem I’d seen throughout my career in edtech: tools were getting better, institutions were not. The technology moved, the GOVERNANCE didn’t. Back then, course management systems became learning platforms while institutions chased releases and faculty expectations. The framework gave leaders a map to discuss MATURITY without mistaking a feature list for STRATEGY.

I rebuilt it in 2026 because the gap has widened. Agentic AI has pushed the center of gravity of learning from content delivery to decision support, and from “what tool do we allow?” to “what OUTCOMES do we accept accountability for?” The campus is already a live lab; students and faculty are using ChatGPT or something like it, policy or not. The question isn’t whether AI is on campus; it’s whether the institution is GOVERNING it or pretending that individual discretion scales.

Look, I’ve been in this game for over three decades. I’ve shipped software, sat through board meetings, and seen pilots become accidental production. The 2010 framework did its job. But we’re past “can it scale?” and into “should it decide?” That’s why the EDUCATIONAL TECHNOLOGY FRAMEWORK: ASSESSING YOUR 2026 ROADMAP is a rebuild, not a reprint. It remains a maturity model for presidents, provosts, CIOs, and faculty leaders—and it is blunt on one point: TECHNOLOGY decisions are GOVERNANCE decisions.

Why the 2010 Framework Needed a 2026 Rebuild

The 2010 edition put structure around a landscape mostly about consolidating systems and bringing the campus online. We had learning management, SIS integrations, identity, and early analytics. The questions were: can we integrate? can we scale? can we support faculty? It worked because the problems were bounded by the campus and the course. GOVERNANCE was largely about acceptable use and procurement.

In 2026, the boundaries are gone. AGENTIC SYSTEMS don’t just recommend; they act—draft feedback, grade, scaffold, match interventions, and run workflows. Data no longer sits neatly within a platform; it flows across systems and out to model providers. The unit of value has shifted from the tool to the LEARNING ENVIRONMENT, with AI in the loop. Our old proxies for maturity—tool adoption, help desk tickets, uptime—are necessary but insufficient. The risk surface now includes MODEL RISK, DATA LINEAGE, and EXPLAINABILITY. If that sounds like banking or healthcare, that’s the point.

The other reason for the rebuild is cultural. A decade ago, you could live with an ambiguous posture about pedagogy because platforms shipped limited affordances. Today, your posture—or your absence of one—shows up in the AI your students and faculty encounter. If you don’t decide what the environment should do, vendors and defaults will decide for you. Plainly, as someone who has built and sold these systems: a VENDOR ROADMAP is not a substitute for institutional STRATEGY. It’s input, not GOVERNANCE.

So I rebuilt the framework to center three truths. First, MATURITY IS A FUNCTION OF GOVERNANCE COHERENCE, not tool count. Second, FACULTY PRACTICE is the control surface that determines whether your stack translates into learning, not your contract terms. Third, AI GOVERNANCE must sit beside cybersecurity and privacy on the same RISK REGISTER and report to the same bodies. The model below is still practical—five phases, five dimensions, a 90-minute diagnostic, and a 12-month roadmap—but it’s anchored in this era: agentic AI, outcome accountability, and board-level visibility.

The Five Phases, Honestly Named

01 Exploratory

What it looks like: Curiosity outpaces coordination. Individual faculty experiment with AI and students compare notes on prompt craft. There’s no shared posture. Tools are used widely with no institutional position beyond a sentence in the academic integrity policy. SUPPORT is ad hoc, DATA flows where it flows, and PEDAGOGY is local. If a dean asks, “Who’s accountable for our AI exposure?” you’ll get five different answers—or none.

2026 reality: Don’t moralize here. Curiosity isn’t the problem; unmanaged RISK is. You see the widest range of practice and outcomes. Some students thrive on AI-augmented feedback; others are confused about what’s allowed. Some faculty run remarkable experiments; others warn without training. The risks include DATA SPRAWL, accidental model training on sensitive content, and unvetted dependencies in credit-bearing work.

Next move: name a small senior group accountable for AI USE, DATA SHARING, and PEDAGOGICAL BASELINE.

02 Supported

What it looks like: IT supports specific tools, but PEDAGOGY is still personal. You’ve got HELP DESKS, SLAs, maybe an AI ACCEPTABLE-USE POLICY borrowed from a peer. There’s a TOOL LIST, a procurement path, and some training sessions. Departments can point to “approved” versus “not approved,” and there’s at least one AI add-on in the LMS people know by name. The public stance is “we’re on it,” while day-to-day practice is “do what works for you.”

2026 reality: Supported can feel comfortable—and that’s the trap. You’ve contained some risks and bought time, but you’re accumulating hidden complexity. Every “approved” tool brings its own data handling, model update cadence, and quirks. Faculty still improvise because the platform posture isn’t explicit about what the environment must consistently deliver with AI in the loop. You can answer “what can I use?” but not “what must this ENVIRONMENT do for learners across the curriculum?”

Next move: move from a TOOL LIST to a LEARNING-EXPERIENCE STANDARD; decide which AI capabilities the institution will operate, permit, restrict, and forbid — and write it down.

03 Strategic

What it looks like: A coherent stack with deliberate GOVERNANCE. The institution acts like one organization instead of a collection of pilots. FACULTY GOVERNANCE PARTICIPATES. MODEL RISK is reviewed alongside CYBERSECURITY and PRIVACY. Vendors disclose TRAINING DATA, RETENTION, and HUMAN-IN-THE-LOOP posture. Faculty are TRAINED, NOT WARNED. You have a minimum LEARNING-EXPERIENCE STANDARD with AI embedded where appropriate, and you’ve written down what the environment must do for learners, instructors, and advisors.

2026 reality: STRATEGIC is not paperwork; it’s behavior change. It shows up in the calendar—recurring cross-functional reviews, structured faculty development, and a shared RISK REGISTER. It shows up in contracts—clauses tying renewal to model transparency and outcomes. It shows up in decommissioning—tools that don’t fit the STRATEGY are retired. Most importantly, it shows up in the language leaders use: from “What’s on the vendor roadmap?” to “Which parts of our LEARNING EXPERIENCE are we accountable for improving, and what will we publish?”

STRATEGIC is the phase where the institution stops being an audience for VENDOR ROADMAPS — and starts being the author of its own.

Next move: tie the stack to a small number of OUTCOMES the institution is willing to publish and begin retiring tools that don't serve them.

04 Mission Critical

What it looks like: Learning depends on the platform; the board treats it accordingly. When the platform is down, CREDIT-BEARING WORK is directly affected. That reality is reflected in RISK REGISTERS, board packets, and budget planning. Procurement, security, and academic affairs SHARE THE SAME RISK REGISTER and reconcile it quarterly. The institution understands the AI/edtech estate is INFRASTRUCTURE, not APPS. Resilience, continuity, and explainability are part of OPERATIONS, not special projects.

2026 reality: You’re managing DEPENDENCIES across learning, advising, assessment, and the agents threaded through them. A change in a model endpoint, a vendor retraining event, or a new regulation can ripple through pedagogy and student support. The MISSION-CRITICAL institution anticipates these ripples. It has RUNBOOKS, not memos; PORTFOLIO DASHBOARDS, not tool lists; and contract structures that allow rebalancing without chaos.

Next move: move from procurement to PORTFOLIO — treat the AI/edtech estate the way the endowment is treated: diversified, rebalanced, and reported on.

05 Transformative

What it looks like: AGENTIC SYSTEMS RESHAPE PEDAGOGY AND UNIT ECONOMICS. The environment moves from static content and scheduled interactions to adaptive, co-creative learning. Advising, tutoring, and assessment are increasingly AI-mediated with HUMAN-IN-THE-LOOP guardrails. The institution experiments with OUTCOME CONTRACTS. Faculty roles shift toward orchestrating learning experiences and validating evidence, and students encounter a consistent, explainable AI presence across their journey.

2026 reality: This isn’t science fiction; it’s careful design. Winning institutions don’t promise magic; they build trust. They publish outcome definitions, measure against them, and make transparent how AI participates in decisions. They mature their ASSURANCE: bias testing, red-teaming, and third-party audits. Economics improve not from hype but because FRICTION drops in the right places and faculty time shifts to the highest-value interactions.

Next move: compete on the TRUSTWORTHINESS of the learning environment — the only durable moat once production cost collapses.

The 90-Minute Diagnostic

The whitepaper includes a structured, 90-MINUTE DIAGNOSTIC that scores your institution across five dimensions. You don’t need weeks of interviews to get signal. You need the right people in the room—provost’s office, CIO, CISO, faculty governance, institutional research, and a registrar who knows where the data really lives. Work through five dimensions, ask specific questions, and agree on a score for each. Here’s the core of it:

• Pedagogy — how is teaching designed around the platform and AI? Are there documented course design patterns that assume AI is in the loop? Do you have a baseline for feedback cadence, formative assessment, and explainability?
• Platform — how coherent and integrated is the stack? Can you describe it as an intentional ARCHITECTURE with clear data flows, or is it a patchwork? Do tools interoperate against a published LEARNING-EXPERIENCE STANDARD?
• AI Governance — who is accountable for MODEL RISK, DATA LINEAGE, and EXPLAINABILITY? Is there a named body that owns these decisions? Are model updates tracked, tested, and reported like security patches? Do vendors disclose training data, retention, and human-in-the-loop posture?
• Faculty Practice — are faculty trained operators of the stack, or reluctant users? Is professional development continuous and tied to platform capabilities? Are incentives and evaluation aligned with desired practices, or do they stop at warnings?
• Contract Structure — are contracts seat-based, outcome-based, or absent? Do agreements align cost with value and include transparency clauses for AI? Can you rebalance the portfolio without burning relationships?

The diagnostic forces a tough, productive conversation and enforces a CRITICAL RULE: the LOWEST dimension score sets the institution’s true phase, not the highest. If your PLATFORM is strong but your AI GOVERNANCE is fragile, you are not Mission Critical—you are whatever your GOVERNANCE allows you to be. I've seen this pattern for decades: a shiny stack masking a brittle institution. In 2026, that pattern—high PLATFORM score with low GOVERNANCE—is the MOST COMMON and the MOST DANGEROUS.

Why lowest-score-sets-the-phase? Because RISK concentrates at the weakest link. If FACULTY PRACTICE lags, your outcomes will lag no matter how clean your architecture diagram looks. If CONTRACTS are seat-based with no visibility into model updates, your governance promises are paper-thin. If PEDAGOGY is still personal and improvisational, your LEARNING-EXPERIENCE STANDARD is a poster, not a practice. The diagnostic prevents self-deception by forcing institutional humility. It is quick on purpose. Revisit it quarterly, adjust scores as you build muscle, and show the board a simple, stable reference over time.

A note on scoring behavior: don’t overcomplicate it. The goal of ninety minutes is ALIGNMENT, not perfect measurement. Document dissenting views, decide on a score, and name the owner who will move that dimension up one notch next quarter. That act—naming an owner and a deliverable—is where culture starts to change. The boring work is the work that survives the cycle.

The 2026 Roadmap: Twelve Months, Four Moves

A roadmap that just lists activities is a calendar, not a plan. The 2026 roadmap compresses the first year into FOUR MOVES, each quarter producing a BOARD-READABLE ARTIFACT with a named owner and an auditable deliverable. Complete these four moves and you’ll make substantive progress no matter your starting phase. Skip the artifacts and you’ll run in place.

Move 1 (Quarter 1): Establish the Baseline and the Owners. Convene the cross-functional group. Run the 90-minute diagnostic. Publish a 6–8 page “Institutional AI and Learning Baseline” with your phase rating (set by the lowest score), the five dimension scores, current AI RISK REGISTER entries, and the names and roles of accountable owners for each dimension. Owner: Provost and CIO as co-sponsors. Deliverable: a document entered into the board packet and archived in your policy repository. Audit hook: link each score to evidence (policies, contracts, training schedules, platform diagrams).

Move 2 (Quarter 2): From Tool List to Learning-Experience Standard. Draft and ratify a LEARNING-EXPERIENCE STANDARD that specifies what every learner should reliably get with AI in the loop: feedback cadence, explainability requirements, accommodations pathways, data-sharing boundaries, and instructor obligations. In the same quarter, decide which AI CAPABILITIES you will OPERATE, PERMIT, RESTRICT, and FORBID—and write it down. Owner: Faculty governance chair and the academic technology lead. Deliverable: a published standard with adoption guidance, example course shells, and a change log. Audit hook: evidence of faculty consultation, version control, and a publish date.

Move 3 (Quarter 3): Governance and Contracts to Match the Standard. Translate the standard into PORTFOLIO MANAGEMENT. Update procurement language to require vendor disclosures on training data, data retention, explainability features, and HUMAN-IN-THE-LOOP controls. Build or update your AI RISK REGISTER and formalize MODEL UPDATE RUNBOOKS (how you test, accept, or roll back). Adjust contracts to include TRANSPARENCY CLAUSES and options for OUTCOME-ALIGNED PILOTS. Owner: CISO for risk and the head of procurement for contracts, with legal and IRB advisors. Deliverable: a “Governance and Contract Addendum” packet with templated clauses and a vendor disclosure appendix. Audit hook: signed amendments, a versioned RISK REGISTER, and evidence of model update testing.

Move 4 (Quarter 4): Portfolio Report and Retirement Plan. Treat your AI/edtech estate like the ENDOWMENT: DIVERSIFIED, REBALANCED, and REPORTED ON. Produce an annual “Learning Environment Portfolio Report” that shows where dollars, data, and risk sit; which tools are core versus experimental; what outcomes you targeted (from Move 2) and how you performed; and which tools will be RETIRED. Owner: CIO for the stack, Provost for outcomes, CFO for financial framing. Deliverable: a report to the board and faculty senate with a clear retirement schedule and transition plan for affected courses. Audit hook: portfolio inventory, cost and usage analytics, outcome measures, and decommissioning runbooks.

Throughout the year, invest in FACULTY TRAINING about operating the stack, not just ethics briefings. Strategic institutions train, not warn. Tie TRAINING to the LEARNING-EXPERIENCE STANDARD and measure participation. Publish coverage statistics next to technical readiness. In the early days of the LMS, the most successful campuses made TRAINING part of the job, not an optional workshop. 2026 doesn’t change that; it amplifies it.

A caution on scope: don’t overhaul everything in twelve months. Aim for CREDIBLE BASELINES and VISIBLE ARTIFACTS. Retiring one or two tools that don’t serve the published outcomes changes culture more than a dozen pilots. Publishing a clear explanation of your AI POSTURE reduces confusion more than another all-campus webinar. This is the work that compounds.

What This Framework Refuses To Do

First, it refuses to treat VENDOR ROADMAPS as institutional STRATEGY. I say this as someone who has built and sold platforms for decades: they are useful inputs, not GOVERNANCE. If your plan is a collage of vendor slides, you don’t have a plan. You have aspirations arranged by a calendar.

Second, it refuses the fantasy that DASHBOARDS ARE GOVERNANCE. Pretty charts don’t substitute for named accountabilities, tested runbooks, or audited contracts. If your AI exposure lives only in a dashboard and not in the RISK REGISTER the board reads, you are performing, not governing.

Third, it refuses to let endless PILOTS stand in for decisions. Pilots are for learning, not for avoiding hard calls. If a pilot runs past two semesters without a decision, it’s not a pilot anymore—it’s production without consent. Strategic institutions time-box experiments and then decide to scale or retire.

Fourth, it refuses the comforting idea that POLICIES BORROWED from a peer are the same as POLICIES OWNED by your institution. A pasted acceptable-use policy without local accountability is a press release. Write down what you will OPERATE, PERMIT, RESTRICT, and FORBID, and then live with the consequences.

Fifth, it refuses to treat FACULTY DEVELOPMENT as optional. We tried optional in the LMS era. It produced islands of excellence and oceans of inconsistency. In the AI era, that inconsistency has real stakes. If students can’t predict how the environment will behave from course to course, we’ve failed at the most basic form of equity—predictability.

Finally, it refuses to confuse FEATURE NOVELTY with institutional MATURITY. The question isn’t whether your LMS has a new assistant or your advising system can summarize notes. The question is whether those assistants and summaries operate within your LEARNING-EXPERIENCE STANDARD, your GOVERNANCE POSTURE, and your CONTRACTUAL ASSURANCES. If they do, you’re building. If they don’t, you’re improvising.

The Map Is the Work

Since the beginning of the platform era, the institutions that did best weren’t the ones with the biggest budgets or the flashiest deployments. They treated the platform as MISSION INFRASTRUCTURE and built GOVERNANCE and FACULTY PRACTICE around it. The same pattern is repeating in 2026. The stack is more powerful, the stakes are higher, and the institutions that win align around a shared MAP—and then do the boring work.

Most institutions don’t fail because they bought the wrong tool. They fail because they bought it before they had a shared map of where they were. Used well, the model becomes the standing reference for procurement, audit, and academic governance—a single artifact the board, the provost, and the CIO are all reading from. That alignment, not the platform itself, is what compounds.

If you take nothing else from this whitepaper and this post, take this: TECHNOLOGY DECISIONS ARE GOVERNANCE DECISIONS. Name the owners. Write down the posture. Publish the OUTCOMES you’re willing to be judged against. Retire what doesn’t serve them. Train your faculty like the operators they are. Move from procurement to portfolio. Compete on trust. I’ve seen cycles come and go for three decades, and I can tell you with certainty: the boring work is the work that survives the cycle, and this is the work that compounds.

Look, the 2010 framework helped a lot of campuses move from systems thinking to platform thinking. The 2026 rebuild asks you to move from platform thinking to INSTITUTIONAL AUTHORSHIP. That’s the shift—from being an audience for someone else’s roadmap to being the author of your own. If we do that, the next decade of teaching and learning won’t just be more automated; it will be more humane, more explainable, and more accountable. And that, as much as any feature, is the future worth building.