The examples offer a grounded look at educational AI: drafting materials, coordinating support meetings and helping staff develop consistent prompting habits. They also invite an important distinction: recognizing a useful application is not the same as independently measuring its results.
Three schools, three approaches to Copilot
Pompano Beach High: everyday staff assistance
According to the district, Pompano Beach High staff use Copilot across academic departments, administration and counseling for lesson drafts, differentiated assignments, family communications and meeting summaries. Staff logged more than 5,300 interactions since April, with AI liaison Colleen Crisan supporting participation through staff training.
That number describes activity, not necessarily effectiveness. Without a definition of “interaction,” a staff headcount or a comparison period, it cannot establish hours saved or improved instruction. The useful question for another school is not simply how many prompts staff submit, but how often the resulting work survives professional review.
Peters Elementary: coordination before instruction
Peters used Copilot to combine teacher schedules, multi-tiered system of supports data and progress-monitoring lists into an RTI meeting schedule prioritizing Tier 3 and third-grade students. Coaches also developed reading plans for grades 2–5 and mathematics plans and focus calendars aligned with B.E.S.T. standards.
The distinction here matters: the district describes scheduling support, not AI deciding which students qualify for intervention. A sensible evaluation would examine whether the draft schedule accommodates required participants, preserves the intended priorities and needs fewer corrections than the previous process.
Indian Trace Elementary: a shared starting point
Indian Trace trained staff serving Pre-K–5 classrooms in Goal–Context–Source–Expectations prompting and privacy-conscious small-group planning. ESE Support Facilitator Katherine Shamoun reported spending less time creating materials from scratch and more time reviewing and adapting them.
Her experience points to a practical way to frame the tool: a first-draft assistant rather than an instructional authority. A shared prompting method is worth evaluating because it gives staff a repeatable process to improve, rather than leaving everyone to improvise at a blinking cursor.
The IT detail worth checking: account and data protection
For school technology administrators, the most consequential missing detail is which Copilot experience the schools used. The district announcement does not identify the edition, licensing arrangement or tenant configuration. Its description of privacy-conscious planning therefore should not be read as independent verification of the schools’ technical safeguards.
Microsoft’s documentation supplies useful context without resolving that local question. Microsoft Support states that Copilot Chat provides enterprise data protection when users sign in with a work or school account, and that prompts and responses are not used to train foundation models. It also says prompts, responses and triggered Bing search queries are logged, with administrative search and audit capabilities available. Not used for model training does not mean not retained.
Microsoft Learn further explains that applicable controls depend on the underlying subscription plan. It also distinguishes Bing web-search queries from prompts and responses: the search service operates separately from Microsoft 365 and has different data-handling terms. Those distinctions are particularly relevant when evaluating a workflow involving student-support information.
For an institution considering similar uses, the practical recommendation is to establish three things before entering sensitive information:
- Approved access: Identify the permitted Copilot service and organizational account.
- Approved inputs: Define which information staff may include and which details must be excluded.
- Human accountability: Assign responsibility for checking schedules, instructional materials and communications before use.
These are evaluation recommendations, not claims about safeguards Broward has implemented.
Recognition is a starting point, not an outcome study
The three examples suggest different evaluation priorities. Broad staff use calls for quality and workload checks; intervention scheduling calls for accuracy and careful data handling; shared training calls for consistency and evidence that staff can apply what they learned.
A useful next step would be to measure time spent including review and corrections, not just the speed of generating a draft. For instructional materials, schools could also assess factual accuracy, suitability for the intended students and alignment with the requested standards.
Broward’s awards identify concrete tasks worth examining. The strongest takeaway is not that Copilot has proved an educational transformation, but that the next questions can now be specific: What improved, how reliably, and under whose supervision?