End-of-year reporting in Sweden is not just “writing nice comments”. Under LGR22, you are expected to communicate what the pupil can do in relation to the curriculum, what they need next, and—when grades apply—how the grade decision is grounded in evidence. In practice, the strongest skriftliga omdömen read like a short, defensible argument: a claim about current attainment, one or two concrete examples, and a next step that matches teaching.
The tricky part is defining “evidence” in a way that stands up to questions later. Evidence is not only tests. It is also annotated work, exit tickets, observed talk in group work, practical outcomes, lab notes, and patterns seen across lessons. If you can point to what was seen, when, and in which task, you can justify your judgement without inflating the workload. If your team is also trying to standardise how that evidence becomes publish-ready text, it helps to treat reporting as a pipeline with checkpoints, rather than a heroic solo writing session. For a broader view of building a safe, auditable comment process, see report-writing pipelines with audit trails.
Most teachers feel the burden most sharply at the writing stage, but the real time sink is usually scattered evidence collection and “finding the story” afterwards. A simple model helps you see where AI can help without pretending it can do your professional judgement.
Imagine 26 pupils. If you spend just 6 minutes per pupil hunting for evidence across books, platforms, and memory, that is 156 minutes (2 hours 36 minutes). If you then spend 10 minutes drafting, revising tone, and formatting, that is 260 minutes (4 hours 20 minutes). Add 3 minutes per pupil for double-checking names, details, and any support arrangements, and you add another 78 minutes (1 hour 18 minutes). You are already at about 8 hours for one class, before moderation.
The goal of an AI-assisted workflow is not to “mass generate” comments. It is to shift time from retyping and rephrasing into higher-value checking: choosing the right evidence, ensuring alignment to LGR22 language, and moderating for fairness. If you want a moderation-led approach that avoids generic bulk-reporting habits, you can borrow the checkpoint logic from this moderation-first reporting pipeline and apply it to Swedish omdömen and grades.
This pipeline uses four tools in sequence, each with a narrow job. The output of one tool becomes the input of the next, and you keep a lightweight audit trail as you go.
You start with Development Talk (Student) to capture structured evidence in LGR22-friendly language while it is fresh. The Summariser then condenses that evidence into “claim + example + next step” blocks, preserving references to the source notes. Student Communication turns those blocks into age-appropriate, pupil-facing comments that build agency. Parent Communication finally produces a guardian cover letter that sets expectations, explains the judgement, and—where needed—supports translation without over-promising.
Human sign-off is non-negotiable. AI drafts; teachers decide. If your school is refreshing its rules for this kind of workflow, it is worth aligning it with an annual policy update such as an AI acceptable use refresh checklist.
The Development Talk (Student) step is an evidence-capture routine, not a counselling script. You are aiming to capture specific “I can…” statements, plus a task reference, plus one next step. Done well, it prevents the end-of-year scramble because you are collecting usable evidence in a consistent format.
A practical script for Åk 2–Åk 9 can sound like this in the classroom or in a short conference:
You begin with a focus area (“Let’s look at how you explain your thinking in maths”). You ask for a recent example (“Which task this month shows that best?”). You probe for process (“What did you do first, and what did you do when it got hard?”). You capture one concrete teacher observation (“In the fraction card sort on 14 May, you compared 1/2 and 2/4 correctly and explained why”). You end with a next step that can be taught (“Next, we’ll practise writing one sentence that justifies your method”).
Write the notes in a consistent structure: date, subject area, task, observation, pupil voice, next step. That structure is what makes the next step (Summariser) reliable.
The Summariser turns raw notes into short blocks that are easy to moderate and easy to reuse across student- and parent-facing versions. The key is to force the output into a defensible pattern: claim, example, next step. You also keep an audit trail by including the evidence reference (date/task) inside the draft, even if you later remove it from the published comment.
A prompt pattern that works well is: “Using only the evidence below, write two ‘claim + example + next step’ blocks. Keep Swedish terms such as skriftligt omdöme, and do not invent results.” You then paste your structured notes.
Your checkpoint here is simple: if you cannot point to the note that justifies a sentence, you delete or rewrite it. This is also the moment to watch for tone drift (too harsh, too vague, or too glowing). If you want a deeper comparison of assistant behaviours and the practicalities of traceability, revisit AI assistants compared for report writing.
Ready to Revolutionise Your Teaching Experience?
Discover the power of Automated Education by joining out community of educators who are reclaiming their time whilst enriching their classrooms. With our intuitive platform, you can automate administrative tasks, personalise student learning, and engage with your class like never before.
Don’t let administrative tasks overshadow your passion for teaching. Sign up today and transform your educational environment with Automated Education.
Student Communication is where you deliberately change the voice. A defensible omdöme can still be discouraging if it reads like a judgement “about” the pupil rather than guidance “for” them. For younger pupils, keep sentences short and concrete. For older pupils, you can name strategies and invite self-regulation.
A useful rule is to preserve the claim and example, but rewrite the next step as an action the pupil can take. “Needs to improve reasoning” becomes “Next, add one sentence that explains why your method works.” You also remove adult-coded language that can feel final (“cannot”, “fails to”) and replace it with growth language anchored to evidence (“not yet consistent”, “needs more practice with…”). If you are supporting pupils with additional needs, this is also where you avoid deficit phrasing and stick to observable learning behaviours.
Parent Communication is a cover letter and clarity layer. It should help guardians understand what the school has seen, what the next steps are, and how they can support at home without turning the report into a negotiation. It is also where translation may matter. AI can help produce a plain-language Swedish version first, then a translation draft for the home language—provided you label it clearly as a draft and you do not promise perfect accuracy.
Be careful with expectations. Avoid language that implies guarantees (“will reach grade C next term”) and stick to what the school will do and what the pupil will practise. If you run parent meetings alongside reporting, you can adapt a short briefing workflow like this one-page parent consultation conversation brief so the message stays consistent across written and spoken communication.
Quality gates are what make this pipeline defensible. Moderation comes first: sample a handful of pupils across attainment levels, compare the “claim + example + next step” blocks, and check that similar evidence leads to similar language. Then run a bias and tone check: look for patterns such as different adjectives used for different groups, or different levels of certainty without different evidence.
SEND-sensitive language matters in Swedish contexts too: describe supports and strategies without labelling the pupil. “Benefits from sentence starters” is safer than “struggles to write”. Finally, enforce “never paste” rules. Do not paste personal data into tools that are not approved. Do not include sensitive health details. Keep identifiers minimal, and store the evidence references in your school system, not inside the AI output. If your team needs a shared training routine to embed these safety habits quickly, an INSET micro-routines implementation plan can help.
In Åk 2 you are writing skriftliga omdömen without grades, so specificity and kindness matter. Below is an example that stays concrete and actionable.
Skriftligt omdöme (Åk 2, svenska):
Under terminen har du blivit säkrare på att läsa korta texter med flyt. I högläsningen den 22 maj kunde du läsa en hel sida utan att stanna, och du rättade dig själv när du såg att ett ord inte passade i meningen. Nästa steg är att träna på att berätta med egna ord vad texten handlade om. Vi kommer att öva med “tre frågor efter läsning”, och du kan hemma prova att säga en mening om början, mitten och slutet.
Åk 4 maths omdömen work best when they include a strategy, not only a topic. This example uses evidence-led phrasing and clear classroom actions.
Skriftligt omdöme (Åk 4, matematik):
Du visar god förståelse för de fyra räknesätten och kan välja en fungerande metod i rutinuppgifter. I problemlösningen “Skolresan” (8 april) ritade du en tabell för att hålla ordning på informationen och kom fram till rätt svar. När uppgifterna har flera steg blir det ibland svårt att visa hela resonemanget, vilket gör att små fel kan smyga sig in. Nästa steg är att skriva en kort plan innan du räknar och att kontrollera svaret med en annan metod, till exempel uppskattning eller omvänd räkning. Vi tränar detta med “plan–räkna–kolla” i varje problemlösningslektion.
Åk 6 is the first point where grades become central, so alignment between the narrative comment and the grade decision must be transparent. The narrative should not contradict the grade, and it should make the next step believable.
Exempel (Åk 6, engelska – betyg och omdöme):
Betygsbeslut: C.
Motivering i omdömet: Du kan förstå huvuddragen i talad och skriven engelska och du deltar aktivt i samtal. I diskussionen om “School rules” (15 maj) ställde du följdfrågor och utvecklade dina svar med exempel. I skriftliga texter är strukturen oftast tydlig, men språklig variation och korrekthet är ännu inte helt jämn, vilket syns i din berättande text från 6 juni. Nästa steg är att arbeta med att variera meningar och att korrekturläsa med en enkel checklista (verb i dåtid, stor bokstav, punkt). Detta gör att dina texter blir mer precisa och lättare att följa.
In Åk 8 chemistry, precision matters. You can and should reference practical lab evidence, misconceptions, and the quality of explanations.
Skriftligt omdöme (Åk 8, kemi):
Du kan beskriva flera centrala begrepp inom kemin och använder dem ofta korrekt när vi arbetar med reaktioner. I laborationen om syror och baser (12 mars) hanterade du utrustningen säkert, dokumenterade pH-värden noggrant och kunde koppla resultatet till indikatorns färgförändring. När vi diskuterar partikelmodellen blir förklaringen ibland för generell, till exempel när du blandar ihop “ämne” och “blandning” i dina resonemang om lösningar (anteckningar 28 april). Nästa steg är att träna på att skriva en förklaring i tre led: vad vi ser, vad som händer på partikelnivå, och varför det ger just det resultatet. Vi kommer att använda exempelmeningar och begreppskort för att göra resonemangen mer exakta.
To make this workable for a team, prepare templates before the rush: one evidence-capture note format, one Summariser prompt that forces “claim + example + next step”, and two output styles (pupil-facing and guardian-facing). Time-boxing matters more than perfection. Many teams find that 12–15 minutes per pupil is achievable once the evidence notes are consistent: a few minutes to review notes, a few to generate and edit blocks, and a few to run quality gates and final checks.
A minimum viable rollout is to pilot one year group first, moderate together using a small sample, and then scale. If you are planning a broader term-by-term adoption, you may also want to map it into a short foundation sprint such as a summer-term AI foundations plan so the routines are in place before next year’s reporting cycle.
Families deserve clarity. A short statement can sit at the end of your cover letter: the teacher collected the evidence, the teacher decided the judgement and any grade, and AI was used only to draft wording from teacher-provided notes. Add that the text was reviewed for accuracy, tone, and alignment with LGR22, and that families can ask to discuss the evidence behind the comment. This simple transparency reduces anxiety and reinforces that professional responsibility stayed with the school.
May your skriftliga omdömen feel lighter to write—and stronger to stand behind.
The Automated Education Team
Black Friday can feel like a rare chance to “save” on AI subscriptions, but …
Three years after ChatGPT’s release, schools have enough experience to …
December in schools brings a familiar spike: cover changes, heightened …
Microsoft Ignite can feel like a firehose of AI updates, but schools need a …
Report writing in 2025 is less about “which chatbot is best” and more about …
Three years into LGR22, many schools report real gains in clarity and …
Anti-Bullying Week works best when it moves beyond awareness and into …
Remembrance teaching asks for careful language, accurate sources, and …
Mock season often fails for predictable reasons: revision plans are …
Aasta lõpu aruandlus LGR22 järgi võib tunduda nagu sprint: pead muutma kuude kaupa kogunenud …
Årsavslutande omdömen enligt LGR22 kan kännas som en spurt: du ska omvandla månader av vardagliga …
Loppuvuoden raportointi LGR22:n puitteissa voi tuntua sprintiltä: kuukausien arjen näytöt pitää …