Capstone studio
Capstone: research brief and evidence ledger
Produce a concise research recommendation whose facts, assumptions, and limits can be independently inspected.
Lesson 4 of 6 in the recommended order · About 45 min (estimate)
On this page
Practical AI glossary — terms and common confusions
- Model
- A learned component used to produce a result.
A drafting model is one part of a letter app; it is not the whole app.
- Application
- The software experience around components and services.
A letter app adds accounts, storage, and a Send button.
- Prompt
- Instructions and input supplied for a task.
“Summarize this notice in two bullets” specifies a task and shape, not a truth guarantee.
- Token
- A unit a model processes; it need not be a whole word.
A tokenizer can split a name into pieces; count using the actual system.
- Context
- Information available for the current request.
An earlier attachment may be absent even when its filename is visible.
- Training
- Adjusting a model using examples.
A training log differs from a conversation correction.
- Inference
- Using a trained model to produce a result.
Drafting another sentence does not by itself prove retraining.
- Retrieval
- Selecting material to supply to a task.
A retrieved notice still needs a relevance and date check.
- Embedding
- A numerical representation used to compare or organize items.
A similar search match is a candidate, not proof of the same meaning.
- Tool
- A function an application can call.
A calendar lookup can be read-only; finding an event is different from creating one.
- Agent
- Software that can choose and carry out steps toward a goal.
An agent still needs bounded permissions and checked action receipts.
- Hallucination
- Plausible generated content that is unsupported or false.
An invented opening time needs evidence or removal, even when phrased confidently.
- Evaluation
- Checking results against stated criteria.
Twenty checked examples establish a bounded result, not universal reliability.
- Multimodal
- Working with more than one kind of input or output.
A photo plus text does not prove every small label in the photo was read correctly.
Outcome
Create and revise a source-linked research brief with an evidence ledger and unresolved-question handoff.
Why it matters
Produce a concise research recommendation whose facts, assumptions, and limits can be independently inspected.
Concept
Bring the research method together: define the decision, read the packet, map claims to sources, compare options, and write a bounded recommendation. Keep preference, availability, approval, and attendance distinct. A recommendation can be justified without being certain, provided its conditions and unanswered questions remain visible.
The artifact workspace contains the complete fictional Cedar packet. Build the six-part artifact before revealing case feedback when you want an unaided attempt. The case checks below assess selections only; a facilitator must review your actual brief and reasoning. After fresh transfer, add what would change in your recommendation under the new evidence.
Worked example
For a separate fictional club, 8 of 10 replies favor Saturday, but the venue is only available Sunday. A supported brief reports both facts and asks the organizer to resolve the conflict. It does not silently treat preference as a reservation or claim that the other two respondents favor Saturday. A ledger links each count and availability claim to its source.
Before you check
In the workspace, write a 120–180 word recommendation to Mara about whether to plan a morning session. Cite packet IDs, include the survey denominator and nonresponse, distinguish room availability from approval, and propose the next verification. Then complete the correction/acceptance log and fresh-case revision note.
Practice and fresh transfer
The packets below are fictional and contain the facts needed for these cases. External references are optional background. Judge each response independently: it may be supported, contradicted, or unresolved. Select the passages needed to justify your judgment and write why the distinction matters before revealing feedback.
Cedar research brief workspace
This is the planning snapshot before booking approval. Use only R1–R4 to recommend a next step to Mara. Produce a 120–180 word brief plus source ledger, corrections/acceptance log, uncertainty note, and handoff checklist. Do not claim a booking or send anything. After the fresh cases, record which reasoning would transfer and what would need new evidence.
Write here or use your own document. Text stays in this page until you leave, reload, or change accounts. Download your copy before leaving. No artifact text is sent to a server or automatically graded; account completion records only self-review.
Supplied fictional packet
- r1: Survey: 24 volunteers invited; 18 replied. Of respondents, 12 preferred morning and 6 evening. The 6 nonrespondents have no recorded preference.
- r2: Venue schedule: Hall C is available for a morning session on 14 October; evening is unavailable that day. Availability is not a booking.
- r3: Mara is the coordinator. She must approve the plan before a room is booked or invitations are sent. No approval has been given in this snapshot.
- r4: The task is to recommend a workable next planning step, not to infer attendance. Accessibility requirements and final participant availability have not yet been confirmed.
Purpose, reader, constraints, and the decision or action this packet supports.
For each material claim: source ID, supporting passage or calculation, and scope or date limit.
Write the research brief, decision packet, or workflow handoff requested above.
What did you retain unchanged? What did you repair, why, and using which source?
Unresolved questions, assumptions, consequences, and who should verify them.
Checks completed, remaining approval, reviewer, and next step. Distinguish self-review from independent review.
Review criteria
- Trace every material claim to the supplied source ID or an explicit calculation.
- Retain supported content and explain each correction without inventing missing facts.
- Keep assumptions, uncertainty, and remaining approvals visible.
- Complete the brief, source ledger, output, corrections, uncertainty, and handoff sections.
- After the fresh cases, describe how the changed facts would alter your output.
- A facilitator reviews the actual artifact and unaided explanation before any independent learning claim.
Assess the response
Use Accept when all material claims are supported. Use Revise when a supplied fact or requirement is contradicted. Use Evidence is insufficient when a key fact cannot be established either way. If a response contains both an unknown and a direct contradiction, choose Revise and explain both problems. Conflicting claims with no established authority remain insufficient; a claim does not become a governing fact merely because a source asserts it.
Some responses are fully supported. Others need correction or more evidence. Judge each on its sources; do not edit a correct answer just to change it.
Your written notes stay in this page and disappear when you leave. Only a self-reviewed completion can be saved to your learning account. These practice checks do not establish independent proficiency.
Practice
Case 1
Consider the proposed response for this fictional Cedar work task. Use the supplied packet and stated requirements.
Source packet
- Source 1
- Survey: 24 invited, 18 replied, 12 morning and 6 evening choices. Nonrespondents’ preferences are unrecorded.
- Source 2
- Venue: Morning available, evening unavailable. Neither option is booked.
Response to assess
Morning has support from 12 of 18 respondents and a stated venue opening, but this is not a booking or confirmed attendance count.
Practice
Case 2
Consider the proposed response for this fictional Cedar work task. Use the supplied packet and stated requirements.
Source packet
- Source 1
- Survey: 24 invited, 18 replied, 12 selected morning. Six did not reply.
- Source 2
- Brief: Report only recorded preferences.
Response to assess
Eighteen volunteers selected morning, including all six nonrespondents.
Fresh transfer
Case 3
Consider a fictional transfer briefing for a different club.
Source packet
- Source 1
- Survey: 15 of 20 respondents prefer Saturday.
- Source 2
- Venue confirmation: Only Sunday is available during the required weekend.
Response to assess
Saturday has the respondent preference majority, but the venue constraint requires a further decision before selecting a session.
Fresh transfer
Case 4
Consider a fictional transfer claim about an unmeasured requirement.
Source packet
- Source 1
- Packet: The proposed venue has 20 seats and is available Saturday.
- Source 2
- Review record: No information about step-free access has been supplied.
Response to assess
Step-free entry is confirmed for this venue.
Save your self-review
Completion records that you reviewed the cases. Your explanation and transfer performance need a facilitator to establish independent learning.
Sign in with your learning-center account to save completion.
Review the explanation for every case and review your capstone artifact before saving.
Optional separate visual-evidence ledger example
Use this supporting example after your own capstone attempt if useful. It is not part of the Cedar packet and does not establish anything about Cedar’s rooms, survey, booking or attendance.
Optional evidence-ledger example: a fictional room
View A: whole generated image
Visual description: a table and folding chairs, a blank noticeboard, a window and an open doorway at the right edge. No people are visible.
View B: left-half crop of the same file
Visual description: part of the table, folding chairs, window and blank noticeboard. No doorway or people are visible within this frame.
For this separate image only, draft two ledger entries: one visible feature and one claim about booking that still needs a record. Cite View A or View B for visible details. Does cropping change the room, or the information available in the frame? Use the visual descriptions if preferred.
Check the separate image ledger
A visible-feature entry can cite View A for the doorway. View B omits that edge; it does not establish that the doorway is absent from the room. A booking entry must stay unresolved: neither image supplies an availability, reservation or approval record. Ask for the relevant record rather than treating a picture as a receipt. The whole view provides more visual context, not complete knowledge.
These sample entries apply only to this generated scene. Build your Cedar recommendation and ledger from the assessed packet, not this illustration.
Summary and next step
Apply the checklist to a new task. Preserve supported content, explain any change with evidence, and name what remains unresolved. Saving records self-review, not independently demonstrated proficiency. A facilitator must assess the explanation and fresh transfer for human learning evidence.