# Training programme — quote TQ-2026-0417 **From:** Northaven Learning, ul. Krótka 14, 30-001 Kraków, NIP PL7788990011 **To:** Kestrel Systems s.r.o., Vodičkova 12, 110 00 Praha, DIČ CZ24681012 **Issued:** 2026-09-04    **Valid until:** 2026-10-04 ## Programme | Module | Format | Seats | Fee | Revenue | Materials | Delivered | |-------------------|----------|------:|-------:|---------:|----------:|----------:| | Foundations | workshop | 24 | 450.00 | 10800.00 | 864.00 | 11664.00 | | Advanced patterns | workshop | 12 | 650.00 | 7800.00 | 624.00 | 8424.00 | | Coaching | 1:1 | 6 | 900.00 | 5400.00 | 432.00 | 5832.00 | | Assessment | async | 30 | 120.00 | 3600.00 | 288.00 | 3888.00 | The programme seats **72** people across four modules. Teaching revenue comes to **27600.00** PLN, and printed and hosted materials are charged at 8% of that, adding **2208.00** PLN — a delivered value of **29808.00** PLN. The average fee per seat is **530.00** PLN. ## Payment schedule | Stage | Share | Amount | Due | |------------|------:|---------:|------------| | Kickoff | 25% | 7452.00 | 2026-10-15 | | Midpoint | 50% | 14904.00 | 2026-11-30 | | Completion | 25% | 7452.00 | 2027-01-15 | The three instalments account for **29808.00** PLN, payable within 14 days of each date above. --- ## Appendix — what `visimark infer` makes of this Everything above is ordinary Markdown, and none of it is checked. The numbers were typed by a person, and they agree with each other only because that person was careful once. There is no formula anywhere in this file. `visimark infer` is for a document in exactly that state. It reads the tables and the prose, works out which rules reproduce the numbers already there, and prints them: ```console $ visimark infer docs/example-quote-plain.md docs/example-quote-plain.md table at line 10 — 4 rows, 7 columns column rules Revenue = Seats * Fee 4/4 rows Materials = Revenue * 0.08 4/4 rows Delivered = Revenue + Materials 4/4 rows constants worth naming 0.08 also appears as "8%" in prose, line 18 scalars matching figures in prose 72 line 17 = SUM(Seats) seats_total 27600.00 line 18 = SUM(Revenue) revenue_total 2208.00 line 19 = SUM(Materials) materials_total 29808.00 line 19 = SUM(Delivered) delivered_total 530.00 line 20 = AVG(Fee) fee_avg no rule found — treating as inputs Module, Format, Seats, Fee also fits, not proposed Delivered = Revenue * 1.08 prefers a rule over materialised columns Delivered = Materials * 13.5 prefers a rule over materialised columns docs/example-quote-plain.md table at line 24 — 3 rows, 4 columns column rules Amount = Share * unnamed1.delivered_total 3/3 rows scalars matching figures in prose 29808.00 line 30 = SUM(Amount) amount_total no rule found — treating as inputs Stage, Share, Due also fits, not proposed Amount = Share * 29808 prefers a rule over materialised columns 4 rules, 0 aliases, 6 scalars, 6 anchors. ``` **Rules are found by exact verification, not by fitting.** A candidate either reproduces every cell of its column, at that column's own precision, or it is not a candidate. `Revenue = Seats * Fee` reproduces all four rows. `Materials = Revenue * 0.08` has its constant solved from the first row and then checked against every remaining one — a constant that satisfies only the row it came from is not a finding. Nothing here is a score or a threshold. **The result is a set, not a list of findings.** `Delivered = Revenue + Materials` fits, and so does `Materials = Delivered - Revenue`; taking both would emit a document that fails `check` with a `CYCLE`. Selection rejects any rule that closes a cycle with the ones already accepted, which is why `Seats` and `Fee` come out as inputs rather than as `Revenue / Fee` and `Revenue / Seats`. Those fit too. They are the same arithmetic read backwards. **Constants are detected and never named.** The report says that `0.08` also appears as `8%` in the sentence under the table, and stops there. Matching the two forms is reliable; concluding that the constant is therefore *called* `materials_rate` is a guess about meaning, and this tool does not guess. **Scalars are named from the column and the reduce, mechanically.** `SUM(Revenue)` becomes `revenue_total` and `AVG(Fee)` becomes `fee_avg`. Reading the surrounding sentence for a nicer name — "teaching revenue" becoming `teaching_revenue` — reads better and is another guess. The boring scheme is predictable, and renaming is one keystroke. **A figure in prose is what makes a scalar worth proposing.** The reduces that match nothing a human wrote down are discarded: `MAX(Fee)` and `COUNT(Seats)` are computable, but no sentence in this quote claims them, so no binding is invented for them. A scalar is anchored at the first figure that follows its own table, and only once — which is why the second sentence stating the delivered value, under the schedule, binds `amount_total` rather than repeating `delivered_total`. **"Also fits, not proposed" is not padding.** `Delivered = Revenue * 1.08` and `Delivered = Materials * 13.5` each reproduce the column exactly. They lose to `Delivered = Revenue + Materials` because a rule over materialised columns beats one that introduces a constant: every intermediate stays a number the reviewer can see. Listing the losers is what makes the proposal reviewable. **The tables have no names, so the tool mints them.** `#unnamed1` and `#unnamed2` are placeholders, and they read as placeholders, which is the point — a generated `#programme` would read as a name somebody chose. The id binds once, at write time, and renaming it afterwards is an ordinary edit. ### What `--write` inserts `visimark infer docs/example-quote-plain.md --write` inserts a block after each table and an anchor after each matched figure. It never rewrites an existing byte, so the prose, the headings and every input column survive untouched: ````markdown ```vmark #unnamed1 Revenue = Seats * Fee Materials = Revenue * 0.08 Delivered = Revenue + Materials seats_total = SUM(Seats) revenue_total = SUM(Revenue) materials_total = SUM(Materials) delivered_total = SUM(Delivered) fee_avg = AVG(Fee) ``` ```` The sentence under the table keeps its words and gains invisible comments: ```markdown The programme seats **72** people across four modules. ``` `visimark check` fails on the document as committed, and the finding is the point: `COVERAGE — a table with no vmark rules`, because a green check on a document with no rules would be a question, not an answer. Run it against the document `--write` leaves behind and `0 problems` is a claim. Change one `Seats` figure there and it stops being true — in both tables at once, because the schedule reads the delivered total. The other way to answer that finding is ``, and this document is exactly the case where it would be a lie: `infer` can see nine rules in here. The marker is for a document with tables and no arithmetic, and `infer --write` is the only thing that writes it — never over a near-miss. ### The half that matters more A rule that fits every row but one is not a failed inference. It is the tool saying the document has a wrong number in it, to someone who has adopted nothing and learned no syntax. Transpose two digits in the `Delivered` column and ask again: ```console $ sed -i 's/5832.00 |/5823.00 |/' quote.md $ visimark infer quote.md near-miss — not proposed Delivered = Revenue + Materials 3/4 rows row 3 Coaching cell 5823.00, rule gives 5832.00 differs by 9.00 ``` That is an excerpt: `Delivered` is now the only column without a rule, and the sentence stating the delivered value no longer matches any total, so that anchor is not proposed either. Every other rule in the document still holds. A near-miss is never proposed and never written, even under `--write`. It is reported, with the row named and the difference given, and what to do about it is the reader's decision.