What AlmondTill is — notes
One engine for a retail business — the catalog, the stock, the till, the customers, the purchasing and the books behind them — on one open API, for one monthly price, with every figure on these slides read from the system itself.
2. One engine, not a patchwork
You have outgrown your retail system. AlmondTill is one engine for everything a shop does: the catalog, the stock, the register, the customers, the promotions, the purchasing and the books behind them — built as one open API, with the back office, the browser register, the command line and your AI assistant all reading the same records.
Nothing is screen-only. Nothing is sold back to you one subscription at a time.
Notes
Say the one idea first: one engine, every area, one open API. Every screen — ours or the merchant's own — is a consumer of it. The rest of the deck is that sentence in detail.
3. Who it is for
Retailers who are replacing an existing point of sale or retail system: fashion and apparel, health and wellness, home and living, hardware and garden, electronics and appliances, hobbies and leisure, vehicles and parts, and general and variety retail — one till, or a team across a hundred stores.
Not restaurants: there are no tables, courses or kitchen screens here, and none are planned.
Notes
The trades are the ones the public site names today. Say plainly what is not there: restaurant features are not a target.
4. Every area, live today
- 132 record types live on the engine — catalog, stock, selling, customers, purchasing, staff, the books
- 1,085 operations on one open API: 395 reads and 690 writes
- 24 ready-made reports, and ad hoc reports drafted from plain words
- every operation reachable from the command line and from any AI assistant that speaks MCP — the same operations, not a subset
Notes
Every number on this slide is counted from the registries when the page is rendered; the notes page lists each one with its source. The what's-live page on the site carries the matrix, area by area.
5. The surfaces
The back office in the browser. The register as a mode of the back office, on any modern browser device you already own. The command line for every operation. The open API for your own tools. AI assistants over MCP — 17 tools, read-first, with only the permissions you assign.
See what is live today on the site for the matrix, area by area.
Notes
The register is a browser; the native phone and tablet apps are coming — the slide near the end says so. The MCP connector is live and read-first today.
6. The price
$15 a month for your whole business, whether you run one register or a thousand stores — and $10 of it comes straight back to you every month as 100 million kernels, the till's prepaid metering unit.
No transaction fees. No percentage of your sales. No per-user, per-register or per-location charges. Whatever your payment processor charges is between you and them.
Notes
The words are the pricing page's own. The fee and the included kernels are read from the rate card at the render, never typed.
7. What a month looks like
A quiet month barely moves the meter: the month's included kernels cover it and the bill stays flat. A busy month pays for the work it had. Four worked examples, from a single register to a chain, stand on the pricing page — priced the way the meter prices them.
Notes
Do not quote the sample months from memory; open the pricing page. They are estimates, not quotes, and they live there alone.
8. An example shop
Linden & Vale is the example shop in every manual. Last month it rang 834 sales for CAD 281315.16 net. Its shoppers paid like this: cash CAD 58484.45 · card CAD 221515.70 · gift card CAD 826.28. 11 people have a seat.
Read from its own records on October 17, 2026.
Notes
The shop is a generated example company; its figures are read by the same reports the merchant runs, and dated. Nothing here is typed.
9. AI with permissions, not a widget
Aldric, the built-in consultant, sits on every screen with your real numbers. It drafts plans and shows its reasoning; a person approves before anything moves. It never changes data on its own.
Your own assistant gets a seat at the till too — over MCP, with only the permissions you assign, every call tracked and budget-capped.
Notes
The hardened-Aldric law in one line: AI drafts, a human approves. The connector is read-first today; the writes follow behind the same approvals.
10. Learning built in
A manual per role, generated from the system and the example shop — start with How the office works and The owner’s manual. 78 use cases in the playbook, each one a course with a test and a certificate. A guide on every page, and Aldric to ask.
Notes
The manuals are regenerated when the office moves, so they cannot drift. The use cases double as the courses.
11. What is not there yet
Point of sale on Stripe's smart readers, and the register as native apps on Android, iPhone and iPad — the browser register is live today. Your own online storefront. Richer reporting screens and BI feeds. A marketplace of integrations.
No dates are promised here. The service is in its controlled first wave: merchants are onboarded by application.
Notes
The coming words are the public site's own, and they carry no dates by law. Say the first wave plainly.
12. Next steps
Apply — no card, no fee. A person reads it. Prove the system on your own data beside your current till, and expand when you are satisfied.
Notes
The door is the application. Nothing is signed, nothing is bought.
Where these numbers come from
Every figure on these slides is read from the system when the page is rendered — the registries, the rate card, the example shop’s own records — or, for the other side, from dated public list prices. Nothing is typed.
| Figure | Value | Read from |
|---|---|---|
| the monthly fee for the whole business | $15 | the rate card — the base fee and the included kernels, derived at the render |
| the kernels included every month | 100 million kernels | the rate card — the base fee and the included kernels, derived at the render |
| the part of the fee that returns as the month’s kernels | $10 | the rate card — the base fee and the included kernels, derived at the render |
| record types live on the engine | 132 | the construct catalogue joined to the store registries and the live schema — counted at the render |
| tools an AI assistant reaches over MCP | 17 | the MCP tool registry — the meta-tools and the typed tools derived from the live schema, counted at the render |
| writes among the operations | 690 | the live schema the API serves — counted at the render |
| operations on the one open API | 1,085 | the live schema the API serves — counted at the render |
| use cases in the playbook, each a course | 78 | the playbook registry — counted at the render |
| reads among the operations | 395 | the live schema the API serves — counted at the render |
| ready-made reports | 24 | the report catalogue — counted at the render |
| the example shop | Linden & Vale | the figures files of the example shop |
| the net sales last month | CAD 281315.16 | the example shop’s own records — the Sales activity report (sales_activity) over 2026-09-01..2026-09-30, the Total row, read October 17, 2026 |
| when the example shop’s figures were read | October 17, 2026 | the figures files of the example shop |
| the sales rung last month | 834 | the example shop’s own records — the Sales activity report (sales_activity) over 2026-09-01..2026-09-30, the Total row, read October 17, 2026 |
| the people with a seat | 11 people | the example shop’s own records — the User access report (user_access), read October 17, 2026 |
| how shoppers paid last month, by tender | cash CAD 58484.45 · card CAD 221515.70 · gift card CAD 826.28 | the example shop’s own records — the Tender mix report (tender_mix) over 2026-09-01..2026-09-30, the Total rows, read October 17, 2026 |
The doors: How the office works (the developer documentation site) · The owner’s manual (the developer documentation site) · the pricing page (the public site) · Start your evaluation (the public site) · what is live today (the public site).