AUTOMATION & AI SCORECARD · 30 AUGUST 2026 · Source: Skill Register (Notion) + skill manifest + team inventory
Automation Build-up Over Time — live automations & avg maturity
What changed since the 8 August assessment
Budget & finance ops
budget-update was rebuilt into a full month-end pipeline. It now reconciles against Fortnox’s own subtotals, redeploys the Cloudflare budget site by git push, writes the workbook back to Drive as a new revision, and posts the monthly and quarterly cost updates to Notion by itself. The human approval step was replaced by a correctness gate: four checks pass or nothing is written.
L3 → L4
Valuations
ipev-valuation-review shipped. The first skill that drafts judgement rather than moving data: the quarterly IPEV reasoning and seven-question checklist, written from investor updates in Gmail, weekly meeting notes in Notion and the company’s quarterly sheet, with the EUR/SEK rate pulled from Riksbanken. It also overlaps with portfolio-valuation, which answers the same seven questions into Excel. Pick one.
L3 → L3
Tools & governance
The Skill Register drifted. It lists 13 skills against 14 custom skills in the manifest, is missing ipev-valuation-review, and still describes the July version of budget-update. The Automation Ideas board has taken one submission in three weeks. The governance layer was built in early August and has not been fed since.
L2 → L2
Everywhere else
No change. The four lead scrapers are still four skills, due-diligence is still marked as a test, and six of eighteen sub-areas have no AI in the workflow at all.
—
Read: three weeks of real work went into depth, not breadth. One sub-area moved and average maturity is essentially flat at 1.3, because the six zeroes are where the average lives. The two cheapest ways to move the number are both governance, not engineering: schedule one skill (nothing runs on a clock today — all 14 are triggered by someone asking) and bring the register back in sync, since it is what this page reads.
Automation Maturity by Area (0–4)
Maturity Levels
L0
Manual
No AI in the workflow. Done by hand, every time.
L1
Exploring
Ad hoc prompting or early experiments. No repeatable skill yet.
L2
Assisted
A dedicated skill exists and works, but a person triggers and supervises each run.
L3
Automated
Runs systematically every cycle (scheduled or embedded in the process). Human reviews output.
L4
Optimized
Automated, measured (runs, hours saved, quality) and improved continuously.
+ Suggest an automation Skill Register ↗
Drill-down — click an area, then a sub-area, then any item
Maturity is scored per sub-area on the L0–L4 scale. Live = deployed skill in the Skill Register · Test = built, not yet official · Idea = proposed next step · Quick win = high value, low build effort. New suggestions go to the Automation Ideas board in Notion and are triaged at the quarterly skill review; this page is rebuilt after each review.
Assessment of 30 Aug 2026 scored against the skill manifest (14 custom skills) rather than the Skill Register, which is currently behind. Level changes and their reasoning are in the panel above.
Build-up chart: automation counts by actual ship date (skill manifest + dashboard launch dates); pre-June milestones from Notion history. Hollow maturity points are back-estimates; filled points are real snapshots. Source of truth: the Automation History database in Notion — add a snapshot row at each quarterly review, then rebuild this page.