diff --git a/public/assets/template-thumbs/case-study.png b/public/assets/template-thumbs/case-study.png index 5887a14..4cdfa4b 100644 Binary files a/public/assets/template-thumbs/case-study.png and b/public/assets/template-thumbs/case-study.png differ diff --git a/public/assets/template-thumbs/daily-digest.png b/public/assets/template-thumbs/daily-digest.png index 345ac96..8fb2b1e 100644 Binary files a/public/assets/template-thumbs/daily-digest.png and b/public/assets/template-thumbs/daily-digest.png differ diff --git a/public/assets/template-thumbs/interview-notes.png b/public/assets/template-thumbs/interview-notes.png index 19d7a82..a17ff4d 100644 Binary files a/public/assets/template-thumbs/interview-notes.png and b/public/assets/template-thumbs/interview-notes.png differ diff --git a/public/assets/template-thumbs/launch-announcement.png b/public/assets/template-thumbs/launch-announcement.png index 85f9b34..779558b 100644 Binary files a/public/assets/template-thumbs/launch-announcement.png and b/public/assets/template-thumbs/launch-announcement.png differ diff --git a/templates/case-study.html b/templates/case-study.html index b630bd4..480df56 100644 --- a/templates/case-study.html +++ b/templates/case-study.html @@ -1,156 +1,22 @@ - - + + Northwind Freight cut invoice disputes by 71% in one quarter - -
+
-
- Case study · Freight & logistics + Case study · Freight & logistics +

Northwind Freight cut invoice disputes by 71% in one quarter

71%

@@ -165,39 +31,40 @@ body{padding:var(--s5) var(--s4) var(--s8);font-size:.9375rem;line-height:1.5}
-
-
“We never had a disputes problem. We had a problem where nobody could see the weight - ticket and the invoice at the same time.”
-
Priya Raman, VP Finance, Northwind Freight
-
- -
- +
+
- - + + - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + -
Before and after · Q2 2025 baseline vs Q2 2026
MetricBeforeAfterChange
MetricBeforeAfterChange
Invoices disputed19.4%5.6%−71%
Clerk hours spent on disputes854 / mo179 / mo−79%
Days sales outstanding61 days43 days−30%
Time to resolve one dispute22 min16 min−27%
Unrecovered accessorial write-offs$310K / yr$88K / yr−72%
Invoices disputed19.4%5.6%−71%
Clerk hours spent on disputes854 / mo179 / mo−79%
Days sales outstanding61 days43 days−30%
Time to resolve one dispute22 min16 min−27%
Unrecovered accessorial write-offs$310K / yr$88K / yr−72%
Bars show the after figure as a share of the before figure — +
Bars show the after figure as a share of the before figure — shorter is better. Every row is pulled from Northwind's TMS and billing ledger, not from our own telemetry.
-
+

Two branches still run a manual accessorial log and are excluded from the figures.

+ + +
+
“We never had a disputes problem. We had a problem where nobody could see the weight + ticket and the invoice at the same time.”
+
Priya Raman, VP Finance, Northwind Freight
+
@@ -228,8 +95,7 @@ body{padding:var(--s5) var(--s4) var(--s8);font-size:.9375rem;line-height:1.5}

The first fortnight was bad. The hold threshold shipped at $10, which held 31% of all invoices and made the queue worse than the disputes had been. Raising it to $25 on 23 March dropped holds to 7% and is the single change that made the numbers above possible.

-

Two branches still run a manual accessorial log and are excluded from the figures.

-
+
“The first month I kept refreshing the dispute queue because I assumed it was broken.”
@@ -254,6 +120,6 @@ body{padding:var(--s5) var(--s4) var(--s8);font-size:.9375rem;line-height:1.5} 2026; the comparison quarter is April–June 2026. Two of twenty-two branches are excluded. Published with permission. -
+
- + \ No newline at end of file diff --git a/templates/daily-digest.html b/templates/daily-digest.html index 0e178b7..61e8785 100644 --- a/templates/daily-digest.html +++ b/templates/daily-digest.html @@ -1,145 +1,14 @@ - - + + AI Engineering Daily — 3 September 2026 - - @@ -186,10 +55,11 @@ summary span{color:var(--subtle);font-weight:400;font-size:.8125rem}
-

Postgres 18 ends the case for a connection pooler

+

Postgres 18 ends the case for a connection pooler

Hexadecimal24 min06:40

-
+

Watch it if you own the database tier.

+

TL;DRBuilt-in connection multiplexing lands in 18. Under ~500 clients it removes the case for PgBouncer; above that the pooler wins on failover.

-

Watch it if you own the database tier.

@@ -206,10 +75,11 @@ summary span{color:var(--subtle);font-weight:400;font-size:.8125rem}
-

We replaced our vector database with one table

+

We replaced our vector database with one table

Latent Space38 min09:05

-
+

Watch it if your retrieval bill has a comma in it.

+

TL;DRA 40M-chunk corpus moved from a managed vector store into pgvector with HNSW. Recall held 0.94, p95 rose 11 ms, the bill fell $9,400/mo.

-

Watch it if your retrieval bill has a comma in it.

@@ -226,17 +95,17 @@ summary span{color:var(--subtle);font-weight:400;font-size:.8125rem}
-

The cost curve nobody budgets for

+

The cost curve nobody budgets for

Infra Weekly17 min11:20

-
+

Watch it if you sign off the cloud invoice.

+

TL;DREgress, not compute, turns a cheap architecture expensive: a $300-a-month cluster carrying $4,100 of cross-zone traffic on one misplaced replica.

-

Watch it if you sign off the cloud invoice.

@@ -245,17 +114,17 @@ summary span{color:var(--subtle);font-weight:400;font-size:.8125rem}
-

Shipping agents other teams actually adopt

+

Shipping agents other teams actually adopt

Build Notes52 min14:02

-
+

Watch it if you are rolling agents out.

+

TL;DRAdoption tracked with one thing — whether the agent produced an artifact a human could forward. Chat-only agents churned in two weeks.

-

Watch it if you are rolling agents out.

@@ -277,4 +146,4 @@ summary span{color:var(--subtle);font-weight:400;font-size:.8125rem} - + \ No newline at end of file diff --git a/templates/interview-notes.html b/templates/interview-notes.html index 410b527..b489522 100644 --- a/templates/interview-notes.html +++ b/templates/interview-notes.html @@ -1,117 +1,17 @@ - - + + Interview notes — Ana Petrović, Rovna Logistics - - -
+
Interview notes · 6 of 9 @@ -119,10 +19,9 @@ blockquote p{font-size:clamp(1.1875rem,2.3vw,1.5rem);font-weight:600;line-height

The output was never the problem. The schedule and provenance were.

-
- +

If it had told me it had skipped a week, I would have kept it.

-
Ana Petrović · 41:02 · on what would have changed her mind
+
Ana Petrović · 41:02 · on what would have changed her mind
@@ -137,7 +36,7 @@ blockquote p{font-size:clamp(1.1875rem,2.3vw,1.5rem);font-weight:600;line-height
-
+
What we came to learn

Why teams turn a scheduled agent off in month two. Ana ran a Monday metrics job for seven weeks, then disabled it.

@@ -145,41 +44,32 @@ blockquote p{font-size:clamp(1.1875rem,2.3vw,1.5rem);font-weight:600;line-height

Findings

-
-
08:12
-
-

The report was fine. The problem was that on Tuesday nobody could tell me whether it had actually run.

-
Ana Petrović, on the first three weeks
-
+
+
08:12

The report was fine. The problem was that on Tuesday nobody could tell me whether it had actually run.

+
Ana Petrović, on the first three weeks
+
What follows

Every churned team so far describes the same gap: the output was trusted, the schedule was not. Ana checked manually for three weeks, decided that was the job she had been automating, and stopped.

-
-
+ -
-
21:40
-
-

I forwarded the first one to my director and she asked me to just paste it into the doc. So I was doing it by hand anyway.

-
Ana Petrović, on sharing the output upward
-
+
+
21:40

I forwarded the first one to my director and she asked me to just paste it into the doc. So I was doing it by hand anyway.

+
Ana Petrović, on sharing the output upward
+
What follows

The artifact was not the unit of sharing in her organisation — the weekly doc was. A correct output that cannot travel the last ten metres fails where no output fails. Second interview with this shape (see 03).

-
-
+ -
-
35:05
-
-

We turned it off when the numbers moved and nobody had changed anything. I could not see what it had read.

-
Ana Petrović, on the week before she disabled it
-
+
+
35:05

We turned it off when the numbers moved and nobody had changed anything. I could not see what it had read.

+
Ana Petrović, on the week before she disabled it
+
What follows

Trust broke on provenance, not accuracy — the figure was right and she still could not tell which query produced it. Three unprompted mentions in six interviews, and no question about it in the guide.

-
-
+
@@ -201,6 +91,6 @@ blockquote p{font-size:clamp(1.1875rem,2.3vw,1.5rem);font-weight:600;line-height
Recorded with consent; transcript held 90 days, then deleted. Quotes are verbatim, lightly de-hesitated; timestamps refer to the recording. Interview 6 of 9 · synthesis 18 September.
-
+
- + \ No newline at end of file diff --git a/templates/launch-announcement.html b/templates/launch-announcement.html index fb5b9a0..b39a574 100644 --- a/templates/launch-announcement.html +++ b/templates/launch-announcement.html @@ -1,362 +1,19 @@ - - + + Fieldnote 2.0 is available today - - + -
-
-

Available today · 4 September 2026

+
+

Available today · 4 September 2026

Fieldnote 2.0 works with no signal at all

Survey the whole site offline; it syncs itself when you get back to the truck. @@ -371,58 +28,76 @@

Upgrade
Free, all plans
Downtime
None — nothing to migrate
-
-
- - -
-
-

The headline change

+
+

The headline change

Offline is the default, not the fallback

-

- Every form, photo and signature is written to the device first and queued. The queue - drains by itself the moment there is a network, in the order the records were created. -

-

- Conflicts are boring now too: two people editing one record produces two versions, both - kept, merged in one tap. Nothing is silently overwritten. -

-
-
- +
+ A record is saved on the device, held in a queue while offline, then uploaded in order when the network returns. - - - - - - - - - - - - - - - - - - + + + + + + + + + + + + + + + + + +

On deviceQueuedSynced

The round trip needs no connection at any point.
+
+

+ Every form, photo and signature is written to the device first and queued. The queue + drains by itself the moment there is a network, in the order the records were created. +

+

+ Conflicts are boring now too: two people editing one record produces two versions, both + kept, merged in one tap. Nothing is silently overwritten. +

+
+ + + + + +
+
+

From nine weeks of beta

+ +
- -
-

Also in 2.0

+
+

Also in 2.0

- -
-
-

From nine weeks of beta

-
    -
  • -

    241,800

    -

    records captured offline by 1,900 crews

    -
    • -
  • -

    18 sec

    -

    median time to drain a full day's queue

    -
    • -
  • -

    0

    -

    records lost to a failed sync

    -
    • -
-
-
- - -
+

Try it on your next site visit

  1. @@ -475,7 +128,7 @@

  2. Download your sites

    Settings → Offline, then pick the sites for the week. About 40MB each.

    - fieldnote sites pull --week current + fieldnote sites pull --week current

  3. Turn the radio off and work

    @@ -487,7 +140,6 @@
-

Availability

@@ -519,4 +171,4 @@

- + \ No newline at end of file diff --git a/templates/manifest.ts b/templates/manifest.ts index a80b797..72c81ee 100644 --- a/templates/manifest.ts +++ b/templates/manifest.ts @@ -4,21 +4,23 @@ export const starterTemplateManifest = [ "slug": "daily-digest", "category": "meetings", "name": "Daily digest", - "description": "Many items condensed with an index rail: a jump index and today's themes on the left, per-item cards with a TL;DR on the right, and the long tail collapsed to one line each.", + "description": "A newspaper-style daily desk: a complete story index beside four open, ruled articles with relevance, summaries and evidence, followed by a native six-item archive.", "type": "html", "content_file": "daily-digest.html", "thumbnail": "/assets/template-thumbs/daily-digest.png", - "slots": [] + "slots": [], + "thumbnail_viewport": 1600 }, { "slug": "interview-notes", "category": "meetings", "name": "Interview notes", - "description": "One conversation, quotes-led. Who they are, what they said in a timestamped pull quote, and your reading kept visibly separate. For user research and hiring debriefs.", + "description": "A voice-and-interpretation spread: a broad serif quotation beside the participant identity, then paired verbatim evidence and analyst readings with timed quotes and owned next actions.", "type": "html", "content_file": "interview-notes.html", "thumbnail": "/assets/template-thumbs/interview-notes.png", - "slots": [] + "slots": [], + "thumbnail_viewport": 1440 }, { "slug": "meeting-recap", @@ -94,7 +96,7 @@ export const starterTemplateManifest = [ "slug": "case-study", "category": "research", "name": "Case Study", - "description": "A customer outcome told with the customer's own numbers: one display figure and what it is measured against, a before-and-after ledger with magnitude bars, the situation that caused it, an honest account of what the change cost, and quotes from the people who lived it.", + "description": "An evidence spread: the customer outcome opposite five before/after metric records, followed by serif customer voices, the operational story and an equally prominent honest-cost section.", "type": "html", "content_file": "case-study.html", "thumbnail": "/assets/template-thumbs/case-study.png", @@ -179,11 +181,12 @@ export const starterTemplateManifest = [ "slug": "launch-announcement", "category": "releases", "name": "Launch announcement", - "description": "The friendly end of the releases family: an inverted black masthead with the headline claim and the facts strip, one lead feature with a diagram, two supporting ones, beta numbers, three steps to try it, and a plain-spoken availability table.", + "description": "A dark launch poster pairs the headline and update links with an offline-flow diagram, then shifts to a light beta-evidence strip, open feature rows and a practical trial sequence.", "type": "html", "content_file": "launch-announcement.html", "thumbnail": "/assets/template-thumbs/launch-announcement.png", - "slots": [] + "slots": [], + "thumbnail_viewport": 1440 }, { "slug": "migration-guide",