<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
<channel>
<title>Spurbuck Advisory Briefings</title>
<link>https://spurbuck.com/insights/</link>
<atom:link href="https://spurbuck.com/insights/feed.xml" rel="self" type="application/rss+xml"/>
<description>Sourced briefings for leaders in Australia and New Zealand deciding what to do about AI.</description>
<language>en-AU</language>
<lastBuildDate>Thu, 19 Nov 2026 06:00:00 +1000</lastBuildDate>
<item>
<title>Management reports that explain their own variances</title>
<link>https://spurbuck.com/insights/management-reports-explain-variances/</link>
<guid isPermaLink="true">https://spurbuck.com/insights/management-reports-explain-variances/</guid>
<pubDate>Thu, 19 Nov 2026 06:00:00 +1000</pubDate>
<dc:creator>Tom Buckley</dc:creator>
<description>Finance teams spend nearly half their effort collecting and checking data, so the commentary a board reads is written last. An agent can trace each variance to its transactions and draft the explanation for a person to approve. Key points: The 2026 FP&amp;A Trends Survey of 475 finance professionals found that 47% of the planning and analysis function's effort goes on collecting and validating data, and only 32% on insight and driving action. APQC's open benchmark puts the median at 10 calendar days from running the first trial balance to completing the period-end management report, across 3,287 organisations. Gartner's 2025 survey of 183 finance leaders found 59% using AI in their finance function; among those, the most common uses were knowledge management (49%), accounts payable automation (37%) and error and anomaly detection (34%). An agent can reconcile the report to the trial balance, find every variance over your threshold, trace each one to the transactions behind it, draft the commentary with those transactions attached, and hold what the ledger cannot explain for the person who owns the cost. A person still approves every word before it reaches the board, and the measure is your own: hours spent on the report, the share of explanations accepted without rewriting, and the questions the board asks that the pack already answers.</description>
</item>
<item>
<title>Match supplier invoices to orders without retyping</title>
<link>https://spurbuck.com/insights/supplier-invoices-without-retyping/</link>
<guid isPermaLink="true">https://spurbuck.com/insights/supplier-invoices-without-retyping/</guid>
<pubDate>Mon, 16 Nov 2026 06:00:00 +1000</pubDate>
<dc:creator>Tom Buckley</dc:creator>
<description>Most supplier invoices still arrive as PDFs that someone reads, retypes and checks by hand. An agent can read each one, match it to the order and the receipt, and hold only the exceptions for a person. Key points: Deloitte Access Economics estimates, published by the ATO, put the average cost of processing a PDF invoice at $27.67, against $9.18 for an e-invoice, with 60% of the cost falling on the business that receives it. The ATO reports that 6,012,297 e-invoices had been sent over Australia's Peppol network by the end of August 2026, and that the first million took four years. In New Zealand, MBIE has put business-to-business invoices at over 280 million a year, against more than one million e-invoices exchanged there in total. The National Anti-Scam Centre counted $166.8 million in reported losses to payment redirection scams in 2025, up 9.3% on 2024; it says false billing reports from businesses generally relate to these scams. An agent can read each PDF as it arrives, match it to the purchase order and the goods receipt, prepare the matched ones for a person to approve in a batch, and hold price differences, short deliveries, missing orders and any change of bank details for a person. Measure it against your own baseline: the share of invoices that pass without anyone editing a field, the days from receipt to approval, and the differences found before payment rather than after.</description>
</item>
<item>
<title>Draft tender responses from your best past answers</title>
<link>https://spurbuck.com/insights/tender-responses-from-past-answers/</link>
<guid isPermaLink="true">https://spurbuck.com/insights/tender-responses-from-past-answers/</guid>
<pubDate>Sat, 14 Nov 2026 06:00:00 +1000</pubDate>
<dc:creator>Tom Buckley</dc:creator>
<description>Bid teams spend about 33 hours on each response, and half say they lack the bandwidth. An agent can draft the repeat answers from approved past ones, with the source shown, and leave people the parts that win. Key points: The Department of Finance says the Commonwealth awarded $104.9 billion to suppliers in 2024-25, the highest value ever reported on AusTender, and 45,556 of those contracts went to small and medium businesses. Loopio's survey of 1,533 companies that respond to requests for proposal found teams spent an average of 33 hours on each response in 2025 and submitted 166 of them, while the average win rate fell to 39% from 45%. Half the respondents in the same survey ranked having enough bandwidth as a top challenge, up from 31% a year earlier, and 49% named working with subject matter experts. Much of a response is not new writing: it is the same company description, policies, method statements and project examples, found and adapted again under deadline. An agent can read the request, list every criterion and mandatory form, draft the repeat answers from an approved library with the source of each shown, and hold anything stale or new for a person, so the bid team spends its hours on pricing and win themes.</description>
</item>
<item>
<title>Carry supplier price rises into your quotes the same week</title>
<link>https://spurbuck.com/insights/supplier-price-rises-into-quotes/</link>
<guid isPermaLink="true">https://spurbuck.com/insights/supplier-price-rises-into-quotes/</guid>
<pubDate>Fri, 13 Nov 2026 06:00:00 +1000</pubDate>
<dc:creator>Tom Buckley</dc:creator>
<description>Australian manufacturers paid 8.9% more for inputs over the year and charged 6.1% more. Some of that gap can be quotes priced on last month's costs. An agent can read each price notice and recost every open quote overnight. Key points: The ABS says prices of inputs to Australian manufacturing rose 8.9% in the year to the September quarter of 2026, while the prices manufacturers received for their output rose 6.1%. Input prices did not rise evenly: the ABS index rose 4.9% in the December quarter of 2025 alone, then fell 1.0% in the September quarter of 2026 as oil came off its peak. In a March 2026 survey of mostly small and medium businesses reported by the Reserve Bank, respondents expected input costs to rise about 10 to 15% over a year and expected to pass on only about 30% of that. Some of what is not passed on is not a pricing decision at all: it is quotes going out on costs that were out of date the day they were sent. An agent can read each supplier price notice, propose the change to the cost file for a person to approve, recost every open quote overnight and hold the ones that fall below the margin floor for an estimator.</description>
</item>
<item>
<title>Get paid sooner: debtor follow-up that happens every day</title>
<link>https://spurbuck.com/insights/debtor-follow-up-every-day/</link>
<guid isPermaLink="true">https://spurbuck.com/insights/debtor-follow-up-every-day/</guid>
<pubDate>Thu, 12 Nov 2026 06:00:00 +1000</pubDate>
<dc:creator>Tom Buckley</dc:creator>
<description>Overdue invoices get chased late, or not at all, because chasing them all takes more hours than a credit team has. An agent can follow up every account on time and hand people the disputes. Key points: Xero's data shows Australian small businesses were paid an average of 6.6 days after the due date in the December quarter of 2025, with wholesale trade (8.8 days) and manufacturing (8.7 days) among the latest-paid industries. In Atradius's 2025 survey of Australian receivables staff, a third (33%) named delays in their own invoicing among the top reasons customers pay late, and 27% named disputes or problems with invoices. Each day of debtor days is worth one day of sales on account: for a business with $60 million a year on account, seven days is about $1.15 million of cash, once (illustrative). An agent can check, prepare and send follow-up for every overdue account each morning, with the invoice and proof of delivery attached, and hold disputes, broken promises and credit decisions for a person. Measure it against your own baseline, not a vendor's claim: debtor days, the share of the ledger over 60 days and the time to settle a dispute, for 13 weeks before and after.</description>
</item>
<item>
<title>Automated decisions: the 10 December privacy deadline</title>
<link>https://spurbuck.com/insights/automated-decisions-privacy-deadline/</link>
<guid isPermaLink="true">https://spurbuck.com/insights/automated-decisions-privacy-deadline/</guid>
<pubDate>Tue, 10 Nov 2026 06:00:00 +1000</pubDate>
<dc:creator>Tom Buckley</dc:creator>
<description>From 10 December 2026, Australian privacy policies must say where computer programs make or shape decisions about people. A person signing off does not take a process out of scope. Here is what to map and what to write. Key points: From 10 December 2026, an organisation covered by Australia's Privacy Act must say in its privacy policy when a computer program makes, or substantially helps make, decisions that could significantly affect people, according to the OAIC's updated guidelines. A person signing off does not take a process out of scope: the OAIC says a program is captured where the human decision-maker relies on or is significantly influenced by its output. The rule is not only about AI. The OAIC reads the term computer program broadly, to include rule-based processes, and the law firm Allens reads the guidance as reaching everyday software such as spreadsheets. In the OAIC's 2026 community attitudes survey, 79% of Australians said they should be informed when AI is used, and only 25% found automated eligibility or risk decisions acceptable. The work is an inventory before it is a rewrite: list where programs feed decisions about people, test each one against three questions, then describe the kinds of decisions and information in plain English.</description>
</item>
<item>
<title>What an AI agent really costs to run</title>
<link>https://spurbuck.com/insights/what-an-ai-agent-costs-to-run/</link>
<guid isPermaLink="true">https://spurbuck.com/insights/what-an-ai-agent-costs-to-run/</guid>
<pubDate>Tue, 10 Nov 2026 06:00:00 +1000</pubDate>
<dc:creator>Tom Buckley</dc:creator>
<description>The price of AI has fallen more than 280-fold, yet AI bills are rising. Agents use far more of it per task, and the model is only one line in the budget. Here is how to price an agent before you build it. Key points: The cost of querying a capable AI model fell from US$20 to US$0.07 per million tokens in about two years, a drop of more than 280-fold. Bills are rising anyway. Gartner says agents need 5 to 30 times more tokens per task than a chatbot, and expects overall inference costs to increase. Gartner also predicts more than 40% of agentic AI projects will be cancelled by the end of 2027, citing escalating costs and unclear business value. The fix is to budget per finished task, across five cost lines, before the build starts. The model is often the smallest of the five.</description>
</item>
<item>
<title>Why most AI pilots never reach production</title>
<link>https://spurbuck.com/insights/why-ai-pilots-stall/</link>
<guid isPermaLink="true">https://spurbuck.com/insights/why-ai-pilots-stall/</guid>
<pubDate>Mon, 09 Nov 2026 06:00:00 +1000</pubDate>
<dc:creator>Tom Buckley</dc:creator>
<description>The research is consistent: the model is rarely what fails. What fails is everything around it. Here is what the numbers say, the five gates a pilot has to clear, and the questions a board should ask before funding the next one. Key points: MIT's 2025 study of enterprise generative AI found about 5% of pilots deliver rapid revenue growth; most stall with little measurable effect on profit and loss. In S&amp;P Global's 2025 survey, 42% of companies abandoned most of their AI initiatives, up from 17% a year earlier, and the average organisation scrapped 46% of proofs of concept before production. The causes named are the same across studies: poor data, weak risk controls, rising costs, unclear value and tools that never connect to how work actually gets done. Australia's Guidance for AI Adoption sets out six essential practices that map neatly onto what separates a pilot from a production system.</description>
</item>
</channel>
</rss>
