S1
HR & operations
Top 3
What changed in HR systems, compliance and workforce data for a team under 500 people, and what it costs or saves.
Operations
CompTIA's analysis of Bureau of Labor Statistics data found the economy added 86,000 technology workers in August while tech companies cut 14,700 roles. About 600,000 technology postings are open, and more than 320,000 ask for AI skills.
86,000
tech workers added in August, against 14,700 cut
Why this mattersThe demand moved out of tech companies and into manufacturers, retailers and everyone else, so you are bidding against them for the same people. Budget a longer time to hire and check your salary bands for technical roles before the next requisition opens.
Operations
The Labor Department's Veterans' Employment and Training Service said schedule changes, transfers, added scrutiny and exclusion from opportunities can be retaliation under USERRA, which was amended in 2025 to add other retaliatory action.
USERRA
retaliation standard now tracks Title VII
Why this mattersManagers rarely intend retaliation, they just move a shift after someone asks for leave. That is now enough to build a claim on. Add one question to your scheduling and transfer approval step asking whether military service is involved, and route those to HR first.
Operations
Researchers from the US, UK and Switzerland, including Dirk Lindebaum at the University of Bath, describe epistemic de-skilling in managers who overrely on generative AI. A GoTo survey found 50% of employees report excessive AI dependence at work.
50%
of employees report excessive AI dependence at work
Why this mattersThis is a process problem, not a tool problem. AI is useful for drafting a performance note and dangerous for deciding one. Write down which people decisions must be made by a named human, put that line in your AI use policy, and leave the drafting to the tool.
S2
Science & technology
Top 3
Research and technology that is making money, saving money, or about to change how work gets done.
Science
Nature reports that OpenAI claims a solution to the Navier-Stokes problem, one of seven Millennium Prize problems open since 1900, after running 1,000 agents for 50 hours on a simplified version and then 10,000 agents on the full one. Teams at Harvard, NYU and Caltech released partial solutions at the same time.
10,000
AI agents run in parallel on one problem
Why this mattersThe lesson is not that AI is brilliant, it is that a problem defined tightly enough can absorb as much machine effort as you can pay for. Pick one recurring task, write down exactly what a correct output looks like, and then let a tool run at it.
Science
Google DeepMind released AlphaGenome, an atlas predicting the effect of all 9 billion possible single-letter DNA changes plus more than 100 million insertions and deletions, built from roughly a petabyte of computation. Work that Nature describes as unfeasible for most research groups is now a database search.
9 billion
possible DNA changes mapped and published free
Why this mattersWork that used to need a specialist and a budget keeps turning into something you look up. The same shift is reaching comp benchmarking, policy drafting and first-pass document review. Before you pay for the next specialist report, check whether the answer already sits in a tool you own.
Science
Nature reports on teams in San Francisco, at the University of Zurich and at the Technion who trained language models only on data from before 1900 or 1930 to see whether they could rediscover relativity. The models showed flashes of intuition, failed in most cases, and generated wrong theories they could not tell from right ones.
Pre-1900
training cutoff used to test AI discovery
Why this mattersThis is the honest boundary. AI is strong on volume and weak on the leap, which tells you exactly where to point it. Give it the first draft, the summary and the data pull, and keep a person on the calls that need judgment about your business.
S4
Three things to do this week
3 actions
Three moves with the best payoff for the least effort, each with what it is roughly worth. Even one of them saves hours of work, cash, or a risk you would rather not carry.
🧲1
Hiring
Refresh salary bands for technical roles before the next req opens
Pull three current postings for each technical role you expect to hire, note the posted ranges, and compare them to your bands. Fix any band sitting below market before recruiting starts. About two hours.
PayoffA band set a year ago produces a search that stalls at the offer stage. Every extra month a technical role stays open costs the output you hired for plus recruiter and manager time, and a failed search usually restarts the whole spend.
6 weeks
of vacancy avoided on one technical role
Effort 2 hours
📋2
Compliance
Add a service-member check before schedule or transfer changes
Add one question to your scheduling and transfer approval step asking whether the employee has military service obligations, and route those changes to HR before they go through. About one hour.
PayoffUSERRA claims carry back pay, benefits restoration and attorney fees, and the standard now covers actions that merely discourage a complaint. One approval field costs nothing and removes the most common trigger.
$15k+
cost of one retaliation claim avoided
Effort 1 hour
⚡3
Speed
Turn your three most screenshotted processes into a written one-pager
Ask two recent hires which screens they screenshot to get work done, then write those three flows as short numbered steps in the handbook or HRIS. About 90 minutes.
PayoffNew hires who screenshot instead of reading retain less and ask more questions, which costs manager time every week. Three documented flows cut repeat questions and pull days off the ramp for every hire after them.
40+ hrs
of manager time a year on repeat questions
Effort 90 minutes
Want this handled, but no time or capacity?Olga takes on HR ops projects like these on a fractional or interim basis. A 30-minute call is enough to scope one.
Reach out to Olga ↗Payoff figures are order-of-magnitude estimates for a company under 500 people, based on typical costs, not quotes for your business.