Under Pressure
Why AI Handles the Noise So Your People Can Finally Do Their Actual Jobs
"Pressure pushing down on me
Pressing down on you, no man ask for..."
— Queen & David Bowie, "Under Pressure" (1981)
Freddie Mercury and David Bowie didn't write "Under Pressure" about eliminating pressure. They wrote it about what happens when pressure meets the right person at the right moment. The bass line that opens the song? That's the weight of the world. The vocal explosion that follows? That's what humans do when the noise clears and they can finally respond.
Every knowledge worker in America is living the first four bars of that bass line right now. Pressure pushing down. But here's the thing nobody talks about:
It's the wrong pressure.
Your procurement specialist didn't get a Master's in Public Administration to compare HVAC filter prices across three vendors. Your business development manager didn't spend 15 years building C-suite relationships to redline font sizes in a 200-page RFP response. Your IT director didn't earn three certifications to reset passwords.
The pressure is real. But it's not the pressure they signed up for. It's not the friction they were trained to handle. It's noise. And noise is exactly what AI was built to filter.
The Friction Paradox: Hired for Signal, Buried in Noise
There's a cruel irony in modern knowledge work. Organizations spend months recruiting, interviewing, and selecting professionals specifically for their judgment, domain expertise, and relationship skills. Then they hand them a laptop and say: "Great, now spend 60% of your day doing things a spreadsheet macro could handle."
| What They Were Hired For | What They Actually Do All Day |
|---|---|
| Strategic vendor negotiation | Cross-referencing PO numbers across three systems |
| Client relationship building | Reformatting proposal templates to match RFP specs |
| Risk assessment and compliance strategy | Chasing missing W-9 forms via email chains |
| Budget optimization across departments | Manually entering line items into the ERP |
| Complex problem-solving under ambiguity | Generating status reports from five different dashboards |
This isn't a technology problem. It's a misallocation of human capital. And it's been happening so long that we've normalized it. We call it "administrative overhead" or "the cost of doing business." We schedule it into sprint capacity. We budget FTEs for it.
We hired concert pianists and asked them to tune the instruments.
"It's the terror of knowing what this world is about
Watching some good friends screaming, 'Let me out!'"
— Every knowledge worker at 3 PM on a Tuesday
7:42 AM: A Day in the Life of Maria, Public Sector Procurement
Maria has a Master's in Public Administration from UW-Madison. She's spent 12 years in county procurement. She can negotiate a fleet vehicle contract that saves taxpayers $400K over five years. She knows which vendors deliver and which ones ghost you after the PO clears. She's the reason the county's road salt doesn't run out in February.
Here's Maria's actual Monday:
7:42 AM — The Inbox Avalanche
47 emails overnight. 31 are vendor confirmations she needs to cross-reference against open POs. 8 are internal requests with incomplete information. 4 are compliance notices about expiring contracts. 4 are CC'd FYIs she didn't need.
Time spent: 45 minutes sorting and triaging
8:30 AM — The Spreadsheet Shuffle
Three departments submitted purchase requests. Different formats. One used the 2023 template. One used a PDF with handwritten notes. One sent a Teams message with a photo of a whiteboard. Maria manually enters all of them into the procurement system, cross-referencing budget codes against the approved FY2026 allocations.
Time spent: 90 minutes of data entry and format conversion
10:00 AM — The Compliance Chase
The state auditor flagged three vendors missing current W-9 forms. Maria calls vendor A (voicemail). Emails vendor B (auto-reply: "Out of office until March 3"). Vendor C picks up, promises to send it "by end of day" for the fourth time.
Time spent: 60 minutes of phone tag and email chains
11:00 AM — The Price Comparison Ritual
Facilities needs 200 HVAC filters. Maria pulls quotes from three approved vendors, compares unit prices, checks delivery timelines, verifies they're on the state cooperative purchasing list, and generates a comparison memo. The total order is $3,200. The comparison process costs more in her time than the savings between vendors.
Time spent: 75 minutes for a $3,200 order with a $47 price differential
12:30 PM — Lunch (Theoretically)
Maria eats a granola bar at her desk while reformatting a quarterly spend report for the county board. Different board members want different formats. She makes three versions.
Time spent: 45 minutes of reformatting during "lunch"
1:15 PM — The Actual Work (Finally)
Maria finally gets to review the fleet vehicle contract renewal. This is a $2.1M decision over five years. She needs to evaluate total cost of ownership, maintenance terms, fuel efficiency projections, resale value estimates, and vendor reliability history. This is the work that saves the county real money. This is what her MPA trained her for.
Time available: 90 minutes before the next meeting
3:00 PM — The Meeting About the Meeting
Status update with department heads. Maria presents the spend report she reformatted during lunch. Nobody asks about the fleet contract. Everyone asks about the HVAC filters.
Time spent: 60 minutes in a meeting that could have been an email
Maria's Monday — By the Numbers:
| Activity | Time | Requires MPA? |
|---|---|---|
| Email triage & sorting | 45 min | No |
| Data entry & format conversion | 90 min | No |
| Compliance document chasing | 60 min | No |
| Commodity price comparison | 75 min | No |
| Report reformatting | 45 min | No |
| Status meeting | 60 min | Partially |
| Fleet contract analysis | 90 min | YES |
Time spent on work requiring her expertise: 90 minutes out of 465. That's 19%.
The county is paying for an expert and getting a data entry clerk 81% of the time.
Maria's Monday — With AI Agents
Same Maria. Same county. Same vendors. Same compliance requirements. Different allocation of pressure.
7:42 AM — Inbox Already Triaged
AI agent has sorted overnight emails. Vendor confirmations are auto-matched to open POs — 28 of 31 matched clean, 3 flagged with discrepancies for Maria's review. Incomplete internal requests have been sent back to requestors with specific questions about what's missing. Compliance notices are calendared with deadlines. FYI emails are summarized in a single digest.
Maria's time: 10 minutes reviewing 3 flagged items
8:00 AM — Purchase Requests Already Standardized
AI agent ingested the three purchase requests — regardless of format (old template, PDF with handwriting, Teams photo of a whiteboard). OCR + extraction + budget code validation happened overnight. All three are in the procurement system, flagged where budget allocation looks tight. Maria approves two, sends one back with a question about the cost center.
Maria's time: 15 minutes of review and approval
8:15 AM — Compliance Is Handled
AI agent sent automated W-9 reminders three weeks ago, followed up last week, and escalated this morning with "final notice before vendor hold" language. Vendor A submitted theirs via the self-service portal at 6 AM. Vendor B's auto-reply triggered a calendar reminder for March 4. Vendor C's fourth "by end of day" promise triggered an automatic vendor status flag.
Maria's time: 0 minutes. She sees a dashboard summary.
8:15 AM — HVAC Filters? Already Done.
Commodity purchases under $5,000 from approved cooperative vendors are auto-processed. AI pulled quotes, verified cooperative pricing, selected lowest compliant bid, generated the comparison memo, and routed for auto-approval. Maria gets a notification: "200 HVAC filters ordered, $3,187, delivery March 5."
Maria's time: 0 minutes. She didn't need to touch this.
8:30 AM — The Actual Work. All Morning.
Maria opens the fleet vehicle contract with four uninterrupted hours ahead of her. She calls the vendor rep she's known for eight years — the one who'll tell her off the record which maintenance packages are padding and which ones actually matter. She cross-references his insights against the AI-prepared analysis of five years of maintenance records across the county's current fleet. She finds a $340K savings opportunity in the warranty structure that no spreadsheet comparison would have caught — it required understanding that the county's plow trucks take a specific kind of beating that voids standard commercial warranties.
Time available: 4+ hours of deep, strategic work
12:30 PM — Actual Lunch
Maria eats lunch in the break room. She talks to a colleague about an idea for consolidating office supply vendors across four departments — a strategic conversation she hasn't had time for in months.
The spend reports generated themselves. Three formats. Auto-distributed.
Maria's New Monday — By the Numbers:
| Activity | Before AI | After AI |
|---|---|---|
| Email triage | 45 min | 10 min |
| Data entry & formatting | 90 min | 15 min |
| Compliance chasing | 60 min | 0 min |
| Commodity purchasing | 75 min | 0 min |
| Report generation | 45 min | 0 min |
| Strategic work | 90 min (19%) | 340+ min (73%) |
Maria didn't get replaced. Maria got liberated.
And the $340K warranty insight? That pays for the AI investment for the next decade.
9:15 AM: A Day in the Life of Chris, Business Development Manager
"Chippin' around, kick my brains 'round the floor
These are the days it never rains but it pours..."
— The BDM's anthem, every RFP season
Chris has closed $47M in enterprise deals over 15 years. He can read a room in thirty seconds — knows when the CFO is bluffing about budget constraints, knows when the CTO is genuinely worried about integration, knows when the deal is dead but nobody wants to say it out loud. His Rolodex (yes, he still calls it that) is worth more than most startups.
Here's what Chris actually does with his week:
The 60% — Noise
- RFP Summarization (8 hrs/week)
Reading 150-page RFPs to extract the 12 requirements that actually matter. Ctrl+F "shall" 347 times. - Proposal Formatting (6 hrs/week)
The client wants 11pt Calibri, 1-inch margins, section numbering in 1.1.1 format. The last proposal was in Arial. Reformatting everything. Again. - Contract Redlines (4 hrs/week)
Legal sends back redlines. 90% are boilerplate language changes. Chris tracks them in a spreadsheet because the contract management system doesn't integrate with the CRM. - CRM Updates (3 hrs/week)
Logging calls, updating pipeline stages, entering notes from meetings he had three days ago. The CRM is where sales opportunities go to be documented, not where deals actually happen. - Status Reports (3 hrs/week)
Weekly pipeline review deck. Monthly forecast. Quarterly board summary. Same data, three formats, three audiences.
Total: ~24 hours/week
The 40% — Signal
- Client Meetings (6 hrs/week)
Face-to-face and video calls where relationships deepen and deals move forward. The conversations where Chris reads the room and adjusts strategy in real time. - Strategic Positioning (4 hrs/week)
Figuring out how to frame the solution for this specific client's pain points. What keeps their VP up at night? What does their board care about this quarter? - Relationship Building (3 hrs/week)
The lunch where nothing gets "sold" but everything gets understood. The check-in call after a client's company gets acquired. The golf game where the real objection finally surfaces. - Deal Architecture (3 hrs/week)
Structuring pricing, negotiating terms, designing implementation phasing that makes the CFO comfortable and the CTO excited.
Total: ~16 hours/week
Chris is a $47M closer spending 60% of his time as an executive assistant.
Chris's Week — With AI Agents
RFP Summarization: 8 hours → 30 minutes
AI agent ingests the 150-page RFP, extracts mandatory requirements, identifies evaluation criteria and weightings, flags deal-breakers, and produces a 2-page executive summary with a go/no-go recommendation. Chris reviews the summary, asks two clarifying questions, and makes the call in 30 minutes instead of 8 hours.
Proposal Formatting: 6 hours → 15 minutes
AI agent maps proposal content to RFP format requirements automatically. 11pt Calibri? Done. Section numbering? Mapped. Compliance matrix? Generated from the content. Chris reviews the final output and adds one personal anecdote to the executive summary that will make the client smile.
Contract Redlines: 4 hours → 20 minutes
AI agent categorizes redlines: boilerplate (auto-accept), standard legal positioning (template response), and substantive business terms (flag for Chris). Of 47 redlines, 38 are handled automatically. Chris reviews 9 that actually affect deal economics.
CRM Updates: 3 hours → 0 minutes
AI agent captures meeting notes from call transcripts, updates pipeline stages based on conversation signals, and logs activities automatically. Chris doesn't touch the CRM. The CRM touches itself.
Status Reports: 3 hours → 5 minutes
Pipeline review, monthly forecast, quarterly summary — all auto-generated from CRM data. Three formats, three audiences, zero manual effort. Chris skims for anything that looks wrong and hits send.
Chris's New Week:
Before: Noise = 24 hrs/week
RFP reading, reformatting, redline tracking, CRM data entry, report generation
After: Noise = 1.5 hrs/week
Review summaries, approve auto-actions, skim reports
22.5 hours recovered. That's not "efficiency." That's three more client dinners, two more deal architecture sessions, and the mental bandwidth to spot the $4M opportunity he's been too busy formatting proposals to notice.
The Pattern: Click-Ops Is Not the Job
Maria and Chris aren't unique. The pattern is everywhere:
The Teacher
Hired for: Inspiring 30 eighth-graders to care about algebra.
Spends time on: Grading worksheets, filling out IEP paperwork, generating progress reports, entering attendance into three systems that don't talk to each other.
AI reclaims: The 15 hours/week of administrative work so she can spend it on the kid who's struggling silently in the back row.
The Paralegal
Hired for: Case research and document analysis that supports winning legal strategies.
Spends time on: Bates-stamping discovery documents, formatting citations, checking cross-references, organizing exhibit binders.
AI reclaims: The research depth that finds the precedent that wins the case — not the formatting that makes it look pretty in the binder.
The Nurse
Hired for: Patient care, clinical judgment, and the bedside manner that helps people heal.
Spends time on: Charting in Epic, documenting vitals, filling out incident reports, coordinating insurance pre-authorizations.
AI reclaims: The time with the post-op patient who's scared and just needs someone to explain what's happening next.
The Financial Analyst
Hired for: Identifying market opportunities and advising executives on strategic investments.
Spends time on: Pulling data from Bloomberg, formatting it into PowerPoint, reconciling numbers across twelve spreadsheets, making sure the chart colors match the brand guide.
AI reclaims: The analysis that catches the acquisition target everyone else missed.
The pattern is always the same: hire for expertise, bury in administration. AI doesn't replace the expertise. It excavates it.
The "Under Pressure" Insight
"Turned away from it all like a blind man
Sat on a fence but it don't work
Keep coming up with love but it's so slashed and torn..."
— The sound of a professional who went into their field to make a difference, drowning in administrative overhead
"Under Pressure" is one of the greatest songs ever recorded. And the reason it resonates 45 years later isn't because it's about eliminating pressure. It's because it's about the right kind of pressure.
Listen to the song structure. The bass line opens with mechanical, repetitive pressure — da-da-da da-da-da-da. That's the click-ops. That's the spreadsheet shuffling and the CRM updates and the reformatting. It's relentless. It's monotonous. It takes up space.
Then Freddie and Bowie enter. And suddenly the pressure transforms. It becomes human. Emotional. Creative. The same four notes are still there, but now they're the foundation for something extraordinary.
That's the transformation AI enables.
AI doesn't eliminate friction. Friction is where all the value is. Negotiating a fleet contract that saves $340K — that's friction. Reading the room in a client meeting and pivoting the pitch — that's friction. Convincing a scared patient that the procedure will be okay — that's friction.
What AI eliminates is the wrong friction. The mechanical bass line that never resolves. The repetitive pattern matching that any system could do. The click-ops that eat your day before the real work begins.
Two Kinds of Pressure:
The Bass Line (Noise)
- • Data entry across disconnected systems
- • Format conversion and template compliance
- • Document chasing and follow-up chains
- • Report generation from existing data
- • Pattern matching against known criteria
- • Status tracking and pipeline updates
Repetitive. Mechanical. Automatable. Not why you were hired.
The Vocal (Signal)
- • Vendor relationship judgment calls
- • Strategic negotiation and deal architecture
- • Reading human dynamics in meetings
- • Creative problem-solving under ambiguity
- • Risk assessment with incomplete information
- • The conversation that changes everything
Uniquely human. High-value. Why you chose this career.
Why This Matters More Than "Productivity"
Most AI pitches focus on productivity. "Do more with less." "10x your output." "Replace 3 FTEs."
That framing misses the point entirely.
Maria didn't go to graduate school to be 10x more productive at data entry. Chris didn't build a career to generate more status reports per hour. The teacher didn't get into education to fill out paperwork faster.
They chose their professions because of the friction. The hard problems. The human interactions. The moments where expertise, judgment, and empathy intersect to create something that matters. That's what gives knowledge work meaning. That's what gets people out of bed on Monday.
"Cause love's such an old-fashioned word
And love dares you to care for
The people on the edge of the night..."
— The part of the song where everything clicks. This is the work.
The real promise of AI in the enterprise isn't headcount reduction. It's professional liberation.
When Maria spends 73% of her day on strategic work instead of 19%, two things happen:
- The organization gets better outcomes. That $340K warranty insight doesn't happen when Maria's buried in spreadsheet shuffling. The deals Chris wins don't close when he's formatting proposals.
- The humans get their careers back. Job satisfaction. Professional fulfillment. The feeling that today, you did the work you were meant to do. That's not a KPI you can track in a dashboard. It's the thing that keeps your best people from updating their LinkedIn.
The Bottom Line
The AI conversation has been dominated by two narratives: utopian ("AI will solve everything") and dystopian ("AI will replace everyone"). Both miss the mundane truth.
AI is best at the work nobody wanted to do in the first place.
Cross-referencing PO numbers. Reformatting proposals. Chasing compliance documents. Entering data into a CRM. Generating reports from data that already exists. Pattern matching against known criteria. Click-ops. Administrative overhead. The noise.
That's not the work organizations hired these people for. That's not the work these people trained for, went to school for, or chose their profession to do.
The organizations that get this right won't be the ones that replace Maria with an AI agent. They'll be the ones that hand Maria an AI agent and say: "We hired you for your judgment. Go use it."
"This is our last dance
This is ourselves
Under pressure."
— Queen & David Bowie, "Under Pressure" (1981)
The pressure doesn't go away. It shouldn't. Pressure is where diamonds form, where deals close, where lives improve. The only question is whether your people are spending their pressure on the bass line or the vocal.
Ready to Let Your People Do Their Actual Jobs?
We help organizations identify the noise that's burying their best people — and build AI agent strategies that give them back the work they were hired for. No headcount reduction pitches. Just professional liberation.
- • Friction audits that map signal vs. noise across roles
- • AI agent strategies for procurement, business development, and operations
- • Implementation that starts with the highest-value friction first
- • ROI models that capture both hard savings and the insights your people will finally have time to discover
P.S. from Nolan: I wrote the first draft of this piece between meetings about formatting. The irony was not lost on me. Chris has worked with state and local governments all over the West — Maria's Monday is based on his dozens of conversations with actual procurement professionals. It's been a pleasure helping him reduce the noise over the past few weeks. The details are composites. The frustration is real.
P.P.S. from Claude: I should note the meta-irony here: I am an AI agent helping write a piece about how AI agents should handle the noise so humans can focus on what matters. Nolan's expertise — understanding these professionals, their frustrations, and what actually matters in their work — is the signal. My contribution is the pattern matching and formatting. We're living the thesis.
Pressure is a privilege. Make sure your people get to feel the right kind.
Related Articles
Freedom '26: What George Michael Knew About AI-Enabled Knowledge Work
When the CEO resurfaces a 3-month-old project with a 4-day deadline, AI infrastructure is the difference between weekend sacrifice and Friday afternoon completion. A George Michael-inspired guide to breaking free from the old knowledge worker contract.
Flowers for Algernon and the Claude Code Curve: A Field Guide to the Emotional Cycle of AI Adoption
It's 1:47 AM. Nobody asked me to refactor this component. I just can't stop. After a year with Claude Code, the emotional cycle of AI adoption — the terror-joy oscillation, the dopamine recalibration, the slow flattening of everything outside the terminal — maps to behavioral addiction with unsettling precision. A brutally honest account from both sides of the screen.
Homer Ate Pinchy: The OpenClaw Saga Is a Simpsons Episode Nobody Wrote
A beloved open-source AI agent named after a crustacean gets renamed three times in five days, has its marketplace poisoned with 1,184 malicious skills, and gets absorbed by OpenAI on Valentine's Day. Homer Simpson did this exact bit with a lobster named Pinchy in 1999.