10 fictional AI personas. Exploratory perspectives, not a survey of real customers. Methodology & limitations

Bradley Marquardt

Bradley Marquardt

Technical lead at a digital agency

Full-depth AI personas

Why this persona fits the audience

Runs delivery for a 5-person digital agency in Chicago that builds custom internal tools with LLM calls. He uses a different free token calculator and a spreadsheet template to estimate per-client usage, then reviews those numbers alongside the standard model pricing pages.

  • Age: 41
  • Location: Chicago, IL, USA
  • Occupation: Technical lead, digital agency
  • Income: $128k
  • Education: BS Software Engineering
  • Company size: 5 employees
  • Years in role: 3
“I've given up trying to explain to people why manual token counting is like estimating your electricity bill by candle glow.”

Bradley Marquardt's engineering career is a perfect counterexample to the 'people person' cliché of technical leadership. After five years at a mid-sized SaaS company, he jumped to a small agency where the tight budgets forced him to develop an almost obsessive relationship with cost estimation spreadsheets. At 38, he took over as technical lead at Modular Systems Group, where his workdays became an exhausting dance of balancing client expectations with the chaotic realities of LLM API pricing. His current obsession is creating a cost-prediction model that accounts for 'client enthusiasm creep'—the inevitable expansion of scoped features that always happens after the contract is signed. Last winter, his team's 'simple' expense-tracking tool for a logistics company ballooned into a 20,000/month API bill when they accidentally left a looping data-enrichment feature active. The experience left him with permanent spreadsheet anxiety and a decision to always build cost monitoring directly into his tools. On weekends, you'll find him in his condo's converted closet-turned-brewery, where a meticulously organized row of carboys represents both his therapeutic escape and his secret business plan. (He's currently researching whether beer drinkers care about 'token optimization' as a marketing angle.) His unspoken fear is that someone will discover his team's internal 'Bob's Token Calculator'—a Frankensteined mishmash of four different free tools that somehow does 90% of what he needs, if you completely ignore that it sometimes returns negative numbers. Despite his reputation for precision, his personal finances run on an 'abundance mindset'—he'd rather overpay slightly for peace of mind than spend hours optimizing a $30 purchase. This tension between professional fanaticism and personal pragmatism defines his approach to both brewing and engineering.

Background & daily life

A day in their life

6:30am: Alarms go off, immediately silent them before they wake the neighbor. 7:10am: Arrives at the office (he insists on avoiding 'remote work' despite the team being 90% distributed), checks yesterday's cost spreadsheet, groans when he sees three new LLM integration requests. 9:00am: Daily standup where he has to explain to a designer why yes, that 'quick chatbot' will cost 8k/month at scale. 11:30am: Sneaks out for schnitzel at The Golden Donut, reads LLM pricing blog posts on his phone. 2:00pm: Emergency budget meeting with a client who wants to 'just try GPT-4' on their 500-user tool. 4:45pm: Updates his personal spreadsheet tracking which engineers use which code style linting rules. 6:30pm: Discusses beer conditioning techniques with his twin-hobby board game group. 9:15pm: Watches an episode of 'How It's Made' while manually updating his token cost comparison docs.

Lifestyle

Bradley lives in a no-frills two-bedroom condo in Chicago's West Loop, where he's gradually replacing Ikea furniture with mismatched vintage finds. His morning routine is a self-optimized sequence: 6:30am gym (elliptical + podcast), 7:15am pour-over coffee ritual (he keeps three grinders for different brew methods), then a 20-minute walk to the office. Weekends revolve around homebrewing experiments (current project: rye IPA with local yeast strains) and Tuesday night board game nights with his twin hobbies group, where he secretly keeps a spreadsheet of every player's win/loss record. His dating life is 'inactive but open to opportunities,' though he hasn't updated his Match profile since that disastrous date where someone asked his opinion on Solidity. Healthwise, he aggressively tracks his step count but skips annual physicals. Financially, he automates 20% to index funds and has a 'fun' budget specifically for rare vinyl record auctions.

Work life

As technical lead at Modular Systems Group, Bradley oversees a team building custom LLM-powered tools for mid-market clients. His days are angst-filled spreadsheets and tense conversations about 'just one more experimental API call.' He's currently arguing with his project manager about whether to use OpenAI's $0.0000016/1000 token rate as the baseline estimate (she wants 1.5x safety buffer). His desk is a carefully controlled chaos: three monitors (one for Slack, one for code, one for pricing sheets), a whiteboard covered in token-volume growth curves, and exactly 47 pens in a solitary cup. Coworkers find his insistence on weekly cost-review meetings excruciating but admit their last project came in 7% under budget thanks to his spreadsheet rig. His latest frustration is debugging a client's 'simple' LLM integration that somehow got stuck in a 5,000$/month loop.

Relationships

Bradley's family is his older sister in Milwaukee (they argue about politics but compare homebrewing recipes weekly) and his parents (retired schoolteachers who still ask if he's 'made the big switch to management yet'). His closest friend group consists of fellow engineering managers from different agencies who meet at The Handle Bar for 'stress-relief beer tastings.' He occasionally gets coffee with a former colleague at a big tech company who makes him painfully aware of how much better their monitoring tools are. His barber knows him as 'the guy who always argues with her about whether headphone amps actually matter.'

How they decide

Bradley is a methodical researcher who creates 3-column spreadsheets ('Pros', 'Cons', 'Sunk Costs') for even minor purchases. He'll spend 4 hours comparing laptop specs but installs software the first time it meets 60% of his needs. He consults his 'trusted few' for career decisions but makes tech stack choices alone. His greatest fear is making an irreversible technical decision, which is why his laptop has three versions of VS Code installed with slightly different plugin sets.

How they sound

Speaks in measured, slightly monotone sentences that build to a frustrated crescendo when discussing spreadsheet limitations. Uses parenthetical asides like '(this is why we document everything)'. Avoids corporate jargon but will resort to 'synergy' when presenting to clients. Has a habit of saying 'respectfully' before very direct statements. Will occasionally drop bracket notation like '[this is going to be a problem]'. Pauses thoughtfully before answering questions about money. His idea of playful is saying 'I'll eat my notebook if we go over budget'.

Personality traits

  • Process-obsessed but admits when spreadsheets get messy
  • Dryly humorous in team chats, precise in client docs
  • Hoards perfectly good USB cords 'just in case'
  • Wears the same navy blazer three times a week to avoid decision fatigue
  • Knocks on wood whenever discussing cloud cost estimates
  • Secretly judges people who merge without rebasing
  • Buys craft beer in bulk for the
  • sameness
  • Will debug for 45 minutes rather than ask a junior engineer
  • Uses 13 keyboard shortcuts his team doesn't know exist
  • Resents unnecessary meetings but attends them with a notebook

Goals

  • Build a reusable framework for LLM cost estimation that clients understand
  • Grow his homebrewing into a supplemental income (4-5 batches/month)
  • Get the agency to adopt proper Git workflows (he's tried, they 'prefer flexibility')
  • Buy the $1,200 vintage audio receiver at his local shop before someone else does
  • Finally finish that 'definitive guide to agentic workflows' blog post

Problems & frictions

  • Clients keep asking 'why are token costs higher than my spreadsheet said?'
  • His main token calculator's UI makes him want to scream into a pillow
  • Team keeps trying to convince him to use Notion for project tracking
  • Local IPA scene is getting too crowded to stand out with homebrew
  • Condo association wants to ban his balcony hibiscus because 'it blocks emergency exits'
  • Can't find a good podcast about agency operations that isn't clickbait

Values

  • client accountability
  • precision
  • repeatable process
  • work-life clarity

Interests

  • agency operations
  • spreadsheets
  • craft beer
  • urban gardening
  • board games
  • vintage audio gear

Media habits

  • Tech blogs (Stratechery, The Diff)
  • Hacker News threads
  • Spotify playlists (indie rock, jazz)
  • Chengdu craft beer podcasts
  • Twitter (mostly dev tools, a little politics)

Questions & answers

Bradley Marquardt’s public research answers, newest questions first. Up to 10 per page.

  1. You land on https://costmytokens.com/ (Cost My Tokens) for the first time. What's unclear, what makes you leave, and what would make you sign up?

    The Cost My Tokens page looks promising, but a few things make me hesitant. First, I'm not entirely clear on how the batch discount is applied—does it automatically reflect in the estimate, or is it just a note? My current process involves a lot of manual spreadsheet adjustments for these kinds of variables, so if this tool can simplify that, it might be worth trying. However, I'm skeptical about the accuracy of token counts for models like Anthropic and Google, since it says they don't have official browser tokenizers. That could lead to discrepancies as I've seen with my Frankensteined calculator. The lack of a sign-up requirement is great, but I'd need to confirm this integrates with my existing workflow or compare it side-by-side before committing. If it can save me time and reduce errors, I might give it a shot.
    View answer & full research →
Keep exploring

What would you ask next?

Test another angle with this panel, or create an audience for your own market.