Est. 1910
Pfirsthand
116 years in. Back in the pfield.
We flip the travel spend, because AI flipped the world, so we put boots in the field, build the trade an app, and prove it on a five-year plan.
Follow the line ↓01 — The Flip / where the money goes
Same spend.
Different result.
Today, product's time and travel flow to supplier oversight in China. We design around what the factory can build, and hear the trade third-hand — through sales reps and warranty claims. Pfirsthand redirects that exact spend into North American field immersion. The whole product team, in person, embedded with plumbers, distributors, and jobsites.
Who watches the factory? Professional inspectors on every run — dedicated supplier-quality ownership, escalation-only for product. Coverage goes up, not down.
02 — Why Now / the moment
AI made
information free.
Every competitor has the same models, the same research, the same factories. When knowing gets cheap, value moves to what's still scarce. Three things can't be scraped:
Relationships
A plumber's trust is slow, physical, and non-transferable. No API returns it. You earn it at the counter and keep it on the jobsite.
Brand equity
116 years, one handshake at a time. The pfunny name plumbers still sing. It cannot be generated — only compounded.
Jobsite truth
How parts really fail, install by install — the data that was never written down. It exists nowhere online. You have to be there.
Photos blocked? You're in a sandboxed preview. On the live site — or this file opened in a browser — the photography loads.
"Anyone can buy from the same factories.
Nobody can out-know the trade."
Being physical isn't the disadvantage anymore. In the AI era, it's the only advantage left.
03 — The Pfield / the engine
Six rituals.
One calendar.
Full product team, in person, rotating in pods so HQ is never dark — across Sun Belt, cold-climate, and coastal waves, because Phoenix hard water and Minneapolis freezes fail differently.
Photos blocked? You're in a sandboxed preview. On the live site — or this file opened in a browser — the photography loads.
Every handshake becomes data.
Capture
Sixty-second mobile form. Photo, voice note, tags. Same day — never "written up later."
Cluster
AI groups thousands of entries into themes across every city. Our own thesis, eating its own dog food.
Rank
Frequency × severity × addressability. Published monthly as Pfield Notes, straight to leadership.
Build
Top themes become product briefs. Governance rule: no brief without field receipts.
"A plumber told me" becomes evidence. That's what protects this from ever being called a road trip.
04 — The Research / the evidence
Plumbers aren't avoiding AI.
They're waiting to be taught.
The barriers are time, cost, and training — not fear. Trust flips through hands-on experience. Our field team delivers exactly that, in person.
05 — Pfix / the app
AI for the
plumber's hands.
They built AI to run the plumber's office. We build it for the plumber's hands — and it ships with the one thing software can't buy: a field team the trade already trusts, putting it in their palm at a cookout.
Point. Know. Pfix.
Photo any part — old, corroded, unlabeled — and get the ID, the exact replacement, and the fix. The "never get stumped again" moment that flips a skeptic in ninety seconds.
Stocked nearby
Live availability at the local distributor branch — reserved for pickup. Through the channel, never around it. Distributors become our loudest evangelists.
Emil, the assistant
A master plumber in every pocket — born 1910. Trained on our catalog, our specs, and Pfield data no competitor's model has ever seen.
Human one tap away
AI first; live video expert when it's stumped. The escape hatch is the adoption strategy — and every call trains Emil on the hardest edge cases.
Estimator
Photo the job, get the takeoff and materials list, priced. The make-them-money feature — where Pfix graduates from helpful to how I run jobs.
Warranty in 40 seconds
Photo, serial, auto-filed claim, live status. A universally hated paper process, deleted.
Free to the trade. Paid four ways.
Pull-through
Every scan, chat, and estimate resolves to our SKU. A spec machine in 100,000 pockets.
Channel velocity
Pfix routes plumbers to distributor counters. Our line becomes the one they feature.
Loyalty
Points on scanned installs. Status tiers. Switching costs, gamified.
The data asset
An installed-base census of the category — including competitor parts. Nobody else has one.
Charging $9.99 a month would strangle all four. The strategic value dwarfs the subscription.
06 — William / the teardown agent
Emil serves the trade.
William watches the market.
Two founders, two agents. William runs continuously on foundation models, ingesting everything competitors publish — then does the one thing no rival can: cross-references it against Pfield data to find the gaps nobody is filling.
William watches
- Competitor pricing pages every move, every SKU, tracked to the day
- Changelogs & launches new products and quiet spec revisions
- User reviews — retail & pro clustered by complaint, ranked by volume
- Social, spec sheets, warranty terms the full public surface of the category
William delivers
- The monthly Gap Report folded into Pfield Notes, straight to the roadmap
- Opportunity briefs with receipts field evidence attached to every claim
- Pricing & launch alerts competitor moves surfaced the same week
- The whitespace map what the trade asks for that nobody builds
Public data × Pfield data = The Gap Map
Competitor data shows where everyone is standing. Field data shows where the trade is hurting. Overlay them and the gaps light up. Anyone can scrape the first half — the second half only exists in our dataset. The fusion is the moat.
Prototype in 90 days · ~$60K year one07 — The Flywheel / how it compounds
One loop.
Every turn easier.
Not five ideas — one loop. The physical program is the ignition; Pfix is the engine; Emil and William are the intelligence.
A competitor can copy any single activity. They can't start the wheel from zero and catch five years of spin.
08 — The Numbers / the spend
Priced like a launch.
Returns a platform.
| Line | Pilot · Year 1 | At scale · Year 2–3 |
|---|---|---|
| Pfield engine — blitzes, pfeasts, council, shows | $160K | $365K |
| Third-party QC — replaces our factory trips | $50K | $70K |
| Pfix — prototype to public MVP | $500K | $2.0M |
| William — competitor teardown agent | $60K | $100K |
| Less: redirected China travel | ($200K) | ($200K) |
| Net new investment | ≈ $570K | ≈ $2.3M |
The field layer runs near cost-neutral on the redirect — quality coverage improves while it does. The software line is the growth investment: free to the trade, paid back four ways, building the moat as it goes.
09 — The Return / the payback
The return on
being there.
Three value streams, one proof method — in the products we ship, the data we own, and the preference we earn in every room where our brand gets chosen.
The right products
Field receipts de-risk the front end: fewer dud launches, faster insight-to-brief, redesigns aimed at real failure modes. Every avoided miss returns its tooling; every field-sourced winner compounds.
Trained data
Every scan, chat, and claim trains Emil and feeds William — an installed-base census nobody else owns. An asset that appreciates instead of depreciating.
Goodwill, three rooms
Plumbers spec us on purpose. Distributors feature the line that fills their counter. Retail wins on ratings, pro-desk pull, and deflected no-fault returns.
| Value driver | Mechanism | Conservative annual · Yr 2–3 |
|---|---|---|
| Field-informed launches | 2 avoided misses + 1–2 incremental winners | $1.5–3.0M |
| Warranty & returns | −0.25 to −0.5 pt on redesigned lines | $0.8–1.5M |
| Pfix pull-through | Scans and estimates resolving to our SKU | $0.7–1.5M |
| William gap win | One whitespace entry landed per year | $1.0–2.0M |
| Retail lift | Ratings, pro-desk pull, deflected "no-fault" returns | $0.5–1.0M |
| Research replaced | Owned field data vs. syndicated panels | $0.2–0.3M |
| Total, conservative | vs. ≈$2.3M net program spend | ≈ $4.7–9.3M / yr |
Call it 2–4× annually — before the compounding. Illustrative model sized to a ~$300M line; plug the real baselines and it hardens fast. Costs stay flat while every stream grows.
How we prove it: blitzed markets vs. untouched control markets — a built-in natural experiment. The scoreboard ships monthly, inside Pfield Notes. No faith required.
10 — The Plan / five years
Prove. Scale.
Own. Compound.
Prove
Pilot on the redirect. QC handoff signed. Pfield capture and monthly Pfield Notes live from month one. Pfix prototype rides along on every blitz. William v1 shipping Gap Reports.
Scale
Full cadence — eight blitzes, twenty-plus pfeasts, every key show, full Pfellowship. Pfix launches publicly at the blitzes. Blitzed markets vs. control: the natural experiment reports.
Own
Field-first is just how product works. Seventy percent of the roadmap carries field receipts. Warranty and returns fall on redesigned lines. The program stops needing a defender.
Compound
The installed-base census becomes the substrate for AI no competitor can train. Trade preference shows up as measurable revenue. The model exports to Canada, Mexico, adjacent lines.
Years 1–2 buy better products. Years 3–5 build the moat. The patience is the strategy.
The Decision / on the table
The travel budget we already spend.
A small squad to build Pfix.
The autonomy to run the pfield.
In return: the most trusted name in the trade.
Year One
They can copy the activity.
They can't copy the years.
RESEARCH GROUNDING: Housecall Pro AI-in-the-trades industry surveys (2025–2026, 400+ home-service contractors); Edelman global AI trust research; NFPA skilled-trades technology survey; CNN Business trades reporting; BLS and industry workforce analyses on the plumber shortage and retirement wave. Figures are directional and cited in full in the working brief.