How it works
From test strip to exact dose, in seconds
PoolAssist replaces guess-and-add with reading-driven, math-verified dosing. Here’s the full workflow — what we read, what we compute, and where we draw the line and send you to a pro.
Snap or describe your reading
Three input modes, all optional in any combination:
- Test strip photo — Aqua Chek, Taylor pad strip, HTH, etc. We read pH, FC, TC, TA, CYA, hardness from the color pads using Claude vision.
- Equipment photo — salt cell display, Taylor K-2006 reagent kit, pump pressure gauge, water surface (cloudy / green / scaly).
- Plain text — “pH 7.8, FC 0.5, TA 110, 18K gal salt pool” works just as well.
Photos run through a quality pre-check before any AI call — blurry, dark, or angled photos are flagged and you’re asked to retake before vision tokens are spent.
AI analyzes — with your context, not just the numbers
This is where PoolAssist diverges from static calculators like PoolMath. Every analysis combines four signal sources:
- Current reading — extracted from your photo or text.
- Pool config — your gallons, type (chlorine / salt / bromine), equipment, indoor / outdoor. Same pH gets different advice for a 12K gal SWG vs a 30K gal manual chlorine pool.
- Your 30-day history — recurring patterns drive different diagnoses. Three FC crashes in two weeks = CYA degradation, not “add more chlorine.”
- Season + UV — July outdoor = 2-3× the chlorine demand of January indoor; we cite the seasonal adjustment in every recommendation.
Exact dose, scaled to your gallons + verified
Every dose recommendation goes through two layers:
- Linear scale to your pool — not “per 10K gal”; the actual quantity for your pool size, with concentration noted (cal-hypo 73% vs 65% matters).
- Math verification — we re-run the arithmetic against an 18-formula verified dosing table sourced from Trouble Free Pool / PoolMath / manufacturer specs. If Claude’s number is off by >15%, we flag or correct it before showing you. (Most chat-only LLMs get pool dosing math wrong ~10% of the time on multi-step problems.)
Multi-day protocols (e.g., TFP SLAM for green water) come back as a sequenced plan, not a single dose. Each step has a re-test target and timing.
Honest escalation when chemistry exceeds DIY
We say “this needs a pro” — and connect you to one — for any of:
- Safety — suspected gas leaks, electrical-near-water, structural cracks > 1/8″, abnormal heater smell or sound.
- Repeated failure — same issue persisting after 3+ corrective attempts in your history.
- Physical damage — visible cracks, persistent leaks, motor / pump / heater failure.
- Chemistry beyond DIY — extreme imbalance (pH < 6.5 or > 8.5; FC > 20; metal staining), suspected mustard or black algae, persistent green water after a TFP SLAM.
When we escalate, you get a one-click handoff to a vetted local pro on HiredLocalPros — pre-qualified, in your ZIP, with the exact issue category routed correctly.
Try it
Open the chat with one tap
Free for 5 queries / day, no account needed. Sign in to keep history and get smarter recommendations.