Water intelligence · multi-agent AI
A vertical panel of expert AI agents for water and aquaculture — routed, argued in parallel, and citation-verified before a word reaches you.
A panel of expert agents that would rather escalate than invent. When the stakes are high, a human expert signs off — by design, not as a fallback.
Below — the register flips to specimen paper. This is where it proves.
The grounding verifier
The dark is where the agents argue. The page brightens to paper the instant a claim has to stand on evidence: each one is checked against Crossref / DataCite, and anything unresolved is escalated, never guessed.
“Nitrite accumulation points to stalled second-stage nitrification, not an oxygen deficit.”
Drawn by the microbiology specialist, then grounded against the literature:
10.1111/jfd.13211Verified · resolves at publisherWhat the gate does
Proof as data-viz
What changed, how strong the evidence was, and who signed it. This is the shape of that proof; the first anonymized cases publish soon.
Textile effluent screen: a 7.2× COD exceedance and a REJECT call, human-gated (illustrative)
A textile dyehouse asked whether its trade effluent clears Singapore PUB limits and whether to green-light a go-ahead. AquaMind V2 flagged three confirmed exceedances — COD at 7.2× the limit, TSS over 3×, and out-of-range pH — plus a wide block of undeclared metals, and returned a NON-COMPLIANT posture with a REJECT recommendation, mandatorily held for human sign-off as a food-safety/compliance call. Illustrative compliance screen on a fictional facility, not a legal determination.
Indoor RAS discharge screen: seven parameters, a PROCEED call, human-gated (illustrative)
A 600 t/yr indoor tilapia RAS asked a two-part question — does the effluent meet Singapore PUB Trade Effluent limits, and does the ROI support a go-ahead? AquaMind V2 screened all seven declared parameters as passing, computed the investment economics, and returned a PROCEED recommendation — flagged tier-5 and held for mandatory human sign-off before any business decision. Illustrative compliance screen on a fictional facility, not a legal determination.
Greenfield tilapia RAS: a sized, costed RAS + MBBR design in one pass (illustrative)
A greenfield 55 t/yr indoor tilapia brief — 550 m³, 110 kg/day feed, grid power, municipal discharge — went in; a route recommendation, unit sizing, a costed bill of materials and internal consistency checks came out. AquaMind V2 selected RAS + MBBR, sized the biofilter, aeration, pump and tankage from the computed nitrogen and oxygen loads, and self-verified that every sized unit maps to a BOM line. Illustrative sample on a fictional facility.
What makes it different
Microbiology is the moat — WaterDoctor’s SND functional microbial engine, a capability no horizontal chatbot has.
A panel of discipline specialists, deliberating in parallel around a Virtual Round Table.
DOI-strict grounding — every load-bearing claim checked against real literature.
High-stakes calls route to a human expert for sign-off. Skip-to-human is a feature.
Two verticals · one engine
The same pipeline runs any water discipline. Aquaculture diagnosis is live to real operators; the industrial-water vertical is live too, with compliance kept human-in-the-loop.
RAS, high-density and green-seafood shrimp & fish — water chemistry, disease, microbiology, nutrition and operations, in production today.
Effluent diagnosis, treatment design and Singapore PUB / WEMP compliance for industrial dischargers — live today. Compliance runs with a human in the loop; treatment design is directional, not a final quotation.
Modules
Built bilingual · EN / 中文
Not English-premium and Chinese-fallback. The display serif is matched by a Noto Serif SC face, so a Chinese headline reads as journal-grade too. Reports and SOPs transcreate on demand, with a fidelity guard that keeps DOIs, numbers and doses intact.
Every high-stakes call is grounded in the literature and signed by a human expert.
expert-signed每一个高风险判断都以文献为依据,并由人类专家签署确认。
专家签署
Founder
Dr Wang Chuansheng — Founder · NUS PhD, Environmental Engineering
Aquaculture’s hardest problems are microbiology problems.
Dr Wang Chuansheng founded WaterDoctor to bring water-quality and microbial expertise to aquaculture operators. NUS PhD in Environmental Engineering, MSc from Peking University, and the inventor of the SND functional microbial engine that anchors WaterDoctor’s products — recognized with the IWA Best PhD Thesis Award 2024. AquaMind V2 packages that work into a system that reads your case and answers like a senior aquaculturist would.
