Understandable
Plain-language answers first; engineering depth remains available.
Open Thermal AI
Our purpose
Make reliable refrigeration and heat-pump knowledge accessible, give engineering decisions a sound basis, and pursue improvements that can be verified in practice.
Understand the problem. Judge with evidence. Verify the improvement.
On a specific plant, first judge whether the change is worth continuing. When evidence is missing, collect it first.
I started Open Thermal AI to make useful engineering experience easier to understand and share, helping people make better-informed decisions.
What you can expect
- Understand the problem — plain-language guidance and a clear next step.
- Judge with evidence — distinguish facts, calculations, assumptions, and unknowns.
- Verify the improvement — keep forecasts separate from field outcomes.
Our public-interest commitment
Open Thermal AI’s core knowledge, basic project screening, and public calculators will remain free through at least September 2028. We do not sell user data. High-compute-cost interfaces retain reasonable rate limits so public access can remain sustainable.
- Suppliers cannot pay to buy a recommendation or ranking.
- Sponsorships and commercial relationships must be clearly disclosed.
Explore our approach
Why we do this
From industrial refrigeration to heat-pump R&D
Help experience inform the next decision
My work in industrial refrigeration and heat-pump R&D has taught me to value safety, reliable operation, and the experience of people on site. I want those lessons to be easier to understand, check, and pass on, so others can avoid unnecessary trial and error.
Read the founder story and experience record → Read the full diagnosis → Media & citation kit
From learning the principles and collecting site data to deciding whether to proceed, people need reliable information and clear reasoning. We organize knowledge, cases, and methods so newcomers can start, site experience is valued, and project owners can make informed choices.
Every calculation, case and feature should help people understand a problem, reduce waste, and leave useful experience for others.
How we use AI
AI helps organize material, explain problems, and compare options. People retain judgment and responsibility. Confirm extracted inputs and review the calculation basis; an AI explanation is not field evidence or engineering approval.
When evidence is missing, collect it first. When a project should pause, say so clearly. Important conditions and risks belong beside the conclusion.
How we work
Auditable
Calculations, assumptions, missing data, and versions stay visible.
Reliable
Respect safety and process constraints, then assess long-term operation, system efficiency, cost, and maintenance. Screening is not a performance guarantee.
Open
Foundational learning and the first screen should remain broadly accessible.
How we check value
Consider safe and reliable operation, whole-system energy and resource use, user costs, and outcomes that can be checked on site. Evaluate auxiliary loads and the operating period as well as equipment efficiency; disclose what is outside the calculation.
What we will measure
Measure whether people understand a decision and its limits, know what to do next, and can check the outcome. Keep forecasts, owner feedback, and verified measurements separate.
- Useful screenings, including justified decisions to pause; a pause alone does not prove avoided investment loss.
- Projected costs, energy and emissions with comparable boundaries and explicit assumptions; projections are not achieved savings.
- Field outcomes verified later, kept separate from screening estimates.
- Reusable field lessons, including failures and limits, and resources reinvested into open knowledge and tools.
Public impact ledger · 22 Aug 2026
Publish evidence before claiming impact
These counts describe the open evidence base—not customer savings or carbon reductions. Projected and verified outcomes remain separate.
Tested learning templates
Public cases or validated calculator duties; completion assumptions are labeled.
Anonymized field references
Field observations and reported outcomes—not independent metering.
Verified measured outcomes
Kept at zero until consent, boundaries, measurements, and review support the claim.
-
1
Screening projection
A deterministic estimate with visible assumptions.
-
2
Field follow-up
The owner reports what changed after the decision.
-
3
Verified outcome
Comparable boundary, metering period, and evidence review.
How open access continues
Open foundational knowledge and tools, together with transparent, independent professional services, support steady progress and help more people benefit from refrigeration and heat-pump technology.
Open foundation
Free to understand and start
Beginner knowledge, the quick estimate, selected cases, and data checklists remain available without a paywall.
Professional depth
Paid when accountability and delivery deepen
Expert review, scenario comparison, integration studies, and decision packages are scoped professional work.
Enterprise collaboration
Sustain the platform through partnerships
Private deployment, training, APIs, and joint R&D fund continued improvement of the open foundation.
Commercial income supports the maintenance of free knowledge, tools, and case preparation.
Editorial & sources
Open Thermal AI (开热智) is an industrial heat-pump go/no-go screening platform plus a bilingual research and knowledge hub—an open place to accumulate reusable engineering experience for refrigeration and heat pumps.
Update cadence
Industry briefings are manually curated and published weekly. Knowledge chapters, standards maps, and Annex notes are revised when sources or practice change—not on a fixed weekly clock.
Source principles
- Prefer primary links (regulators, IEA HPT TCP, standards bodies, named company filings).
- Annex and services pages translate and distill Task reports; the authoritative text remains on heatpumpingtechnologies.org and the cited PDFs.
- Cases declare a data status (illustrative, public-source, verified, or anonymized field reference)—never invent measured numbers.
- Site content is engineering orientation, not legal, tax, or compliance advice.
Start here: Research · Standards & policies · Services · Latest briefing.
How we build and update the site
Jing Yanrong maintains Open Thermal AI with engineering experience and AI-assisted development. Sources, calculations and field evidence have separate roles.
AI assists the work
ChatGPT and Cursor assist website design, programming, information organisation and testing. Using these tools is not an endorsement, certification or partnership with their providers. Development tools are distinct from any AI service a visitor may invoke.
Evidence supports the judgment
Read primary publications and their dates and scope. Research, supplier claims, calculations, plant-reported results and independently verified outcomes remain separately labelled. A citation is not an institutional endorsement.
From new information to an actual improvement
Discover → check the original source → assess who and what it affects → update the relevant guide or tool → validate and record. Industry briefings follow the existing editorial process; source changes trigger targeted revisions. This is an editorial method, not a claim of unattended automatic monitoring.
Adopt improvements after evaluation
A newer model is not automatically better for an engineering task. Before adopting changes, check units, missing inputs, source citations, calculation boundaries and both languages. AI explanations do not turn assumptions into measurements or replace qualified review.
Sources and development tools: OpenAI brand guidance · Cursor · HPT Project 68. See our briefings and guides and data commitments.