Five slides drafted from the Wastepoint and Chris Young calls (May 18, 2026) plus Philip Galland's two infographics. These are intended to be reviewed standalone and then manually copied or rebuilt into the live deck. Nothing here overrides the existing 19-slide deck at element-iq-deck.pages.dev.
ElementIQ is designed to support the broader critical mineral and material value network. The initial commercial focus is sourcing intelligence for material sourcing professionals and resource owners, where supply-chain risk, procurement urgency, and budget availability are most immediate.
ElementIQ identifies viable products, customers, and processing pathways for the material a resource owner is sitting on. The platform turns disposal cost into a revenue stream.
ElementIQ supports flowsheet evaluation, feedstock-to-product matching, and offtake pathway analysis for the conversion of conventional and unconventional mineral resources. ElementUSA Inc. is itself a midstream processor.
The platform matches technology owners' proprietary processes with feedstocks that can scale. This opens up licensing revenue and lowers scale-up costs.
ElementIQ helps material sourcing professionals find domestic supply sources that reduce supply-chain risk and procurement volatility. Initial commercial focus.
ElementIQ enables strategic partners to run joint pilot testing, flowsheet co-development, and commercialization pathway design. Both sides share the development risk.
ElementIQ supports government and defense work on domestic supply chains and industrial waste reuse. The work serves national security and supply-chain resilience.
ElementIQ produces every answer through a defined workflow rather than a single prompt. Each question is broken into structured steps that draw on curated industry knowledge and verified data sources. The same workflow runs every time, which keeps outputs traceable and reviewable across teams.
The platform receives a defined objective along with the relevant context for the decision being made.
Each question is broken into sub-tasks. The platform routes those sub-tasks to the relevant knowledge modules and tools.
Curated modules supply the methods, data, and decision rules that apply to the question. Modules are peer-reviewed and source-tracked.
Models and tools execute within the boundaries set by the retrieved knowledge. The platform controls what the language model produces and how.
Every answer comes with provenance and confidence indicators. Outputs are delivered as supply maps, briefs, flowcharts, and reports.
Tools like ChatGPT are smart generalists. They guess at the answer based on patterns they learned from public information. Ask the same question twice and you may get two different answers, and you usually can't see why. That doesn't work for industrial decisions, where the same question needs to give the same answer and where teams need to be able to check the work.
| Dimension | Generic LLM ChatGPT · Claude · Perplexity |
LLM + RAG Retrieval-augmented generation |
ElementIQ Specialized industrial intelligence |
|---|---|---|---|
| Intelligence type | Predicts the next likely word. | Predicts from retrieved chunks of text. | Produces answers using curated industry knowledge. |
| Transparency | Black box. | Shows which chunks were retrieved. | Every answer is source-backed and traceable. |
| Knowledge source | Public training data only. | Public training data plus your documents. | Curated industry knowledge plus ElementUSA operating data. |
| Memory | Session only. | Session only. | Persistent across projects, users, and time. |
| Audit & ownership | No audit trail. | No audit trail. | Audit trail across every step. |
| Output | Text only. | Text with citations. | Supply maps, briefs, flowcharts, reports. |
| Improvement over time | No domain learning. | No feedback loop. | Learns from corrections and outcomes over time. |
| Cost at scale | Token cost rises with usage. | Token plus vector-search cost per query. | Per-query cost stays roughly flat. |
Note: ElementIQ is designed for industrial workflows where the same question should produce the same answer and every output can be reviewed against its sources.
The language model itself is not what makes ElementIQ work for industrial questions. The platform is built around curated industry knowledge developed by domain experts and captured in structured modules that the AI applies to specific questions. Much of this knowledge sits with retiring engineers and operating specialists, and is not represented in public training data.
A structured representation of an expert's domain knowledge, including methods, decision rules, and reference data. Modules are peer-reviewed before being added to the platform, and they work with any underlying language model.
Recovery chemistry, flowsheet design, processing economics, and supply-chain risk patterns are domains where the foundational knowledge is held by working engineers and senior researchers rather than published in public sources. Curated modules make that knowledge available to the platform.
[N modules in production] covering [domains, e.g. recovery chemistry, flowsheet design, sourcing intelligence], developed with [named expert advisors]. Coverage expands across gallium, germanium, scandium, rare earth elements, and lithium during the seed-funded build-out.
With most AI tools, the more questions you ask, the more it costs, and the answers stay best guesses. For a handful of questions a day that barely matters. But when these tools get embedded into internal team workflows and run thousands of queries a day, the cost adds up quickly and the inconsistent answers become a real liability.
ElementIQ is built differently. Each question is routed directly to the right specific knowledge instead of searching everything at once, so the cost per question stays steady and the answers stay consistent at any scale. That makes the platform practical to plug into the daily tools sourcing, engineering, and procurement teams already use.
Because the platform retrieves the same knowledge in the same order, the same question produces the same answer across users and queries. That is what makes outputs reviewable in regulated workflows.
Every answer is tied back to the specific knowledge modules and data sources used to produce it. Teams can validate findings against the underlying material before acting on them.
Because each question is routed directly to the relevant knowledge module, the platform does not need to search across an entire vector index on every query. That keeps the cost profile relatively stable as usage and concurrency rise.