Built by Experts, for Experts.
We're analytical scientists who know a lot, about a lot, with over 15 years of instrumentation, laboratory, and field expertise. We build AI agents grounded in truth, and peer-reviewed literature, not chatbots trained on the open internet.
Let us show you what an AI agent built on your knowledge base can do.
Every KB is built from published literature and validated by working scientists. That's why the answers hold up to scrutiny.
Your logo, your colors, your domain. Customers see your brand — not ours. 12 configurable branding fields, custom system prompts, and full CSS control.
Every answer is grounded in verified documentation and published literature — not the open internet. FAISS vector retrieval + your choice of LLM (Grok, Claude, OpenAI) = accurate, citable responses.
We handle everything — Docker containers, auto-SSL, analytics, user auth, feedback capture. You focus on your expertise; we turn it into a production AI agent.
From docs to live AI agent in four steps.
Markdown, PDF, or wiki exports. We structure and index everything automatically.
Set your company name, logo, colors, and custom domain. Your agent, your identity.
Grok (cheapest), Claude (smartest), or OpenAI (most popular). Switch anytime.
Your AI support agent is deployed at your custom domain with SSL. Start helping customers.
Same question, two very different answers.
"I'm seeing high F counts in my metal sample in EPMA — what's happening?"
High F counts in a metal sample are usually not real fluorine. The most common causes are:
1. Surface contamination — polishing residues, mounting materials, cleaning chemicals…
2. Overlap / background problems — F Kα is very low energy, vulnerable to poor background placement…
3. Secondary fluorescence — near inclusions, pores, epoxy, or grain boundaries…
4. Carbon coat issues — uneven coating, oxidation, scratches…
5. Instrument setup — incorrect crystal choice, poor peak centering…
Generic checklist. Doesn't identify the actual cause or give diagnostic steps for your software.
High fluorine counts in a metal are often due to spectral interference — specifically, overlap between F Kα (0.677 keV) and Fe Lα (0.705 keV), which mimics high F signals when iron is in the matrix.
To diagnose: run a wavescan in your EPMA software to visually confirm the peak. Then use Standard → Misc Options → Interferences → Calculate Nominal Interferences, enter your sample composition, and review the overlap report.
If interference is confirmed, apply corrections or select a different analytical line. Otherwise, consider contamination from polishing compounds and verify your standards.
Sources: spectral_interference.md, quantitative_analysis.md, standards_setup.md
The expert agent identifies the specific physical cause (Fe Lα / F Kα overlap at known energies), gives exact menu paths in your software, and cites its sources. Generic AI gives you a study-guide checklist.
Try the Live Demo →Electron microprobe analysis (EPMA) — 2 knowledge bases built from published literature and decades of hands-on CalcZAF experience. Multi-LLM, with cited sources on every response.
Scanning Electron Microscopy — a complete textbook transformed into an expert AI agent. 99 KB articles covering imaging, sample prep, detector physics, and analytical techniques.
We'll build a knowledge base from your literature and documentation, then deploy an AI agent that cites its sources on every answer.
Or email us directly at hello@kbforge.com