reasonance lab

Dr. Emil Jivishov

Software for
understanding.

I build practical tools for high school learning, scientific exploration, and robotics. Explain.az turns students’ answers to ordinary open-ended questions into visual feedback they can explore as they refine their responses.

Explore Explain.az Get in touch

High school science · Student reasoning · Teacher-led assessment

A LOOK INSIDE THE WORK / 01
An open-ended question leads to a student's written answer, then an interactive simulation of that explanation A FAMILIAR QUESTION. YOUR ANSWER AS VISUAL FEEDBACK. 01 / QUESTION How does a push change a ball's motion? Explain in your words. 02 / YOUR ANSWER A stronger push makes the ball speed up faster. Student reasoning 03 / FEEDBACK Change the push See what your answer says. Reflect. Refine it. ANSWER → SEE → REFLECT → REVISE

Explain.az

Visual feedback from the student’s own answer

Workflow illustrationsExplore the live projects below

Across three connected fields

AI & learning Scientific software Educational robotics

01 / Selected work

Ideas you can explore.

Start with high school learning. Explore Explain.az, its teacher plugin, and the science and robotics tools below.

01 / AI & learning
An open-ended question leads to a student's written answer, then an interactive simulation of that explanation A FAMILIAR QUESTION. YOUR ANSWER AS VISUAL FEEDBACK. 01 / QUESTION How does a push change a ball's motion? Explain in your words. 02 / YOUR ANSWER A stronger push makes the ball speed up faster. Student reasoning 03 / FEEDBACK Change the push See what your answer says. Reflect. Refine it. ANSWER → SEE → REFLECT → REVISE Workflow illustration
High school learningLive app

Explain.az

Answer a familiar question.
See your thinking made visible.

An original approach to assessment feedback. Students answer ordinary open-ended questions. Explain visualizes their answer as an interactive simulation, giving them feedback they can use to reflect on their understanding and refine the response. How it works

Explain plugin for ChatGPT & Codex. Teachers draft assessments, refine rubrics, and review student work in conversation, then save the results they approve.

Explore Explain.azExplain plugin
Project notes

The assessment task is to answer the original subject question. Explain uses that written response to generate visual feedback. Students inspect what their answer communicates, reconsider their reasoning, and refine the answer. The simulation is generated by the system. In the OpenAI workflow, GPT Image 2.5 Flare first generates an image, then an OpenAI language model generates the HTML, CSS, and JavaScript that run the simulation on the page.

The Explain plugin connects a teacher’s account to ChatGPT or Codex for assessment drafts, rubric revisions, and review of submitted evidence. Teachers approve saved changes and publish grades from Explain. Plugin setup guide · View source

02 / Scientific software
Illustration of a titration setup with a burette, flask, and teacher review step Set up.Stage the inventory Rehearse.Explore the procedure 01 / PLAN PLAN → REHEARSE → TEACHER REVIEWWorkflow illustration
Scientific softwareLive demo

Lab Studio WebMCP

A rehearsal space for the chemistry lab.

Plan and rehearse a supported titration workflow, with an AI agent assisting and the teacher making the final decisions.

React · TypeScript · WebMCP

Explore demoView code
Project notes

Inspect the inventory, stage a pre-lab, rehearse the procedure, and run protocol checks. An agent can assist through WebMCP, while the teacher retains the final Apply or Discard decision. The supported activity is a titration workflow.

03 / Educational robotics
Illustration of a robot arm in a browser physics workspace xyz Build. Program.Observe. PYTHON → MUJOCO → OBSERVATIONWorkflow illustration
Educational roboticsPhysics preview

RoboBuddy IDE

Build a workcell.
Program, simulate, and observe.

Write Python, build OpenArm laboratory scenes, and inspect simulated motion and contacts in MuJoCo. Opt-in WebMCP lets an AI agent help build scenes and programs.

Python · MuJoCo · WebMCP

Explore RoboBuddyView code
Project notes

Six robot workspaces: OpenArm, SO-101, LeKiwi, Asimov 1, MicroDuck, and Unitree G1. OpenArm supports two-arm manipulation and laboratory scene building with rotated parts, hollow profiles, and validated meshes. Supported experiments vary by robot.

A multimodal agent can use a reference photo to help describe new equipment. The app validates the submitted geometry; dimensions remain explicit estimates. Preview and approve a scene before applying it.

Opt-in WebMCP lets an agent inspect telemetry, build scenes, and help draft programs in a compatible host. You retain review, Stop, and access controls. This is a physics preview; simulated outcomes are not hardware validation.

OpenArm manual · WebMCP & photo guide

Explain.az / Assessment feedback

Familiar questions.
Visual feedback.

An original approach to open-ended assessment feedback.

Students answer ordinary science questions, including the kinds familiar from AP Chemistry and AP Environmental Science. Explain generates an interactive visualization from the answer itself, so the student can inspect what their response communicates.

The simulation serves as feedback on the response. Students use it to reflect on the correctness and completeness of their explanation, then refine their answer. The aim is deeper comprehension through reflection and revision.

How OpenAI turns the answer into a simulation

  1. Generate the image. OpenAI’s GPT Image 2.5 Flare creates a visual draft from the student’s answer.
  2. Generate the interactive code. An OpenAI language model uses the image as a visual guide and the student’s answer as the content source to generate HTML, CSS, and JavaScript. The code runs the interactive simulation directly on the assessment page.

How the idea began. The idea grew out of teaching my 13-year-old son to use Codex to build a simulation of his immunology knowledge.

02 / The collection

More ways into the work.

The full project index, from classroom tools to experiments still taking shape.

Showing 13 of 13

Projects are at different stages of development. AssayLens provides exploratory image-derived measurements, not calibrated absorbance or validated MIC values.

03 / Behind the projects

Different disciplines.
A shared curiosity.

I'm Dr. Emil Jivishov. My independent projects bring together education technology, chemistry and scientific software, robotics, and applied AI.

The common thread is making complex workflows easier to explore and understand. Sometimes that means a virtual laboratory. Sometimes it means a tool that helps a teacher review an assessment, or a workspace for experimenting with a robot.

Dr. Emil JivishovCreator of Reasonance Lab
Find me on GitHub

04 / Start a conversation

Let's talk about
what comes next.

Connect with me on LinkedIn to discuss opportunities in educational technology, scientific software, or applied AI.

Connect on LinkedIn