For researchers and partners

What we are building - and how you can collaborate

We are building open-source tools that help rare-disease researchers reach answers faster. We start with fibrous dysplasia (FD/MAS), but the architecture is universal.

Research Canvas is an iterative AI–researcher collaboration platform - not an autonomous "AI scientist", not an ELN/LIMS, not yet another workflow builder for developers. AI proposes workflow variants and hypotheses; the researcher decides, modifies, clones; wet-lab results feed back into the AI. It is controlled autonomy: full AI freedom inside a block, deterministic skeleton between blocks, auditable trace of every run.

Three directions

Open-source tools and clinical research

01 In progress

Research Canvas

Agentic workflows for researchers: deterministic steps + AI blocks + a knowledge base (Karpathy-style wiki). The researcher defines the pipeline; the system executes it step by step with full control over each stage.

First use case: drug target prioritization for FD - ranking ~1000 molecular candidates down to the 5 most promising, with provenance for every score.

CLI MVP: June 2026
License: open-source (Apache 2.0)

02 Concept

Live Guidelines

AI monitors PubMed and drafts clinical-guideline updates. A clinician approves changes in a git-style PR review - every change is versioned and auditable.

Instead of a static PDF every few years - continuously updated, version-controlled clinical knowledge. Currently a post-MVP concept.

03 Research

FD research directions

CRISPR / targeted editing - blocking the mutation in affected somatic cells.

Molecules blocking aberrant signaling - a preclinical direction with documented efficacy in mouse FD models. The path to clinical translation in humans.

Organoids / iPSC-derived models - tissue models for testing therapies in vitro.

Natural mechanisms - studying cases of spontaneous regression and natural protective mechanisms.

Research roadmap

What we deliver, and when

The plan has six parallel streams. Some are done, some are in progress, some are upcoming. We update this timeline live - no glossing over, with a concrete status on each item.

March 2026 · done Mapped the AI-for-science landscape: 13-platform analysis (Biomni, Kosmos, Google AI co-scientist, ScienceClaw, AI-Researcher, and others). Identified the gap: iterative AI ↔ researcher collaboration with workflow as the first-class artifact. First sketches of Research Canvas architecture.
April 2026 · in progress Grant application to Google.org for Nonprofits (Open Source AI round). Decision expected in H2 2026. In parallel, we are studying researcher context (problem framing, reasoning paths, iteration patterns) as a foundation for the pilot.
June 2026 · planned Research Canvas CLI MVP - first working workflow runner with a deterministic sequence of blocks and AI freedom inside each block. Open-source, Apache 2.0. Public repo on GitHub.
Q3 2026 · planned First FD case study - full diagnostic pipeline from histopathology through GNAS testing to therapeutic decision, with provenance for each step. In parallel: Live Guidelines pilot - Polish FD diagnostic path as a versionable git-style document.
Q4 2026 · planned First formal scientific partnership with an academic FD/MAS research center. Seed of an FD audit corpus - organized clinical and imaging data for internal model training. 5–10 fork-able workflow templates for rare-disease research.
2027 H1 · planned Patient Registry FD/MAS Poland - GDPR-compliant patient registry with onboarding through the foundation. First 5–10 budget-constrained academic labs onboarded onto Research Canvas.
2027 H2 · planned Knowledge Navigator - search across the FD corpus with full provenance (every fact has an auditable source). Open API for integration with external research systems.

This roadmap is ambitious and may slip - especially milestones tied to the grant decision. We update each item's status on this page. If a specific stream interests you (CLI MVP, Live Guidelines, Patient Registry), get in touch - we'll happily share details and current progress.

What sets Research Canvas apart

Three things no existing system delivers

After reviewing 13 AI-for-science platforms (Stanford Biomni, FutureHouse Kosmos, Google AI co-scientist, MIT ScienceClaw, HKU AI-Researcher, Benchling, and others) we identified a gap that nobody fills this way:

Workflow as a first-class artifact - the researcher authors their methodology as a versioned document, instead of merely describing a goal in natural language and hoping the agent interprets it correctly.
Controlled autonomy - AI has full freedom inside a block, but the skeleton between blocks is deterministic (sequence, gates, I/O validation). This is the precondition for reproducibility, auditability, and acceptance in clinical and regulatory contexts.
Open-source overlay on top of existing tools - we orchestrate AlphaFold, Evo 2, ChEMBL, OmicsBox, Galaxy, Benchling via MCP - we don't replace them. A lab doesn't need to abandon what it has; we add a reasoning layer on top of the ELN/LIMS and bio models.

12–18 month targets

What we plan to deliver after the first year

5 Validated FD candidate targets - drug target prioritization in an internal pilot

1 Wet-lab tested target - in a partner academic laboratory

50+ Public NGS tool inventory - a forkable catalog with MCP integration assessments

20 Fork-able workflow templates - open-source workflow library for rare-disease research

5–10 Custom MCP wrappers - OmicsBox, Galaxy, AlphaFold, Evo 2, ChEMBL as public goods

5–10 Budget-constrained academic labs onboarded - primarily Polish and Eastern-European academic teams

Who builds this

Track record

GeneQuest doesn't start from zero. The founder has 30 years of experience building enterprise systems, including commercial products deployed in large organizations.

ARSmacros

A no-code tool automating work in BMC Helix ITSM / Remedy. Acquired by IBM, deployed at Telstra (savings on the order of 30 FTEs). Used in many countries, maintained for years.

arsmacros.com →

TicketClip

A current portfolio product. A tool supporting ticket management in ITSM environments.

The same skills - building tools that solve real problems in large organizations - are now directed at rare diseases. Research Canvas and Live Guidelines aren't academic prototypes; they are products built by someone who knows how to ship software.

Waitlist

We will let you know when the CLI MVP goes public

The first open release of Research Canvas is planned for June 2026. Leave your email - we will write when the repo goes public. Occasionally we will share progress in the meantime. No spam, no data trading.