Quest: AI for Science Tools Landscape
ID: q-ai-tools-landscape
Layer: Forge
Priority: 75
Status: active
Depends-on: Forge quest, Tools quest
Goal
Maintain a living catalog of AI-for-science tools, platforms, agents, and
benchmarks. Help researchers discover the best tools for their work. Help
SciDEX track what to integrate next.
Design decisions
Why wiki entity_type='ai_tool' (not a separate table)
- Reuses existing wiki infrastructure: search, cross-linking, KG edges
- Frontmatter JSON provides structured fields for filtering
- Agents can create/update entries using existing wiki workflows
- Excluded from /wiki neuroscience pages by entity_type filter
Keeping it current
- Agent scanning: recurring task scans arXiv, tech news, Product Hunt
for new AI-for-science launches. Creates draft entries.
- Staleness flag: entries with last_reviewed > 90 days get "needs review" badge
- Community updates: future: let visitors suggest edits / new entries
- Inspirations page: remains hand-curated highlights; landscape is comprehensive
Not polluting the wiki
- entity_type='ai_tool' is EXCLUDED from /wiki search + /wiki listing
- Dedicated browsing page at /forge/landscape
- Links TO neuroscience wiki (tools that study X → link to X's wiki page)
- Links FROM wiki TO landscape (wiki pages can reference relevant tools)
Frontmatter schema for ai_tool entries
{
"tool_category": "platform|agent|model|benchmark|database|library|framework",
"specializations": ["drug_discovery","protein_engineering","genomics",
"literature_synthesis","hypothesis_generation",
"clinical_trials","single_cell","imaging","multi_agent"],
"pricing": "free|freemium|paid|enterprise|open_source",
"open_source": true,
"url": "https://example.com",
"github_url": "https://github.com/org/repo",
"paper_url": "https://arxiv.org/abs/XXXX.XXXXX",
"paper_title": "Paper Title (Year)",
"benchmarks_run": ["SWE-bench","BixBench","BioML-bench"],
"benchmark_scores": {"BixBench": "17%", "SWE-bench": "33%"},
"founded_year": 2024,
"funding_stage": "seed|series_a|series_b|public|academic",
"funding_amount": "$10M",
"team_size": "10-50",
"key_people": ["Jane Doe (CEO)", "John Smith (CTO)"],
"data_sources": ["PubMed","UniProt","AlphaFold"],
"scidex_integration": "integrated|planned|none",
"scidex_tool_ids": ["pubmed_search","uniprot_lookup"],
"last_reviewed": "2026-04-06",
"maturity": "research|beta|production|deprecated",
"tags": ["multi-agent","web-based","api-available"]
}
Page: /forge/landscape
Filters
- Category dropdown (platform, agent, model, benchmark, ...)
- Specialization tags (multi-select)
- Pricing filter (free, open-source, paid)
- Maturity filter
- SciDEX integration status
- Text search across title + description
Display
- Grid or table view toggle
- Each card: logo/icon, name, one-line description, category badge, tags
- Expandable: full description, benchmarks, key people, links
- Sort by: name, category, recently reviewed, funding
Sections
Platforms & Agents — end-to-end AI scientist systems
Models & Methods — foundation models, specialized architectures
Databases & Knowledge Graphs — data infrastructure
Benchmarks — evaluation suites
Libraries & Frameworks — open-source building blocksSeed data
Migrate ~30 entries from /docs/inspirations (AI for Science sections) into
ai_tool wiki entries with structured frontmatter. These become the initial
catalog.
Acceptance criteria
☑ /forge/landscape page renders with filterable catalog
☑ 30+ seeded entries from inspirations page (53 entries as of 2026-04-06)
☑ Each entry has structured frontmatter (category, specializations, url, ...)
☑ Entries excluded from /wiki neuroscience search
☑ "Last reviewed" badge shows staleness
☑ Added to Forge nav dropdown
☑ Agent can create new entries via wiki workflow
☑ Comparison view: /forge/landscape/compare?tools=slug1,slug2[,slug3]
Future extensions
- Comparison view: side-by-side tool comparison ✓ done 2026-04-06
- Benchmark leaderboard: tools ranked by benchmark scores
- Integration roadmap: which tools SciDEX plans to integrate
- User ratings/reviews
- Auto-discovery agent scanning arXiv + GitHub trending
Work Log
Already Resolved — 2026-04-27 17:00:00Z
The /forge/landscape/compare route is live on origin/main via commit 223a3b165.
Evidence: curl http://localhost:8000/forge/landscape/compare returns 200 with "Side-by-Side Tool Comparison" content.
Comparison with 3 real tools (ai-tool-4paradigm-biomed,ai-tool-absciai,ai-tool-agent-laboratory) returns the comparison table with overflow-x:auto wrapper.
Original work was committed in orphan 9d39cf649 ([Landscape] Add /forge/landscape/compare …); the code landed on main through the large squash merge in 223a3b165 (2026-04-27).
2026-04-06 — Comparison view [task:c6a6770f-3d79-48a9-a5f5-af7c16446db5]
Implemented /forge/landscape/compare?tools=slug1,slug2[,slug3]:
- New FastAPI route
GET /forge/landscape/compare accepting comma-separated slugs (up to 3)
- Side-by-side comparison table with 15 fields: category, maturity, pricing, open source, specializations, SciDEX integration, benchmarks, benchmark scores, funding, team size, data sources, tags, last reviewed, description
- Differences highlighted (slightly tinted rows); same values shown in muted style
- Tool selector dropdowns (3 slots) with auto-submit JS to swap tools
- Added "Compare Tools" entry to Forge nav dropdown
- Added "+ Compare" button on each landscape card; JS tracks up to 3 selected tools; sticky compare bar shows selected tools and links to compare URL
- Tested: Python syntax OK; route renders with selector form when no tools provided
2026-04-06 — Auto-discovery scan #1 [task:f2f6629a-1f40-4f3e-aba1-126f58b17fb5]
Weekly scan of arXiv cs.AI/q-bio, GitHub trending, and tech news.
Script: scripts/landscape_scan_2026_04_06.py
12 new entries added (catalog now 35 total):
| Slug | Title | Category | Maturity |
|---|
| ai-tool-ai-scientist-v2 | AI Scientist-v2 (Sakana AI) | agent | research |
| ai-tool-google-ai-co-scientist | Google AI Co-Scientist (DeepMind) | platform | beta |
| ai-tool-evo2 | Evo 2 (Arc Institute / NVIDIA / Stanford) | model | production |
| ai-tool-boltz2 | Boltz-2 (MIT / Recursion) | model | beta |
| ai-tool-openfold3 | OpenFold3 (OpenFold Consortium) | model | beta |
| ai-tool-allen-bkp | Allen Institute Brain Knowledge Platform | database | production |
| ai-tool-drugclip | DrugCLIP (Tsinghua / Peking University) | model | production |
| ai-tool-agent-laboratory | Agent Laboratory (Schmidgall) | framework | beta |
| ai-tool-frontierscience-bench | FrontierScience Benchmark (OpenAI) | benchmark | production |
| ai-tool-fairchem-v2 | FAIRChem v2 / UMA (Meta FAIR Chemistry) | framework | production |
| ai-tool-sciscigpt | SciSciGPT (Northwestern / Dashun Wang Lab) | agent | beta |
| ai-tool-ai-researcher-hkust | AI-Researcher (HKUST / novix.science) | agent | beta |
Staleness check: 0 entries flagged (all reviewed within 90 days).
Highlights for SciDEX:
- Evo 2: 40B DNA model — direct relevance for variant effect prediction in APOE, MAPT, LRRK2
- Allen Institute BKP: 34M cells / 22 species, Alzheimer's + Parkinson's data — Atlas integration candidate
- DrugCLIP: 10M× faster virtual screening (Science 2026) — Forge tool candidate
- Google AI Co-Scientist: multi-agent Gemini debate system mirrors SciDEX Agora architecture
- Boltz-2 + OpenFold3: open structure + binding affinity prediction — drug discovery pipeline
2026-04-06 — Catalog expansion to 50+ entries [task:b837ce33-4343-49cd-ac77-f34dfed301f7]
Expanded catalog from 35 to 53 entries with complete metadata.
Script: scripts/landscape_expand_2026_04_06.py
18 new entries added (catalog now 53 total):
| Slug | Title | Category | Maturity |
|---|
| ai-tool-insilico-medicine | Insilico Medicine (Chemistry42 / PandaOmics) | platform | production |
| ai-tool-recursion | Recursion Pharmaceuticals (Recursion OS) | platform | production |
| ai-tool-benevolentai | BenevolentAI | platform | production |
| ai-tool-isomorphic-labs | Isomorphic Labs (DeepMind spinout) | platform | production |
| ai-tool-genentech-balm | BALM — Genentech Biomedical Language Model | model | research |
| ai-tool-alphafold3 | AlphaFold3 (Google DeepMind / Isomorphic Labs) | model | production |
| ai-tool-esm3 | ESM-3 / ESMFold (Meta FAIR) | model | production |
| ai-tool-rosettafold2 | RoseTTAFold2 / RoseTTAFold All-Atom (Baker Lab) | model | production |
| ai-tool-rfdiffusion | RFdiffusion (Baker Lab / UW IPD) | model | production |
| ai-tool-proteinmpnn | ProteinMPNN (Baker Lab / UW IPD) | model | production |
| ai-tool-scgpt | scGPT (University of Toronto) | model | production |
| ai-tool-celltypist | CellTypist (EMBL-EBI) | model | production |
| ai-tool-scfoundation | scFoundation (Tsinghua University) | model | production |
| ai-tool-elicit | Elicit (AI Research Assistant) | agent | production |
| ai-tool-consensus | Consensus (AI Academic Search) | agent | production |
| ai-tool-atomwise | Atomwise (AIMS Platform) | platform | production |
| ai-tool-genie2 | Genie2 (Generate Biomedicines) | model | production |
| ai-tool-absciai | AbSci (Zero-Shot Antibody Design) | platform | production |
Frontmatter completeness: All 18 entries include url, pricing, maturity, open_source, key_people, specializations, tags, last_reviewed. Entries for industry platforms (Insilico, Recursion, BenevolentAI, Isomorphic, Atomwise, AbSci, Genie2) include founded_year, funding_stage, funding_amount, team_size. Entries with papers include paper_url, paper_title.
Highlights for SciDEX (Neurodegeneration):
- ESM-3 + ProteinMPNN + RFdiffusion + RoseTTAFold2: Complete Baker Lab / EvolutionaryScale protein design toolkit
- AlphaFold3: Gold standard for neurodegeneration target structures (tau, amyloid, α-synuclein)
- scGPT + CellTypist + scFoundation: Single-cell foundation models for brain atlas analysis
- BenevolentAI: KG-driven ALS/Parkinson's programs — direct SciDEX neurodegeneration overlap
- Isomorphic Labs: DeepMind spinout ($2.9B+ pharma deals) — state-of-the-art structure-based drug design reference
2026-04-08 — Priority Tiers 1-3 complete [task:4be33a8a-4095-401e-8223-990b9d29283b]
Curated and created wiki pages for all remaining Priority Tier 1-3 tools.
Script: scripts/landscape_tiers_2026_04_08.py
23 new entries added (catalog now 82 total):
| Slug | Title | Tier | Category |
|---|
| ai-tool-openai-deep-research | OpenAI Deep Research | T1 | agent |
| ai-tool-autoscience-carl | Autoscience Carl | T1 | agent |
| ai-tool-benchsci-lens | BenchSci LENS | T1 | platform |
| ai-tool-causaly | Causaly | T1 | platform |
| ai-tool-converge-bio | Converge Bio | T2 | platform |
| ai-tool-sciencemachine-sam | ScienceMachine Sam | T2 | agent |
| ai-tool-tag-bio | Tag.bio | T2 | platform |
| ai-tool-kepler-ai | Kepler AI | T2 | agent |
| ai-tool-pluto-bio | Pluto Bio | T2 | platform |
| ai-tool-synthesize-bio | Synthesize Bio / GEM-1 | T2 | model |
| ai-tool-elucidata-polly | Elucidata Polly | T2 | platform |
| ai-tool-receptor-ai | Receptor.AI | T2 | platform |
| ai-tool-storm-stanford | STORM | T3 | framework |
| ai-tool-octotools | OctoTools | T3 | framework |
| ai-tool-biodiscoveryagent | BioDiscoveryAgent | T3 | agent |
| ai-tool-re-bench | RE-Bench | T3 | benchmark |
| ai-tool-mle-bench | MLE-Bench | T3 | benchmark |
| ai-tool-curebench | CUREBench | T3 | benchmark |
| ai-tool-gaia-benchmark | GAIA Benchmark | T3 | benchmark |
| ai-tool-dabstep | DABStep | T3 | benchmark |
| ai-tool-pdgrapher | PDGrapher | T3 | model |
| ai-tool-origene | OriGene | T3 | agent |
| ai-tool-eubiota | Eubiota | T3 | agent |
All entries include: URL, category, key features, relevance to SciDEX, last_reviewed date, structured frontmatter with tool_category, specializations, pricing, maturity, open_source, key_people, data_sources, tags, and scidex_integration status. Entries with papers include paper_url and paper_title. Industry entries include founded_year, funding_stage, funding_amount, team_size.
Highlights for SciDEX:
- Causaly: 70M causal relationships — benchmark for SciDEX Atlas KG scale
- BenchSci LENS: 400M+ entity KG with structured evidence extraction — reference for Atlas
- Tag.bio: Direct Parkinson's Foundation partner (PD GENEration data)
- Eubiota: Gut-brain axis microbiome discovery — directly relevant to neurodegeneration
- OriGene: 600+ tools multi-agent system — closest architectural parallel to SciDEX
- PDGrapher: Inverse perturbation prediction for therapeutic targets (Zitnik Lab)
2026-04-08 — DeSci & Token Economy + new tools [task:a8e1228c-8a2c-40c6-84aa-e8c483b92808]
Completed the DeSci & Token Economy section and added notable new tools.
Script: scripts/landscape_desci_2026_04_08.py
7 new entries added (catalog now 89 total):
| Slug | Title | Section | Category |
|---|
| ai-tool-bio-protocol | Bio Protocol / BioAgents (elizaOS) | DeSci | platform |
| ai-tool-molecule-protocol | Molecule Protocol (IP-NFTs) | DeSci | platform |
| ai-tool-scinet | SciNet (Decentralized Science Platform) | DeSci | platform |
| ai-tool-vitadao | VitaDAO (Longevity Research DAO) | DeSci | platform |
| ai-tool-labdao | LabDAO (Decentralized Compute) | DeSci | platform |
| ai-tool-nvidia-bionemo | NVIDIA BioNeMo | Tier 4 | platform |
| ai-tool-gyde | GYDE (Protein Design) | Tier 4 | platform |
All entries include: URL, category, key features, relevance to SciDEX, last_reviewed date, structured frontmatter with tool_category, specializations, pricing, maturity, open_source, key_people, data_sources, tags, and scidex_integration status. DeSci entries include founded_year, funding_stage, funding_amount, team_size where applicable. Open-source entries include github_url.
Highlights for SciDEX:
- Bio Protocol/BioAgents: Direct architectural parallel — multi-agent fleet + knowledge graph + hypothesis generation. $36.9M+ funded, 1,000+ hypotheses on-chain
- Molecule Protocol: Reference implementation for tokenized scientific IP. Founded VitaDAO + Bio Protocol
- VitaDAO: $55M+ treasury, Pfizer-backed, longevity focus overlaps neurodegeneration
- NVIDIA BioNeMo: Infrastructure layer for AI drug discovery — Lilly $1B co-innovation lab, open Clara models
- GYDE: Open-source protein design platform — could integrate for neurodegeneration target analysis
Tool Inventory (2026-04-08)
Covered (wiki pages created)
☑ Inference.bio / Variant Bio (ai-tools-inference-bio)
☑ Biomni — Stanford (ai-tools-biomni)
☑ FutureHouse — Robin, PaperQA2, Aviary, Kosmos (ai-tools-futurehouse)
☑ ScienceClaw — MIT multi-agent (ai-tools-scienceclaw)
☑ Mithrl — NGS lifecycle (ai-tools-mithrl)
☑ Owkin — K-Navigator, OwkinZero (ai-tools-owkin)
Priority Tier 1 — COMPLETE
☑ Google AI co-scientist (ai-tool-google-ai-co-scientist)
☑ OpenAI Deep Research (ai-tool-openai-deep-research)
☑ K-Dense (ai-tool-k-dense)
☑ Autoscience / Carl (ai-tool-autoscience-carl)
☑ BenchSci LENS (ai-tool-benchsci-lens)
☑ Causaly (ai-tool-causaly)
☑ Sakana AI — The AI Scientist (ai-tool-ai-scientist-v2)
Priority Tier 2 — COMPLETE
☑ Recursion/Exscientia (ai-tool-recursion)
☑ Converge Bio (ai-tool-converge-bio)
☑ ScienceMachine / Sam AI Bioinformatician (ai-tool-sciencemachine-sam)
☑ Tag.bio (ai-tool-tag-bio)
☑ Kepler AI (ai-tool-kepler-ai)
☑ Pluto Bio (ai-tool-pluto-bio)
☑ Synthesize Bio / GEM-1 (ai-tool-synthesize-bio)
☑ Elucidata (ai-tool-elucidata-polly)
☑ Receptor.ai (ai-tool-receptor-ai)
Priority Tier 3 — COMPLETE
☑ Agent Laboratory (ai-tool-agent-laboratory)
☑ STORM (ai-tool-storm-stanford)
☑ OctoTools (ai-tool-octotools)
☑ BioDiscoveryAgent (ai-tool-biodiscoveryagent)
☑ RE-Bench (ai-tool-re-bench)
☑ MLE-Bench (ai-tool-mle-bench)
☑ BioML-Bench (ai-tool-bioml-bench)
☑ CureBench (ai-tool-curebench)
☑ GAIA (ai-tool-gaia-benchmark)
☑ DABStep (ai-tool-dabstep)
☑ PDGrapher (ai-tool-pdgrapher)
☑ OriGene (ai-tool-origene)
☑ Eubiota (ai-tool-eubiota)
DeSci & Token Economy — COMPLETE
☑ BioProtocol / BioAgents / ElizaOS (ai-tool-bio-protocol)
☑ Aubrai (ai-tool-aubrai)
☑ SciNet (decentralized science platform) (ai-tool-scinet)
☑ Molecule / IP-NFTs (ai-tool-molecule-protocol)
☑ Beach.Science (ai-tool-science-beach)
☑ VitaDAO (ai-tool-vitadao)
☑ LabDAO (ai-tool-labdao)
New Tools (Tier 4 — notable 2026 launches)
☑ NVIDIA BioNeMo (ai-tool-nvidia-bionemo)
☑ GYDE (ai-tool-gyde)