the questions we actually get asked.

start in chat or curl without an account, within the bounded guest allowance. ask for what you want in natural language, then let Ivy choose and chain the tools. Accounts add saved sessions, higher quota, and the full catalog. For terminal and agent workflows, use the ivybloom CLI and its MCP server; the public docs cover the inputs, outputs, and workflow recipes.

we support a wide range of molecular and data formats:

protein structures:
• pdb, mmcif, cif — standard structure formats
• pdbqt — docking-ready structures

small molecules:
• sdf, mol, mol2 — 3d molecular structures
• smiles, inchi — chemical notation
• safe — sequential attachment-based fragment embedding
• xyz — atomic coordinates

sequences:
• fasta, fastq — sequence data
• genbank — annotated sequences

data:
• csv, json, xlsx — tabular data
• parquet — large datasets

diffraction:
• xrd patterns for crystallography tools

genomics:
• bam, vcf, bed — ngs and variant data

each tool indicates its specific format requirements in the interface.

the platform works best on modern browsers: chrome 120+, firefox 120+, safari 17+, or edge 120+. we recommend using chrome for optimal performance, especially for mol* 3d molecular visualizations. javascript must be enabled, and we suggest at least 8gb ram for large molecular datasets and complex visualizations.

individual accounts:
sign up with your email and verify your account. you'll start on the seed (free) tier with monthly credits and access to core tools. upgrade to grow or flourish for more features.

team accounts:
the team tier (per-seat pricing) enables shared projects with:
• admin dashboard for user management
• role-based access control (owner, admin, editor, viewer roles)
• per-seat credit allocation
• invitation management

enterprise accounts:
• sso/saml authentication
• dedicated customer success manager
• custom integrations and on-premise options
• unlimited concurrent jobs

students:
verify with .edu email for the student tier — free monthly research credits with advanced workflow access.

prepaid credit packs:
buy a credit pack and spend it as you go — no subscription, no bill after the fact. packs include access to all tools and 5 model families.

see [/pricing](/pricing) for current plan details and rates.

ivybloom is the command-line surface for the platform:

installation:
• npm install -g @ivybiosciences/ivybloom-cli
• curl -fsSL https://www.ivybiosciences.com/install.sh | sh

key capabilities:
• submit and monitor jobs from the terminal
• batch processing and scripting
• interactive job monitoring
• session-aware authenticated workflows
• MCP delivery for coding agents

example:
• ivybloom run esmfold protein_sequence=MKTAYIAK...
• ivybloom jobs follow JOB_ID
• ivybloom batch submit proteins.json --parallel 5

authentication:
use browser-based login or an API key for the account-bound CLI and MCP surfaces.

output formats: table, json, csv, yaml.

we provide multiple structure prediction tools:

single-chain prediction:
• fast, high-quality predictions comparable to alphafold2 for most globular domains
• plddt confidence scores indicate per-residue reliability
• best for proteins under ~1500 residues

complex prediction:
• protein-protein and protein-ligand complexes
• improved accuracy on multimeric assemblies

pre-computed structures — a stored structure is returned for known proteins without compute cost

protein design:
• sequence design for ligand-binding proteins
• general protein sequence design
• diffusion-based protein structure generation

limitations:
• intrinsically disordered regions have lower confidence
• very large proteins may require segmentation
• use confidence scores to identify reliable vs. uncertain regions

we provide multiple molecular docking approaches:

primary docking:
• diffusion-based flexible docking (diffdock)
• cnn-scored docking (gnina)
• collaborative docking for difficult targets

complementary tools:
• binding pocket detection (p2rank)
• binding free energy refinement
• drug-target binding affinity prediction

capabilities:
• blind docking — no binding site required (auto-detected)
• batch docking — screen multiple ligands against one target
• pose visualization with scoring and interaction details

input requirements:
• receptor: pdb format protein structure
• ligand: sdf format or smiles notation
• binding site: optional (auto-detected if not specified)

output: top-scoring poses ranked by confidence, with interactive 3d visualization.

our generative drug design platform combines multiple ai approaches:

molecule generation:
• diffusion-based molecular generation
• scaffold-based de novo design
• fragment growing from binding sites
• constraint-based generation (pharmacophore, similarity)

fragment-based design:
• native fragment notation handling
• fragment growing, search, and extraction
• scaffold decoration with functional groups
• linker design for fragment connections

optimization goals:
• target binding affinity
• admet properties
• synthetic accessibility
• novelty vs. known chemical space

output:
• ranked molecule library with scores and molecular properties
• 2d structure thumbnails
• admet profiles and retrosynthesis routes

available on flourish tier and above.

we provide a crystallography suite:

xrd analysis:
• diffraction database search and analysis (opxrd)
• powder diffraction pattern analysis and matching (pywpem)

crystal structure prediction:
• polymorph prediction and crystal structure generation (xtalnet)
• crystal structure prediction for novel compounds

these tools support pharmaceutical solid-state research, polymorph screening, and materials characterization workflows. upload your diffraction data or smiles strings to predict crystal forms and match against reference databases.

available on power tier and above.

we provide retrosynthesis and synthesis planning:

retrosynthesis:
• retrosynthetic route analysis (aizynthfinder)
• multiple synthetic routes ranked by feasibility
• identifies commercially available starting materials

reaction analysis:
• reaction type classification and prediction (reaction_classifier, rxnfp)
• bulk reaction database search

compound sourcing:
• purchasable compound search across major libraries (molport, zinc)
• supplier availability and pricing comparison
• building block identification

output:
• interactive reaction network visualization
• exportable synthesis schemes
• supplier links for starting materials
• cost estimates for synthetic routes

available on grow tier and above.

job completion times vary by tool and input size:

typical times:
• esmfold (protein folding): 2-15 minutes
• diffdock (molecular docking): 5-30 minutes
• admet prediction: 1-5 minutes
• genmol (drug design): 1-10 minutes
• retrosynthesis: 10-60 minutes

all jobs include real-time progress updates via the web app, cli (ivybloom jobs follow JOB_ID), or sse streaming.

concurrent job limits:
• seed (free): 3 concurrent jobs
• grow: 5 concurrent jobs
• flourish: 10 concurrent jobs
• team: 3 concurrent jobs
• enterprise: unlimited

priority queue:
• grow: priority level 1
• flourish/student/team: priority level 2
• enterprise: priority level 4 (highest)

we provide extensive genomics and bioinformatics capabilities:

sequence analysis:
• sequence homology search (local and remote)
• structural similarity search
• protein language model embeddings (esm3)

variant analysis:
• deep-learning variant calling
• variant effect prediction
• clinical variant annotation and interpretation
• polygenic risk scoring

crispr:
• guide rna design and off-target prediction
• crispr screening analysis

bioinformatics workflows:
• genome assembly
• rna-seq differential expression
• variant calling pipelines
• metagenomics and microbiome analysis
• phylogenetics
• chip-seq/atac-seq, gwas, hi-c, nanopore, and more

we integrate multiple ai providers for reliable, high-quality analysis:

available models:
• leading foundation models for scientific reasoning, summarization, and analysis
• access to a broad selection of models across multiple providers
• self-hosted model options for private inference

built-in capabilities:
• web search — real-time scientific literature lookup
• code execution — inline data analysis
• tool calling — automatic orchestration of computational tools

specialized models:
• biomedical text mining and entity extraction
• molecular property prediction (admet, solubility, toxicity)

model access scales with your tier — higher tiers unlock premium models and higher rate limits.

our target discovery platform integrates multiple data types:

omics analysis:
• single-cell rna-seq analysis and dimensionality reduction
• molecular machine learning models
• bulk rna-seq differential expression analysis

structural biology:
• structure prediction from sequence
• complex structure prediction
• druggable pocket detection
• pre-computed structures returned for known proteins

literature & database mining:
• biomedical entity extraction
• target-disease associations
• protein interaction networks
• literature search and analysis

regulatory intelligence:
• multi-jurisdiction approval data
• curated safety signals and drug warnings
• scheduling and classification data

target passport — comprehensive report aggregating druggability, structures, literature, disease associations, and known drugs into a single assessment.

we provide comprehensive omics analysis across many domains:

single-cell rna-seq:
• standard single-cell analysis pipelines
• advanced dimensionality reduction
• cell type annotation, clustering, differential expression

spatial omics:
• spatial transcriptomics analysis
• tissue-level gene expression mapping

metagenomics:
• microbiome classification, taxonomic profiling
• diversity analysis, differential abundance

additional domains:
• epigenomics — chip-seq, atac-seq, methylation
• metabolomics — metabolite identification and pathway analysis
• cytometry — flow and mass cytometry analysis
• proteomics — protein quantification and interaction mapping
• systems biology — flux balance analysis, metabolic modeling, gene essentiality

cell therapy:
• car-t and nk-car design workflows
• candidate selection and safety profiling

targeted degradation:
• ternary complex scoring
• linker design and degradation prediction

we use multiple ml models for safety profiling:

admet prediction:
• admet property prediction across dozens of endpoints (admet-ai)
• fast predictions for absorption, distribution, metabolism, excretion, and toxicity
• toxicity class prediction (admet-ai)

binding & solubility:
• protein solubility prediction (deepsol, graphsol)
• drug-target binding affinity prediction (deeppurpose)
• binding free energy refinement (mmgbsa)

regulatory intelligence:
• fda/ema/pmda approval histories
• curated drug withdrawals and boxed warnings
• adverse event reports and recalls

safety passport — aggregated report combining adverse event data, drug interactions, label warnings, recalls, and risk assessment.

these predict drug-drug interactions, pharmacokinetic parameters, and safety endpoints with high accuracy.

yes! we provide multiple literature analysis tools:

data sources:
• biomedical literature databases (pubmed, europe pmc, pubmed central)
• preprint repositories
• patent databases

ai capabilities:
• biomedical entity recognition and relationship extraction (biobert)
• deep analysis, hypothesis generation, and scientific reasoning
• process thousands of papers to extract relationships
• identify trends across publication timelines
• generate hypotheses based on literature gaps
• summarize key findings with auto-generated citations
• compare methodologies across studies

literature passport — intelligence report covering research gaps, competitor activity, and emerging targets.

passports are comprehensive entity intelligence reports that aggregate data from multiple tools and databases into a single assessment.

core passport types:
• compound passport — admet, synthesis routes, vendor pricing, docking summary, molecular properties
• target passport — druggability, structures, literature, disease associations, known drugs
• safety passport — adverse events, drug interactions, label warnings, recalls, risk level
• literature passport — research entities, gaps, competitor activity, emerging targets
• therapeutic area passport — disease landscape, target/compound intelligence, clinical pipeline

additional passport types — biologics design briefs, biomarker dossiers, drug repurposing reports, patent landscapes, pathway analysis, and pharmacovigilance dashboards.

passports are generated via the ai orchestrator or api. results include provenance tracking and completeness metrics.

available on grow tier and above (some passport types require enterprise).

we provide multiple visualization options:

3d molecular viewers:
• interactive 3d viewer for protein structures and complexes
• per-residue property visualization (confidence scores, contacts)
• lightweight viewers for quick inspection

specialized result views:
• docking pose selector with 3d co-display and scoring columns
• generative chemistry results with 2d thumbnails and sorting
• analytics dashboards for dose-response, survival, enrichment, and safety data
• systems biology and cell therapy result views

network & graph:
• pathway and interaction network visualization
• reaction network visualization for synthesis planning

data plots:
• admet radar plots, kaplan-meier curves, forest plots, waterfall charts, and more

export: publication-quality figures in png, svg, html, pdf, docx, and pptx.

advanced visualization features available on grow tier and above.

yes! batch processing is supported across all tools:

input options:
• csv files with smiles columns
• sdf files with multiple molecules
• fasta files with multiple sequences
• programmatic submission via api or cli

batch capabilities by tier:
• seed: 3 concurrent jobs
• grow: 5 concurrent jobs
• flourish: 10 concurrent jobs + priority queue
• team: 3 concurrent jobs
• enterprise: unlimited concurrent jobs

daily job limits (studio submissions):
• seed: 10/day
• grow: 50/day
• flourish: 200/day
• team: 500/day
• enterprise: unlimited

cli batch processing:
• ivybloom batch submit jobs.json --parallel 5
• supports job dependencies with depends_on chains
• agent mode output for ci/cd pipelines

use cases: virtual screening campaigns, admet profiling of compound libraries, protein family analysis.

we provide a flexible document generation system:

document formats:
• json — structured data for programmatic use
• csv — spreadsheet-compatible tabular data
• markdown — documentation and notes
• html — interactive shareable reports
• pdf — publication-ready documents with charts
• docx — microsoft word reports
• pptx — powerpoint presentations

report templates:
compound passport, target passport, safety profile, literature review, therapeutic area, trial comparison, and more — each with embedded vega charts (admet radar, kaplan-meier, forest plots, waterfall, phase funnel).

molecular data:
• pdb, mmcif, sdf, mol2 — structural data
• smiles, inchi, safe — chemical notation
• fasta — sequences

all exports include provenance metadata. api and cli access available on flourish tier and above.

yes! our rest api enables programmatic access to all computational tools:

who gets it: every account, including the free seed plan. create an ivy_ api key in settings; plans differ in credits, concurrent jobs and model families, not in api access. team and enterprise add team management.

key capabilities:
• submit and monitor computational jobs
• batch job submission
• real-time streaming for job progress
• tool discovery and schema inspection
• result retrieval in multiple formats

ivybloom cli:
our cli provides terminal access to the full platform:
• ivybloom run esmfold protein_sequence=...
• ivybloom batch submit jobs.json --parallel 5
• ivybloom jobs results JOB_ID --output-format json

ai agent integration:
the cli includes an mcp server for integration with ai coding assistants and other mcp-compatible clients.

authentication: api keys or session tokens. full documentation available in your dashboard.

yes! we integrate with dozens of scientific databases across key domains:

structure databases:
• protein structure repositories and pre-computed predictions
• protein sequence and annotation databases

compound databases:
• public chemical databases and bioactivity data
• purchasable compound libraries and sourcing
• regulatory approval databases

genomics/variant databases:
• clinical variant interpretations and population frequencies
• target-disease associations and pathway databases
• genome annotations and functional genomics data

literature:
• biomedical literature and preprint repositories
• full-text access and patent search

we also integrate with protein interaction networks, expression databases, antibody databases, and more.

all databases are queryable by asking Ivy or via the api.

yes! webhooks enable real-time notifications for workflow automation:

supported events:
• job completion — triggers when jobs finish (success or failure)
• job progress — milestone updates during long-running jobs
• session updates — state changes in persistent sessions

cli integration:
the ivybloom cli supports callback urls on job submission:
• submit jobs with a callback_url for completion notifications
• agent mode (--agent) returns structured json for ci/cd pipelines
• sse streaming for real-time progress updates

payload:
• job id and status
• result summary and download urls
• timing and resource usage
• error details (for failures)

availability:
api keys work on every plan, including the free seed plan.

configure webhooks in your project admin dashboard.

authenticated accounts can save sessions and tool results; guest runs have bounded quota and should not be treated as persistent storage. check the current plan and storage documentation for retention and export details.

api & cli access:
• create, list, and manage sessions programmatically
• real-time streaming for tool execution
• cli session management with checkpoint rollback

enterprise extensions:
advanced conflict resolution, compliance logging, and priority queuing available for enterprise customers.

we use a tiered subscription model with credit-based compute pricing:

individual tiers:
• seed (free) — monthly credits, 3 concurrent jobs, core tools (esmfold, diffdock, admet)
• grow — more credits, 5 concurrent jobs, advanced visualization, retrosynthesis, openai models
• flourish — higher credit allocation, 10 concurrent jobs, drug design (genmol), priority queue, api access

team & enterprise:
• team (per-seat pricing) — per-seat credits, shared projects, rbac, admin dashboard, full api access
• enterprise (custom) — unlimited jobs, sso/saml, sla guarantees, dedicated csm, on-premise option

flexible options:
• student (free) — monthly research credits, priority queue, full api access — verified .edu email
• prepaid credit packs (no subscription) — all tools and 5 model families. buy a credit pack and spend it as you go — no subscription, no bill after the fact.

yearly billing costs less than twelve monthly payments — the current saving is shown on the yearly toggle at [/pricing](/pricing).

current dollar amounts and credit allocations live on [/pricing](/pricing).

yes! we provide dedicated support for academic users:

student tier (free):
• generous monthly research credits
• advanced workflow access (gpu folding, generative design, standard pipelines)
• verified with .edu email address, no payment required

academic institutions:
• team plans with academic pricing
• bulk licensing for research groups
• collaboration features for labs

non-profit research:
• special pricing available
• contact our academic team for custom arrangements

verification is automatic with institutional email addresses.

usage is measured by credits and concurrent jobs:

monthly credits:
each tier includes a monthly credit allowance, from 500 (seed) up to 15,000 (flourish), with team plans offering credits per seat. credit packs carry no monthly allowance: buy a credit pack and spend it as you go — no subscription, no bill after the fact.

concurrent job limits:
ranging from 3 concurrent jobs (seed) to unlimited (enterprise). higher tiers get more parallel capacity.

compute pricing:
credits scale with gpu tier — lighter workloads cost less, high-performance gpus cost more. see the pricing page for current rates.

credits are calculated based on gpu tier, duration, and workload type.

step 1: check job logs
view the error message in your job details — most issues are explained there. in the cli, run ivybloom jobs status JOB_ID for details.

common issues:
• invalid input format — verify pdb/sdf/smiles syntax
• sequence too long — esmfold limit is ~1500 residues
• missing required fields — check tool documentation
• timeout — large jobs may need retry with smaller input
• credit limit — check remaining credits with ivybloom credits

step 2: retry
• fix identified issues and resubmit
• try with a smaller test input first
• check input file encoding (utf-8 recommended)
• use ivybloom doctor for cli diagnostics

step 3: contact support
if problems persist:
• include your job id
• describe what you expected vs. what happened
• attach input files if possible

support response times:
• seed: community forum
• grow/flourish: email support (24h)
• team: priority support (4h)
• enterprise: dedicated support (1h)

job speed depends on several factors:

input complexity:
• large proteins (>1000 amino acids) take longer
• complex molecular systems require more compute
• batch jobs scale with input count

queue position:
• peak hours may have longer queue times
• priority queue (grow+ tiers) gets faster scheduling
• enterprise tier gets highest priority (level 4)

optimization tips:
• break large jobs into smaller batches
• use the cli for batch processing: ivybloom batch submit jobs.json --parallel 5
• upgrade tier for higher concurrent job limits and priority queue
• monitor jobs in real-time: ivybloom jobs follow JOB_ID

contact support if performance issues persist — we can help diagnose bottlenecks.

check these common issues:

format validation:
• pdb — valid atom records, proper chain ids
• sdf/mol — correct bond orders, 3d coordinates
• smiles — valid syntax, no invalid characters
• safe — valid safe notation for fragment-based inputs
• fasta — proper headers, valid sequences
• csv — utf-8 encoding, correct delimiters

troubleshooting:
• try a different browser (chrome recommended)
• check file isn't corrupted (open locally first)
• ensure file extension matches content
• for large files, compress or split into batches
• use the cli for large batch uploads: ivybloom data upload FILE

contact support with file details if issues persist.

we implement enterprise-grade security:

encryption:
• aes-256 encryption at rest
• tls 1.3 for data in transit
• encrypted backups

authentication:
• user accounts with multi-factor authentication
• api keys for programmatic access
• sso/saml for enterprise single sign-on
• device and location-based session validation

compliance posture:
• framework-by-framework status is published at /help/compliance — 21 cfr part 11 and soc 2 type ii are in progress; hipaa is planned
• our infrastructure providers hold their own soc 2; that is their attestation, not ours
• gdpr — eu data protection

access controls:
• hierarchical rbac (org → project → campaign → chat roles)
• audit logging on authenticated data access
• session validation with device and location context

data ownership:
• you retain full ownership of your data
• we never share data with third parties
• we don't use your data to train models without permission
• delete your data anytime

data is encrypted in transit and at rest. exports and deletion are available through the documented controls; retention and backup details depend on the service and plan.

data controls:
• download your data in multiple document and molecular formats
• delete data on demand
• export complete project archives

your intellectual property is protected through multiple measures:

data isolation:
• your data is stored in isolated tenant environments
• no cross-customer data access
• strict access controls with audit logging

ai training policy:
• we never use customer data to train models
• your molecules, structures, and results remain yours
• no data mining or pattern extraction

legal protections:
• standard terms include confidentiality provisions
• custom nda arrangements for sensitive programs
• enterprise customers: tailored ip agreements

technical measures:
• aes-256 encryption at rest
• tls 1.3 in transit
• secure deletion on request
• no data retained after account closure

enterprise options:
• on-premise deployment (your infrastructure)
• private cloud instances
• private model inference for sensitive data
• air-gapped environments for classified work

yes! team collaboration uses a hierarchical role-based access control system:

organization roles: owner, admin, member
project roles: owner, admin, editor, viewer
campaign roles: owner, admin, member, viewer
chat roles: owner, admin, member, viewer

team features include:
• shared projects — centralized project management with visibility controls (private, team, public)
• admin dashboard — user management and analytics
• invitation system — invite members with specific roles
• resource sharing — share projects, campaigns, and chat sessions

available on:
• team tier — per-seat pricing (see [/pricing](/pricing))
• enterprise — sso/saml, dedicated csm, custom integrations

individual tiers (seed, grow, flourish) are single-user. upgrade to team for collaboration.

custom model training is available for enterprise customers:

training options:
• transfer learning from our base models
• fine-tuning on proprietary datasets
• custom model development for specific therapeutic areas
• self-hosted models on dedicated gpu infrastructure

typical use cases:
• proprietary admet models trained on internal data
• target-specific binding predictors
• custom molecular generation models
• disease-specific classifiers

process:
1. data assessment and preprocessing
2. model architecture selection
3. training on cloud GPU infrastructure
4. rigorous validation (cross-validation, external test sets)
5. deployment into your organization's private environment

data privacy:
your training data remains confidential and is not used for other customers.

contact enterprise sales for custom model development.

our platform runs on managed gpu infrastructure with per-second billing:

gpu options:
• latest-generation gpus for highest performance
• high-memory gpus for large models and complex simulations
• balanced options for most standard workloads
• cost-effective options for lighter tasks
• cpu-only for database queries and simple tasks

credits by the second:
all tiers access the same gpu infrastructure. a job draws credits for the compute time it uses, counted per second. even seed (free) tier users can run gpu jobs using their monthly credits. see the pricing page for current rates.

priority queue:
faster job scheduling is available from grow tier and above, with enterprise getting the highest priority.

the licence and the release: the platform is licensed FSL-1.1-Apache-2.0, and a public source release is planned; the repositories are not public yet. once released, researchers can inspect the implementation and use it for permitted work, while the license keeps competing-service resale restricted.

fsl-1.1-apache-2.0 licensing:
- functional source license: free for non-competing use, including internal research and teaching
- automatic apache 2.0 conversion: each release becomes Apache 2.0 two years after release
- enterprise licensing: regulated deployment and commercial features may require separate terms

what this means for you:
• researchers and academics: use the software for permitted research and teaching
• teams: review the license before incorporating the software into a commercial service
• contributors: report bugs through /help/support and send feature requests through /suggestions until the repositories open

planned for release:
• web application
• backend compute infrastructure
• ivybloom cli
• tool definitions and documentation

fsl-1.1-apache-2.0 protection: the functional source license prevents competitors from directly repackaging our platform while allowing free use for research and non-competing purposes. after two years, code converts to apache 2.0.

trade secrets: keep proprietary datasets, training methods, and competitive advantages closed. publish the platform infrastructure under fsl while keeping proprietary data assets closed.

licensing tiers:
- fsl core platform: free for non-competing use, converts to apache 2.0
- enterprise features: 21 cfr part 11, sso/saml, white-label under separate license
- custom agreements: tailored partnerships with pharma for specific needs

contributor agreements: contributor license agreements (clas) maintain legal protections while accepting community contributions.

enterprise features: proprietary capabilities (advanced scoring, enrichment, compliance) are available via enterprise licensing. self-hosters get sensible defaults for all core functionality.

transparency advantage: pharma companies prefer vendors with transparent, auditable technology. the source release is planned, not yet public; once released, it enables:

• due diligence: partners can audit codebase for security and compliance
• customization: ability to modify platform for specific therapeutic areas
• integration: easier integration with existing pharma workflows
• validation: community adoption provides third-party validation

partnership pipeline:
- academic research: builds credibility through scientific validation
- pilot programs: low-risk testing with startups and mid-size pharma
- enterprise deals: high-value partnerships with big pharma

regulatory compliance:
- transparent codebase simplifies fda/emea regulatory review
- community testing increases quality assurance
- open standards enable better interoperability
- academic validation provides scientific credibility

current status:

the platform is licensed fsl-1.1-apache-2.0, and a public source release is planned:
• the repositories are not public yet
• permitted research and teaching use is available under the license
• each release converts to apache 2.0 after 2 years

planned for release:
• web application
• backend compute infrastructure
• command-line interface
• session runtime and tool registry packages
• tool definitions and documentation

contribution guidance:
• report bugs through /help/support and send feature requests through /suggestions
• contact the team for research or partner integrations

enterprise features (separate terms):
• regulated deployment support
• SSO/SAML authentication
• dedicated assistance