BenchSci vs Phylo

Before a biologist can analyze a dataset, there's usually a lot of setup. They have to find the right public datasets, install and configure the analysis software, write the code that runs it, and read the relevant literature. AI agents for biology now handle much of that work, so the scientist can go straight to the question.

Phylo and BenchSci both build this kind of AI agent. The biggest difference between them is the data each agent can use. Phylo grew out of Biomni, an open-source project at Stanford. Its Biomni Lab platform connects agents to 300+ databases, software packages and tools, plus partner data sources such as COSMIC for cancer genomics and Addgene for plasmids.

EMET by BenchSci draws on the same kinds of public resources. It also gives its agents biology no other platform holds: 16M closed-access papers that BenchSci licenses from publishers, and 85M reagent and model-system data points curated by BenchSci's PhD scientists.

This comparison scores the two platforms on five criteria. They are the data each one reasons over, the depth of its biology-specific workflows and integrations, the service that drives adoption, the freedom to use the best model for each task, and whether the vendor develops drugs of its own. The two are evenly matched on model flexibility and on neutrality. BenchSci leads on the other three.

What to look for in AI for preclinical R&D

Many AI agents for biology can now read papers, run code and return a chart in minutes. That makes a demo a weak guide to which platform will still carry a program a year from now. Five traits underneath decide it.

• Proprietary, biology-specific evidence to reason over. An agent can only reason from the evidence it can reach. Curated biology that other platforms can't use gives it more to work with on the target and safety questions that decide a program.
• Depth of biology-grounded workflow and integration. Preclinical work runs from a literature question to an omics analysis to a decision at the bench. A platform earns its place by doing that biology and by connecting to the systems that hold a lab's own data.
• A service model that drives adoption. Buying a platform doesn't change how a lab works. Scientists who build workflows around each team's projects decide whether it becomes part of daily research.
• The freedom to use the best model for each task. No single model is best at every scientific job, and new ones ship every few weeks. A platform tied to one provider is limited by that provider's roadmap and pricing.
• A clear position on whether the vendor develops its own drugs. Your targets and results pass through this system. You should know whether the vendor could ever use what it learns to compete with you.

The sections below take EMET and Biomni Lab through each criterion in turn.

Criterion 1: The data and evidence the platform reasons over

Give two agents the same model and the same tools, and they'll still reach different conclusions if one can see more of the evidence. This criterion asks what each platform's agents can read when they answer a question.

Phylo connects its agents to a broad set of public and partner resources

Biomni Lab's strength is breadth. Phylo's launch announcement describes 300+ databases, software systems and analytical tools in one environment. Partner connections include Consensus for academic literature, COSMIC for cancer genomics and Addgene for plasmids. The platform also checks its reasoning against a biology knowledge graph before returning an answer.

Biomni Lab can also work on a customer's own data. In its Chugai collaboration, its agents connect to proprietary research data inside Chugai's secure environment. What Phylo doesn't bring is a proprietary, licensed biology dataset of its own. Its agents reason over public resources, partner providers and whatever each customer supplies.

BenchSci reasons over closed-access biology and experimental data no one else holds

EMET reaches the same kinds of public sources, then adds a layer BenchSci spent a decade building. The platform reasons over a biological knowledge graph of 858M nodes and 2.2B relationships drawn from 38M+ publications. That corpus includes 16M closed-access papers licensed through eight years of partnerships with publishers such as Elsevier, Springer Nature, Wiley and Oxford University Press. Those papers are in EMET and nowhere else.

Under the literature sits a reagent and model-systems database of 85M data points: 16M antibodies, 22M RNAi entries, 18M CRISPR records, 500K cell lines and 700K animal models. A team of 60 PhD scientists curates it. That layer answers the practical questions behind an experiment, such as which antibody has been validated in which model system.

EMET also reasons over your own data, including ELN entries, study reports and structured systems. That data stays isolated to your organization. A neuro-symbolic evaluation loop checks each claim against the knowledge graph and reaches 95%+ accuracy on biological questions.

Both platforms read public sources and customer data. Only EMET adds closed-access biology on top.

Criterion 2: Depth of biology-grounded workflow and integration

A preclinical decision draws on literature, omics data, safety signals and a lab's own past experiments. Joining them usually falls to the scientist. This criterion asks how much of that work each platform does itself, and how well it connects to the systems a research organization already runs.

Biomni Lab runs a wide range of computational biology

Biomni Lab is a capable analysis environment. Phylo lists 100+ pre-built workflows from early discovery to the clinic. They include single-cell analysis, GWAS fine-mapping and Mendelian randomization, virtual screening, de novo binder design and ADMET prediction. Its Ginkgo Bioworks case study reports cutting more than 10 cell-painting and transcriptomic analyses from weeks to hours.

Those workflows center on running computational analyses with the tools and datasets inside Biomni Lab. The platform is strongest when an agent can answer the question from data it can load and process.

EMET runs the biology and connects it to the systems your lab runs on

Scientists spend a large share of their week finding and cleaning data scattered across hundreds of sources. EMET carries 200+ proprietary scientific skills and 250+ agents and workflows. They cover the preclinical journey from target identification and validation through lead optimization, safety and IND-enabling studies. EMET queries live public databases and cites every claim to a PMID, database record or URL.

The platform does the analysis too. It runs bioinformatics pipelines such as FastQC and RNA-seq differential expression inside the conversation. It also retrieves ADMET data and visualizes protein structures. Through a lab-in-the-loop design, it connects to automated lab infrastructure and acts in the lab.

EMET connects live to your internal systems and databases with no data ingestion required. Every deployment also includes custom agents built around the customer's therapeutic areas and internal workflows.

Biomni Lab runs a broad set of analyses. EMET runs the preclinical science and ties it to the systems where your organization records its work.

Criterion 3: The service model that drives adoption

A platform only changes research if scientists fold it into their real projects. For most organizations, that depends on someone fitting the tool to each team. This criterion asks who does that work.

Phylo grows through broad, product-led adoption

Phylo has built a large user base quickly. It reports 60K+ scientists at 6K+ life science organizations using Biomni Lab, and anyone can start with the hosted product online. Pharma companies such as Ono Pharmaceutical and Chugai have signed enterprise collaborations to deploy it.

Phylo's public materials describe the product, its security and its enterprise deployment options. They don't describe a dedicated scientific team assigned to each account.

BenchSci embeds a dedicated scientific team in every account

BenchSci staffs each deployment with PhD scientists who know your biology and your therapeutic areas. They work alongside your researchers and build to your stack. The team runs training, manages the change, and builds the connectors, agents and workflows specific to your organization.

That team stays on as a scientific partner after launch. Its job is to turn the way your programs actually run into workflows your scientists use every day.

Individual scientists can adopt either platform quickly. BenchSci adds the people who make it work across a whole organization.

Criterion 4: Model flexibility and vendor independence

The best model for a scientific task changes every few months. A platform worth committing to has to adopt the next one without making customers rebuild their work. On this criterion, BenchSci and Phylo are on even footing.

Phylo matches each task to the best-performing model

Phylo doesn't tie Biomni Lab to one model provider. It describes the platform as model-agnostic, matching each task to the model that gives the best result at the best price. That lets customers use top-performing models and keep costs in check.

BenchSci is model-agnostic and routes every task to the best model

EMET builds no model of its own, so each step of a question can go to whichever of its 40+ models performs best on it. The platform uses frontier models for open reasoning and specialized models such as ESM-2 and AbLang2 when the question is about sequence or structure. New models join the roster without customers rebuilding their workflows.

Neither platform ties your research to a single model provider. Both can adopt a better model as soon as one appears.

Criterion 5: Whether your AI partner also develops drugs

An R&D platform sees a program's targets, assay results and failed experiments. So it's fair to ask whether the vendor could ever turn what it learns into a drug program of its own. The two platforms stand on the same ground here too.

Phylo is an AI research lab that sells to drug developers

Phylo describes itself as an applied AI research lab building technology for biologists. It sells Biomni Lab to drug developers such as Ono and Chugai and describes no drug pipeline of its own. It also states that customer data is never used to train models, and that Biomni Lab can be deployed into the customer's own VPC.

BenchSci has committed never to develop its own drugs

BenchSci makes the same stance a permanent commitment: it will never discover its own drugs. Your data is isolated to your organization. It is never pooled with another customer's, never added to a shared graph and never used to train shared models. EMET is SOC 2 Type II certified and encrypts data with AES-256 at rest and TLS 1.2+ in transit.

For a team weighing IP risk, both vendors sit on the tooling side of the industry. What separates them is the data, the biology-specific workflows and the service model.

Deciding between EMET and Biomni Lab

Biomni Lab is a strong choice for scientists who want an agent that runs a wide range of computational biology on public tools and data. It grew from a widely used open-source project, and it covers analyses from single-cell work to virtual screening. It matches EMET on model flexibility and on neutrality.

EMET leads on the other three criteria. It reasons over 16M closed-access papers and 85M curated reagent and model-system data points that no other platform holds. It connects that biology to the internal systems where your organization records its work. And BenchSci embeds a standing team of PhD scientists in every account to build workflows around your programs.

If your programs depend on evidence that public sources don't contain, EMET by BenchSci is built for that work. Book a demo to see it on your own targets.