How we make human microbiology computable
How we make human microbiology computable
How we make human microbiology computable
Outpost is developing foundation models trained on purpose-built data combined through our continuous lab-in-the-loop.
Outpost is developing foundation models trained on purpose-built data combined through our continuous lab-in-the-loop.
The Foundation Model
The Foundation Model
The Foundation Model
We're building a new class of foundation models for human microbiology - starting with Waypoint, our open-source model and the largest of its kind. It encodes broad patterns across microbial ecosystems, giving teams a pre-trained foundation to fine-tune for their own cohorts, indications, and research questions.
We're building a new class of foundation models for human microbiology - starting with Waypoint, our open-source model and the largest of its kind. It encodes broad patterns across microbial ecosystems, giving teams a pre-trained foundation to fine-tune for their own cohorts, indications, and research questions.
Atlas
A 539,308 data point pretraining corpus curated from public data through a rigorous quality pipeline , the largest dataset assembled for microbiome modeling.
A 539,308 data point pretraining corpus curated from public data through a rigorous quality pipeline , the largest dataset assembled for microbiome modeling.
Waypoint
A family of open-source GPT-style causal language models from 6M to 170M parameters, trained on Atlas.
A family of open-source GPT-style causal language models from 6M to 170M parameters, trained on Atlas.
Compass
An eight-task benchmark spanning biome classification, drug-microbiome interactions, drug degradation, infant gut development, and more.
An eight-task benchmark spanning biome classification, drug-microbiome interactions, drug degradation, infant gut development, and more.
The Data Engine
The Data Engine
The Data Engine
We combine state-of-the-art pipelines for ingesting and harmonizing existing public data with proprietary wet lab data. Our in vitro, high throughput compositional and functional readouts from human-derived microbial communities capture how these ecosystems respond to drugs, ingredients, and formulations under controlled conditions.
We combine state-of-the-art pipelines for ingesting and harmonizing existing public data with proprietary wet lab data. Our in vitro, high throughput compositional and functional readouts from human-derived microbial communities capture how these ecosystems respond to drugs, ingredients, and formulations under controlled conditions.
Expanse
In vitro microbiome screening and ML validation service, based at LabCentral in Cambridge, MA. Collaborators use it to substantiate strain or formulation claims, triage candidate libraries, and characterize donor variance.
Ingestion Pipelines
A cloud-based engine that harmonizes multi-omic microbiome data across sites, sequencing technologies, and protocols. Produces ML-ready corpora and rapidly accelerates bioinformatics workflows.
Secure & Flexible Deployment
Available as cloud or on-premise deployment, depending on your compliance requirements and existing bioinformatics and ML infrastructure.
Expanse
In vitro microbiome screening and ML validation service, based at LabCentral in Cambridge, MA. Collaborators use it to substantiate strain or formulation claims, triage candidate libraries, and characterize donor variance.
Ingestion Pipelines
A cloud-based engine that harmonizes multi-omic microbiome data across sites, sequencing technologies, and protocols. Produces ML-ready corpora and rapidly accelerates bioinformatics workflows.
Secure & Flexible Deployment
Available as cloud or on-premise deployment, depending on your compliance requirements and existing bioinformatics and ML infrastructure.
Explore or get in touch
Explore or get in touch
Sign up to our newsletter!
Sign up to our newsletter!
©OUTPOST BIO 2026
BAKED BY THE SOURDOUGH
Legal Pages
©OUTPOST BIO 2026
BAKED BY THE SOURDOUGH