Enterprise
AI in your delivery path, on your terms
TaskForce keeps a human accountable at every checkpoint and can run entirely on your own infrastructure — so you get agent leverage without giving up control, attribution or ownership of your data.
Self-hosted
Run TaskForce inside your own walls
Authentication in front, your systems and models around it, and everything in between left auditable. Nothing has to leave your network.
- Self-host the whole platform on your own network, with Docker.
- Keep model inference local with Ollama — prompts and data never leave.
- Your infrastructure, your models, your data — no exceptions.
Governance
A human on every consequential step
Approval isn't a settings toggle. It sits where a decision compounds — and it leaves a record you can read months later.
- A human approves each consequential checkpoint before the run continues.
- RBAC and per-project access scope exactly who can do what.
- Every approval and model call is attributable — who, what, when, which model.
Postgres over Mongo — tenant isolation
Proposed by the CTO agent, with the rejected option and its reason kept alongside.
Illustrative — a governed decision as it's recorded.
Built for your environment
Fits the infrastructure you already run
Connect the identity provider, repositories, trackers and model providers your organization already uses. The platform itself runs on modest hardware — the models are the heavy part.
Identity
Keycloak · OIDC · SAML
Runtime
Docker · Docker Compose
Models
Anthropic · OpenAI · Ollama · local
Data
PostgreSQL · S3 / MinIO
Integrations
GitHub · GitLab · Linear · Slack · +125
- 2+
- vCPU for the platform
- 4 GB
- RAM for TaskForce itself
- 16 GB+
- RAM to run local models (Ollama)
- GPU
- recommended for the larger models
Deploys with Docker Compose today; Kubernetes is on the roadmap. Real requirements scale with the local models you choose to run.
Enterprise controls
Control at every layer
- Identity OIDC / SAML / IdP federation
- Access RBAC + project-level permissions
- AI Hosted or local models, per step
- Data Private infrastructure + retention controls
- Delivery Human approval at every checkpoint
- Audit Run, approval and model-call history
- Deployment Self-hosted (Docker)
- Security Security review + DPA
- Operations Dedicated support + SLA
Security & compliance
No certification claims we can't substantiate
TaskForce documents the controls, data flows and security roadmap so your team can evaluate the current state before deployment.
Open the trust centerGDPR
Technical controls for export, erasure and retention are in place.
SOC 2
Not certified. SOC 2 groundwork is on the roadmap.
DPA
Available on request, with our subprocessor list.
Adoption
From security review to first run
A clear path, so adopting TaskForce is a project with steps — not a leap of faith.
- 01
Architecture review
We map identity, data flows, connectors and model providers.
- 02
Deployment
Your team deploys TaskForce in your environment, with our assistance.
- 03
Security validation
Review controls, permissions, data flows and retention.
- 04
First workflow
Start with one delivery workflow, validate the model, and expand.
Common questions
What a security team asks first
- Can we run TaskForce fully self-hosted?
- Yes — self-hosting is a first-class Enterprise deployment. TaskForce runs inside your own infrastructure with Docker, and local models through Ollama keep prompts and outputs on your network.
- Do our prompts or data train your models?
- No. TaskForce does not use your data to train its own models. On hosted providers, data handling follows the provider and your configured terms; local models keep inference inside your infrastructure.
- Why not just assemble this ourselves?
- You could wire model providers, an approval system, audit logs and connector permissions together yourself. TaskForce puts those behind one governed delivery workflow — so your team ships instead of maintaining an orchestration layer.
- How does access control work?
- SSO and SAML through Keycloak, RBAC with per-project permissions, and connector scopes you grant explicitly. Agents can only invoke what you allow.
- What about compliance?
- GDPR-oriented technical controls (export, erasure, retention) are in place. SOC 2 groundwork is on the roadmap; a DPA is available on request. We document the current state rather than claim certifications we can't substantiate.
- How do we get started?
- It's a guided project: architecture review, deployment with our assistance, a security validation pass, then a first workflow. Book a call and we'll map it to your environment.
Bring TaskForce to your organization
Book a call and we'll map the architecture, the controls and a deployment that fits your environment.