Decawork: Pricing, Features & Best Use Cases

Our take: Consider Decawork if your company runs several AI agents and needs a gateway that checks every API request or MCP tool call against plain-English policies before it executes, accepting that access is via a demo with no public price.

Main limitation: No public pricing or self-serve signup; evaluation starts with booking a demo.

By the AI Tool Atlas editors. Based on the vendor's official documentation; we have not tested this product hands-on. What “documented overview” means

Who Decawork fits

Consider Decawork if your company runs several AI agents and needs a gateway that checks every API request or MCP tool call against plain-English policies before it executes, accepting that access is via a demo with no public price. Based on the vendor's documentation, not on a hands-on test.

Best for: Enterprises deploying multiple internal AI agents

Also a good fit for

Security and IT teams governing agent access
Companies needing audit trails for agent actions

Consider another tool if no public pricing; demo booking required

Or if

No self-serve signup or free tier shown
Requires integrating agent traffic through the Decawork gateway

Decawork pricing and usage limits

Prices checked Oct 1, 2026 on the official US pricing page. Official pricing ↗ Prices and limits change often.

Vendor pricing page re-checked automatically 11 days ago: no changes detected.

Free access and trial conditions

No free tier shown.

Pilots are mentioned in the privacy policy; no trial terms published.

Paid plans: what changes as you upgrade

Sales-led web

Custom pricing source (opens in a new tab)

For: Enterprises deploying many agents

Limits

No public pricing page (404)

Includes

Agent registry
Policy model

1 more

Approvals

Usage caps, credits and extra costs

Unknown.

What is Decawork, and how does it work?

Decawork sits between AI agents and company systems as a gateway for API requests and MCP tool calls. Each agent gets its own identity, and a small model trained for your company evaluates every proposed action against your written policies, the agent's current task and its past actions, deciding to allow, block or route it to a human. Access is scoped to the task and revoked when it ends, and every decision is logged with its policy context for audit. The product is a Y Combinator S26 company (Scale Technologies Inc.), claims SOC 2, and is sold via demo booking; no self-serve signup or public pricing is shown.

Decawork's official site: how the product presents itself. decawork.ai, captured Oct 1, 2026

What you provide and what you get back

What you give Decawork and what comes back, from its documentation

You provide

Agent API requests and MCP tool calls
Plain-English company policies
Current task and past actions

You get back

Allow, block or hold-for-human decision
Per-agent activity records
Audit log with policy context

Where you can use it

Platforms: Web
Account: Required
Intercepts: API requests and MCP tool calls
Decisions: Allow, block, or hold for human
Inputs to decision: Policies, current task, past actions
Compliance: SOC 2 claimed

Best use cases for Decawork

Scenarios are worked out by our editors from the documentation; the actual result depends on your input and plan.

Approving or blocking agent tool calls before they reach production systems

Who: Security and IT teams
You provide: Agent tool calls plus written policies
Expected result: Risky calls blocked before reaching production systems
Plan needed: Sales-led
Limitation: Requires routing traffic through the gateway source (opens in a new tab)

Giving agents task-scoped credentials behind a gateway

Who: Platform teams running agents
You provide: Agent identities and tasks
Expected result: Access scoped to a task and revoked when it ends
Plan needed: Sales-led
Limitation: Setup detail not public source (opens in a new tab)

Human-in-the-loop review for sensitive agent actions

Who: Compliance and people-ops owners
You provide: Sensitive actions such as HR record reads
Expected result: Held actions routed to a person for approval
Plan needed: Sales-led
Limitation: Approver workflow details not published source (opens in a new tab)

What Decawork does, and where it stops

Feature overview on the official site. decawork.ai, captured Oct 1, 2026

Per-agent identity and registry

Assigns each agent an identity and lists owner, team, tools and last activity in one registry fed by routed API and MCP traffic. Know which agents are running

Access: WorkspaceLimit: Only agents routed through the gateway appear source (opens in a new tab)

Company-specific policy models evaluating actions in real time

Trains a small model on your company that scores each proposed action against policies, the agent's current task and its earlier actions. Context-aware decisions

Access: Sales-ledLimit: Training and tuning process not detailed source (opens in a new tab)

Plain-English policy definitions

Takes written rules in everyday language and applies them as the basis of decisions, including conditions that require a human. No policy code

Access: Policies screenLimit: Wording precision affects results source (opens in a new tab)

Allow / block / hold-for-approval decisions before execution

Intercepts the call before execution and returns allow, block or hold; held items wait in Approvals for a person. Prevents bad actions rather than logging after

Access: Approvals screenLimit: Adds a gateway hop source (opens in a new tab)

Decawork: strengths and limits

Reasons to choose Decawork

Works with agents built on any framework via API/MCP routing Existing agents need no rebuild, only rerouting. documented source (opens in a new tab)
Human approval step configurable per policy Sensitive actions can pause for a person. documented source (opens in a new tab)
SOC 2 report available through a trust center Security reviews can ask for the report. documented source (opens in a new tab)
Agent registry keeps each agent's identity, owner and activity visible across API and MCP traffic An IT or security team can see which agents are running and who owns them. documented source (opens in a new tab)
Policies are set in plain English and a company-specific model approves actions in real time against current task and past actions Access can be allowed for one task, such as reading only an assigned hire's profile, and revoked afterwards. documented source (opens in a new tab)

Check before paying for Decawork

No public pricing; demo booking required Budgeting requires a sales conversation. documented source (opens in a new tab)
No self-serve signup or free tier shown You cannot self-evaluate. documented source (opens in a new tab)
Requires integrating agent traffic through the Decawork gateway Adds integration work and a dependency in the call path. documented source (opens in a new tab)
Early-stage product (YC S26) Fewer public references or docs. documented source (opens in a new tab)
Policies must be written and maintained by your team in plain English for the model to apply Approval quality depends on how completely your company's rules are described. documented source (opens in a new tab)

How to use Decawork for governing AI agent actions

Set up Decawork

Click Book a demo on decawork.ai or email the founders address to start
Connect your agents so API requests and MCP tool calls route through the Decawork gateway
Open Agents in the workspace to see each agent's owner, team, tools and activity
Open Policies and write rules in plain English, including when a person must step in

Example: governing AI agent actions

Write a policy such as Read only the assigned hire's profile
Give an agent a task, for example onboard a new engineering hire
Watch Approvals: the model weighs the policy, current task and past actions and shows Allow, block or hold
Approve or reject held actions as the human reviewer

Review and export the result

Confirm the Agents registry shows the agent as live with correct owner and tools
Review the decision log entry for the policy context behind each allow or block
Confirm access ends when the task completes

Decawork API

API access and separate billing

Decawork has no public API in its official documentation. /docs returned 404; gateway integration details not published

Privacy, data use and commercial rights

Summarized from the vendor's current policies; this is not legal advice and a vendor's badge is not an independent audit.

How your inputs and outputs are handled

Pilot and deployment data is governed by the customer agreement and scope; the site collects contact and basic usage info.

Kept only as long as needed for the purpose, relationship, security or legal needs.

Controls for business users

SOC 2 claimed on homepage

Policies: decawork.ai/privacy

Other tools to consider instead of Decawork

"Better suited" is based on documented features and pricing, not on a head-to-head test.

n8n: better suited to

Zapier: better suited to

Make: better suited to

Alternatives
Alternative	Pricing	Limitation
n8n	—	—	Compare
Zapier	—	—	Compare
Make	—	—	Compare

Explore Decawork further

Best Decawork alternativesTools for the same job, by need and starting price.Best AI writing and productivity toolsOur picks in this category, compared by price and free tiers.

If you are looking at Decawork

More productivity & writing tools

Guides and comparisons

Canva plans and prices

Decawork user reviews

No user reviews yet.

Used Decawork? Be the first to share your experience.

Write a review

Decawork FAQ

What does Decawork decide on each agent action?

It allows, blocks or routes the action to a human, judging it against your policies, the agent's current task and its past actions. source (opens in a new tab)

What traffic does it sit in front of?

API requests and MCP tool calls from agents built on any framework are routed through the Decawork gateway. source (opens in a new tab)

Is there public pricing or a free tier?

No. The pricing page returns 404 and the site only offers Book a demo. source (opens in a new tab)

How are policies written?

In plain English, including when a person must step in; Decawork trains a model specific to your company that applies them. source (opens in a new tab)

Does it sell the personal information it collects?

No. The privacy policy says personal information is not sold and is shared only with service providers such as Cloudflare, PostHog and Cal.com or as required. source (opens in a new tab)

Sources and review methodology

Official sources and verification dates

Decawork homepage homepage · checked Oct 1, 2026 · product description, agent registry demo
Decawork Privacy Policy policy · checked Oct 1, 2026 · collection, sharing

Not verified: Pricing; Latency and supported agent frameworks list; Data retention for gateway logs; Training use of customer data; API docs.

How we documented and maintain this page

Every fact about Decawork above comes from the 2 official pages listed here, last compared on Oct 1, 2026; we have not run Decawork ourselves, so it has no editorial score. How we check and re-check pages.

Article updated Oct 10, 2026; prices checked Oct 1, 2026. Outbound links to vendors are plain links; we do not earn commissions. How we review · Report an error