[ Pricing ]

Pay for the work agents do.Not for how many people log in.

Every plan is a flat monthly fee with a monthly credit allotment, pooled across your organization. Every plan includes both agent types, the full agent catalog and all 15 industry packs. Plans differ by execution capacity and operating mode, not by which agents you can use.

Start free, 2,000 credits Talk to us about Enterprise

All prices in USD at $0.01 per credit. Prices shown are per month, billed annually. Monthly plans are available with no commitment.

Run cost

Credits · pooled across your organization

$0.01/credit

No per-user fees · pause cleanly at zero

Free Trial
Free
2,000 credits, one time
2,000 credits
Evaluate one or two real workflows in your own Azure DevOps environment.
- No card required
- Founding customers get 2,000 credits
- Both agent types and the full catalog
- Output verification and job logs
- Hybrid model selection
Start Free Trial
Starter
$500/month
/month, billed annually ($6,000/yr)
50,000 credits / monthSave 9% vs monthly
A single team starting with manual and event-triggered workflows.
- Manual and event-triggered execution
- Single-agent mode on a hosted LLM
- Execution Agents at a flat 4 credits a run
- All agents, skills and industry packs
- Audit logs, approvals and output verification
- 1 priority support call a month
Start with Starter
Pro
Most popular
$2,000/month
/month, billed annually ($22,000/yr)
220,000 credits / monthSave 17%, 1 month free
Most popular. Growing teams that want full autonomous coverage.
- Everything in Starter
- Scheduled (autonomous) execution
- Multi-agent orchestration for high-value workflows
- Hybrid model selection per agent
- 99.5% SLA
- 4 priority support calls a month
Book a demo
Enterprise
$5,000/month
/month, billed annually ($50,000/yr)
600,000 credits / monthSave 24%, 2 months free
Organization-wide deployment across multiple teams.
- Everything in Pro
- BYO LLM: your model and key
- 99.9% SLA
- Dedicated customer success manager
- Unlimited priority support calls
- Negotiated high-volume credit pricing
Contact sales

All prices in USD. Prices are per month, billed annually. Monthly plans available with no commitment. Every plan includes both agent types, the full catalog, and every industry pack.

[ Compare plans ]

Capabilities by plan.

Plan comparison
Capability	Free Trial	Starter	Pro	Enterprise
Manual execution	✓	✓	✓	✓
Event-triggered execution	✓	✓	✓	✓
Autonomous (scheduled) execution	✓	—	✓	✓
Multi-agent orchestration mode	✓	Single-agent mode	✓	✓
Execution Agents	✓	✓	✓	✓
Hybrid model selection	✓	Hosted LLM	✓	✓
BYO LLM (your model and key)	—	—	—	✓
Output verification	✓	✓	✓	✓
Audit logs and governance	✓	✓	✓	✓
Priority support calls / month	—	1	4	Unlimited
SLA	—	—	99.5%	99.9%
Dedicated CSM	—	—	—	✓

[ How credits work ]

Two agent types. Two very different costs.

Assisted Autonomous Agents use a language model each run, so they consume credits by the work performed. Execution Agents run deterministic code with no model call at run time, so every run costs a flat 4 credits, about $0.04, however many Azure DevOps calls the rule makes. Put high-frequency, rule-shaped work on Execution Agents and spend credits where reasoning adds value.

ILLUSTRATIVE · A MONTH COMPRESSED INTO EIGHT SECONDS

Execution Agent run
A deterministic rule executes on an event or schedule. No model call at run time.
4 credits, flat
Light Assisted Autonomous run
Reads one work item and its links, and drafts a short output such as acceptance criteria.
Low
Deep Assisted Autonomous run
Investigates across items, code, pipeline logs or test results, and produces an analysis or report.
Scales with context
Premium-model run
Any Assisted Autonomous run on a top-tier model. Same work, higher token rate.
Highest per-run rate

What a run consumes

A small invocation credit to start the run: lowest for manual requests, higher for event-triggered and scheduled runs.
Model calls and tokens: the context read and the output produced.
Tool calls against Azure DevOps.
Orchestration steps as the agent decides what to do next.
Execution Agents created from a natural-language prompt use standard model credits once, at creation. Write the code yourself and there is no model cost at all.
Scoping an agent tightly, to a project, area path or work item type, is a cost control as well as a governance one.

[ Spend controls ]

Predictable by design.

Nothing runs that your team did not configure. Every run starts from a manual request, an Azure DevOps event you selected, or a schedule you defined, and four controls keep consumption visible and bounded.

CONTROL 1
Real-time visibility
The Jobs view shows consumption by agent and module over the period you select, plus used and remaining credits for the organization. The dashboard shows credits left and the runway at the current burn rate.
CONTROL 2
Consumption alerts
Administrators set a utilization threshold. An alert email goes out as consumption approaches it, with time to adjust agents or add capacity.
CONTROL 3
A hard ceiling
At zero credits, agents pause until the next cycle or a top-up. Nothing is billed beyond your plan unless you turn on auto top-up, and then you set the monthly cap.
CONTROL 4
Per-agent model choice
Each model carries its own credit rate, visible at selection. Lighter models conserve credits on routine agents; the premium model is there when the task warrants it.

Top-ups

More capacity when a month needs it.

Buy top-up credits at any time at 1.5× your in-plan credit rate, so the effective top-up price is lowest on higher tiers.
Top-up credits never expire. They carry forward on top of your monthly allotment until used.
Top-ups are non-refundable.
Your monthly allotment is provisioned fresh at the start of each billing cycle.
Auto top-up refills the pool below a threshold you set, within a monthly spending cap.
Buying top-ups every month? The next tier is usually better value, and we will tell you so.

Bring your own LLM

Your model, your key, on Enterprise.

Enterprise plans can bring their own model and key. The model-token portion of each run is then charged at a small flat infrastructure rate instead of the model's rate, and you pay your provider directly for the reasoning compute. The invocation credit, model-call credits and tool calls against Azure DevOps are charged as they would be on a hosted model. Token-related credits fall sharply and total credits per run fall too, though less, so model both line items when you compare cost. You also control the model, region and data path directly, which is often the deciding factor in regulated environments.

[ What you reclaim ]

Real workflows, organized by who benefits.

Starting points, not fixed templates. Both LLM-based Assisted Autonomous Agents and deterministic Execution Agents can serve any of these, alone or combined.

Product Owner / BA
- Triage incoming bugs by area, severity and assignee
- Draft sprint reports and release notes from completed work items
- Keep wiki pages current from work item descriptions and acceptance criteria
- Flag stale backlog items for cleanup or re-prioritization
- Catch new items missing acceptance criteria, descriptions or links
Developer
- Review pull requests with inline suggestions before human review
- Link commits and PRs to their work items
- Diagnose CI pipeline failures and propose code fixes
- Draft unit test stubs from new function signatures
- Flag style, complexity or security concerns when a PR opens
DevOps / SRE
- Fix common pipeline failures: environment variables, dependencies, transient flakes
- Investigate incidents with several agents correlating logs, commits and past incidents
- Tag build artifacts by branch, version or release status
- Clean up stale build artifacts and pipeline runs on a schedule
- Keep on-call dashboards current with build, deploy and incident status
QA / Test Engineer
- Draft test cases from the acceptance criteria on new stories
- Link test cases to requirements and surface coverage gaps
- Create defects from failed test runs with full context
- Spot flaky tests from recent result patterns
- Prepare test summary reports for sprint review and release sign-off
Engineering Manager / Lead
- Weekly team status reports from work item activity and commits
- Blockers and risks across the team's items, ready for standup
- Engineers with high WIP or sprint overcommitment
- Velocity, scope creep and on-time delivery trends
- PRs open beyond your team's review SLA
Compliance / Audit (V-model orgs)
- Traceability reports from requirement to design, code, test and deployment
- Check every code change has a linked work item, reviewer and traceable approval
- Audit branch policies, mandatory reviewers and protected-branch enforcement
- Assemble validation evidence packs for regulatory submissions (FDA, IEC 62304, ISO 26262)
- Detect missing or outdated SOPs, design documents and validation records

A typical mid-size engineering team of ~50 engineers replaces several hundred hours of manual DevOps toil a month across triage, code review, testing, incident response and reporting. At a $75/hr engineer cost, that is tens of thousands of dollars of monthly engineering capacity returned to product work, for a Pro plan that costs less than one engineer-week. Your result depends on workflow volume, complexity and which agents you configure; ask us for an estimate based on your own patterns.

[ Founding Customer Program ]

50 places. Built with the product team.

We are accepting a limited cohort of 50 founding customers who want to deploy a real Azure DevOps workflow and shape what comes next.

Apply to the program

25% extra credits every month for the first year of your subscription
A double-size free trial: 2,000 credits, no card
Priority access to the product team
Direct influence on the roadmap
An optional case study, only with your permission

[ Trusted by Microsoft and industry leaders ]

[ FAQ ]

Pricing questions.

$0.01. Starter's $500 a month buys 50,000 credits. Pro and Enterprise include more credits per dollar, and Enterprise can negotiate high-volume credit pricing.

No. Credits are pooled across your organization. Everyone your Azure DevOps permissions allow can configure and trigger agents, so you can add reviewers, approvers and auditors without paying per seat.

Agents pause cleanly and no new runs start, so there is no surprise overage. Buy top-up credits, upgrade your plan, or wait for the next cycle. Adding credits resumes agents immediately. With auto top-up on, agents keep running within the monthly cap you set.

No. Top-ups carry forward until used and stay available for twelve months even if you cancel. They cost 1.5× your in-plan rate and are non-refundable.

They run deterministic code with no model call at run time, so every run is a flat 4 credits, about $0.04. A rule that fires on every work item update costs the same each time, a fraction of a reasoning run.

Yes. The free trial is 2,000 credits, no card required, run against one or two real workflows in your own environment. Founding customers get 2,000. Review consumption per agent afterwards to see which agents belong on Execution Agents.

Prices shown are per month, billed annually: Starter saves 9%, Pro includes one month free and Enterprise two. Monthly plans are available with no commitment.

Starter suits one team running manual and event-triggered agents. Pro adds scheduled execution and multi-agent orchestration for growing teams. Enterprise is for organization-wide rollout with BYO LLM, a 99.9% SLA and a dedicated success manager. Every plan has every agent and industry pack.

Size one real workflow with us.

Bring your event volumes and the agents you would deploy. We will model your consumption live, show which work belongs on Execution Agents, and map it to a monthly allotment. Or start the free trial today: 2,000 credits, no card.