Agent operations & governance

The control plane for agents you can trust.

See what every agent did, what it cost, and where it paused for a human. Policies and audit trails keep data and actions inside guardrails — your orchestration framework stays under the hood.

Works with CrewAI · LangGraph · custom orchestration

run_8f3a21c invoice-reconciler Awaiting approval Completed Stopped 5.2s · 18,442 tok · $0.0417.9s · 19,610 tok · $0.0445.2s · 18,442 tok · $0.041
  1. 14:02:11.204 Scheduled trigger cron · weekdays 14:00
  2. 14:02:11.390 postgres.query 142 records 0.8s · $0.002 0.8s · $0.002
  3. 14:02:12.212 Match invoices to bank feed LLM · 12,904 tokens 3.1s · $0.031 3.1s · $0.031
  4. 14:02:15.318 policy/spend-limit $4,280 exceeds $1,000 threshold triggered triggered
  5. 14:02:15.322 Approval: post 3 entries Waiting on finance-lead 2m 2m
  6. 14:02:15.322 Approved: post 3 entries Selin K. · finance-lead 1m 48s 1m 48s
  7. 14:02:15.322 Rejected: post 3 entries Selin K. · finance-lead 1m 48s 1m 48s
  8. — gmail.send_message finance@acme.com · run summary queued queued
  9. 14:04:03.117 gmail.send_message finance@acme.com · run summary 0.4s · $0.001 0.4s · $0.001
  10. — gmail.send_message skipped · run stopped skipped skipped
Approval required
Approved
Rejected
Post 3 adjusting entries to ERP
Amount
$4,280.00
Policy
spend-limit > $1,000
Approver
finance-lead · Selin K.
Requested
2 min ago
+ JE-1042 bank_fees 1,120.00
+ JE-1043 fx_adjustment 2,410.00
+ JE-1044 rounding 750.00
Approved by Selin K. · entries posted
Rejected by Selin K. · agent notified
Decision is written to the audit log
audit · evt_31d9 · approval.granted
audit · evt_31d9 · approval.denied
Connects to
  • OpenAI
  • Anthropic
  • CrewAI
  • LangGraph
  • GitHub
  • Postgres
  • Google Workspace
01 / Features

Built for how you ship.

From first prototype to production traffic, keep agents aligned with your tools, policies, and stakeholders.

Agent operations

Run and observe agents and crews from one cockpit: live status, handoffs, and execution history you can replay step by step.

tracesreplayhandoffs

Tools & integrations

Connect the systems you already use so agents can read context, take action, and report back — every connection scoped by policy.

scoped credentialsper-tool limits

Human-in-the-loop

Pause before spend or sensitive actions. Approvals route to the right owner in one operations view instead of scattered inboxes.

approval queuesownerstimeouts

Governance & compliance

Policies, compliance checks, and audit trails that answer what ran, who approved it, and whether the rules were enforced.

policiesaudit logexport
02 / How it works

From signup to agents your team can trust.

  1. 1

    Connect your workspace

    Sign up and link the data sources and tools your team relies on.

    → workspace connected
  2. 2

    Configure agents

    Define personas, constraints, and tools so behavior matches your standards.

    → policies active
  3. 3

    Run and refine

    Ship to production, monitor outcomes, and iterate as your stack evolves.

    → every run traced and costed
03 / Security & compliance

Answer the security review before it’s sent.

Built for teams selling to regulated customers — in Türkiye, the EU and beyond.

  • KVKK aligned

    KVKK-aligned processing

    Data processing agreement and VERBİS-ready records for personal data processed in Türkiye.

  • GDPR aligned

    GDPR-ready by default

    EU data residency, DPA with SCCs, and deletion / export requests handled from the dashboard.

  • SOC 2 in progress

    SOC 2 Type II

    Audit underway. Security questionnaire and policies available on request.

  • TLS · AES enforced

    Encryption everywhere

    TLS 1.3 in transit, AES-256 at rest. Provider keys stored in a dedicated secrets vault.

  • AUDIT enforced

    Immutable audit log

    Every run, approval and policy decision recorded append-only and exportable for review.

04 / Pricing

Start free. Pay when agents ship.

Model and tool costs stay on your own provider keys — KolayStack meters them, it doesn’t mark them up.

Free

To be announced

For prototypes and side projects.

Get started
  • 3 agents
  • 1,000 runs / month
  • 7-day trace retention
  • Community support

Enterprise

To be announced

For regulated teams and larger fleets.

Talk to us
  • Unlimited agents
  • EU data residency
  • DPA, KVKK agreement and SLA
  • Dedicated support
05 / FAQ

Questions, answered.

Something else? hello@kolaystack.com

Do I have to replace CrewAI or LangGraph?

No. KolayStack sits on top of your orchestration framework and adds traces, approvals, budgets and policies without changing your agent logic.

How long does setup take?

Most teams see their first trace within an hour: install the SDK, wrap your agent, run it. Policies can be tightened over time.

Where is our data stored?

In the EU, processed in line with GDPR and KVKK.

How do approvals work?

When a policy triggers, the agent pauses and an approval request appears in the queue and by email. Once approved it resumes where it stopped; if rejected, the agent is told why.

Are model costs included in the price?

No. You bring your own provider keys; KolayStack meters token and tool spend per agent, run and team and enforces the budgets you set.

What happens when I hit the Free plan limits?

We warn you as you approach them. Runs beyond the limit are paused, never billed — there is no charge until you add a card.

Ready to run your stack?

Create an account to explore the platform, or log in to continue where you left off.