All WFO demos
ConceptIllustrative scenario · not a recording

Keep every certificate renewed

Specialist agents diagnose a failed renewal. After approval, Workflow Orchestrator renews the certificate across eight services, and the same pattern scales to a crypto-agility campaign.

Agent investigation · orchestrated renewal See the setup

Certificates now expire faster than people can renew them

Validity periods keep shrinking, from over a year towards under two months by 2029. That is roughly eight times as many renewals per year, for hundreds of internal and external certificates spread over many sites. Renewing by hand does not scale, and an expired certificate is an outage.

Rounded, illustrative scale for a large organisation.

  1. Problem

    Failures stay hidden until it is almost too late

    Automatic renewal fails quietly: a rotated DNS credential, a changed firewall, an owner who left. Someone then spends hours finding out who owns the certificate, where it runs and why renewal stalled.

    With this setup

    Agents investigate in plain language

    Specialist agents read the CMDB, the scanner and WFO at the same time and explain the cause, the owner and the targets. They are read-only: they propose, they do not change anything.

  2. Problem

    Every renewal touches many systems

    A renewal needs an approved change, a key pair, a CA order, a DNS challenge, a rolling deployment and a check. Scripts and tickets cover parts of it and break at the seams.

    With this setup

    Workflow Orchestrator runs it as one workflow

    After a person approves, WFO executes every step across the services involved. It deploys one target at a time, rolls back on a failed check and resumes where it stopped, with every step on record.

  3. Problem

    A policy change becomes a project

    A new algorithm policy, for example to prepare for post-quantum cryptography, means finding and renewing every affected certificate by hand.

    With this setup

    Crypto-agility is a product parameter

    The algorithm and key size are set once in the product. The agent selects the certificates that still use the old profile and groups them in waves; WFO runs the waves, so a policy change is a campaign instead of a project.

Hypothetical scenario. This is not a recording or a client case. The certificate, the alert and the figures are invented, the services are examples and nothing here was run

The exact request

The certificate for the customer portal expires in 12 days and automatic renewal failed. Why did it fail, who owns it and where is it installed? What is the safest way to renew it, and does it meet our new algorithm policy? Make no changes.
  1. Alert1 / 9

    A scanner raises an alert

    The customer-portal certificate expires in 12 days. Last night’s automatic renewal failed.

    • Scanner
  2. You2 / 9

    You ask in plain language

    Why did it fail, who owns it, what is the safest way to renew, and does it meet the algorithm policy?

    • LibreChat
  3. Agent3 / 9

    Three specialists investigate at once

    The Inventory agent finds the owner and three targets behind two load balancers. The PKI agent reads the scanner: the last renewal stalled at the DNS challenge. The WFO agent sees the renewal workflow suspended at its validation step, not lost.

    • CMDB
    • Scanner
    • WFO
  4. Agent4 / 9

    The planner explains and proposes

    A rotated DNS API credential broke the challenge. The certificate also uses the old algorithm, so the proposal renews with the agile profile. Nothing has changed yet.

  5. You5 / 9

    You approve

    One click approves the proposal. The agent starts the renewal workflow in WFO and steps back.

    • LibreChat
    • WFO
  6. WFO6 / 9

    WFO orchestrates the renewal

    1. Open a change and suspend until it is approvedITSM
    2. Read owner and deployment targetsCMDB
    3. Generate the key pair and CSRVault / HSM
    4. Order, answer the DNS challenge, validatePublic CA · DNS
    5. Store the certificate and chainVault / HSM
    6. Deploy one target at a time; roll back if a check failsDeploy targets
    7. Verify the TLS handshakeScanner
    8. Update the record and close the changeCMDB · ITSM

    If a target fails its check, WFO restores the previous certificate there and pauses. After the fix, the workflow resumes at that step instead of starting over.

  7. Result7 / 9

    Renewed everywhere, with a trail

    The subscription shows the new expiry, algorithm and owner. Every step, retry and approval is on record.

  8. Agent8 / 9

    The agents confirm the result

    The WFO agent reads the finished subscription: new expiry, algorithm and owner, with the change closed. The PKI agent asks the scanner and sees the new certificate on all three targets. The Inventory agent finds the CMDB record updated.

    • WFO
    • Scanner
    • CMDB
  9. Agent9 / 9

    The planner answers you in chat

    The certificate is renewed on all three targets with the agile profile. The planner lists what ran, what the scanner confirmed and what it did not touch, and links the audit trail.

    • LibreChat

Agents decide. WFO orchestrates.

Hover, focus or tap a node to explore

One planner. Three specialists. One orchestrator.

A2A connects the agents; MCP exposes their tools. The agents investigate and propose. After approval, WFO runs the renewal across eight services: change approval, keys, public and internal CAs, DNS, deployment targets and verification. Select a node to follow its connections.

Why orchestration, not a script

  • A lifecycle per certificate

    Each certificate is a subscription with states such as requested, active, renewing and retired. You can ask what is expiring and where it stands.

  • Resume where it stopped

    Steps are durable. A failed DNS challenge or a busy target pauses the workflow; it continues at that step after the fix.

  • Approvals as a step

    A change request is part of the workflow, and the workflow waits for it.

  • Rolling, with rollback

    Targets are updated one at a time, verified, and restored if a check fails.

  • One audit trail

    Who approved, what ran, when and with which result, recorded for every certificate.

  • Campaigns, not just renewals

    The same workflow runs in batches, so a policy change such as a new algorithm is a campaign instead of a project.

Validation and execution boundaries

  • The agents are read-only: they investigate and propose.
  • A person approves before anything changes. WFO then executes and records every step.
  • Intent (WFO, CMDB) and evidence (scanner) stay separate. A renewal is only done once the scanner confirms the new certificate.

Explore the implementation

The agent and orchestrator stack is open source. The certificate product, workflows and integrations in this scenario are examples of what can be modelled on top of it.

How many renewals will you face when certificates live for months, not years?

Discuss your certificate lifecycle