Expertise
Operations that survive the restart
TechFabric is a Temporal partner. We put durable execution under the operations that must not half-complete: a payment that charges but does not fulfil, a repossession that starts and is never recorded, a pipeline that dies mid-run. A step that fails at three in the morning resumes rather than restarts.
TechFabric is a Temporal partner. We build durable execution into APIs and long-running operations, with workflow history as the audit trail, and we run our own products on it.
Workflows and activities
The split that makes the rest work. Workflow code is deterministic and replayable, activities are where the world gets touched, and getting that boundary wrong is the source of almost every problem teams hit later.
- Deterministic workflow code, with time, randomness and IO pushed into activities
- Existing services called as activities, so nothing has to be rewritten to get history
- Idempotency keys on anything that charges, sends or provisions
- Versioning from the first release, because a running workflow outlives the deploy that started it
Long-running processes
Processes measured in days rather than seconds. An onboarding, a repossession, a claim, an agent run that waits on a person, all of which behave badly when modelled as a job.
- Timers and signals for waits that outlast any process holding them
- Human approval gates that survive a deploy and record who approved
- Child workflows for the parts that fan out and have to be tracked individually
- Continue-as-new for histories that would otherwise grow without bound
Failure and compensation
Designed at the start, when the alternative is cheap. Retrying is easy, and the hard question is what to undo when a step three stages back already committed to a system you do not control.
- Retry policies per activity, tuned to what the downstream system actually does under load
- Saga compensation for the steps that cannot be rolled back by a transaction
- Poison workflows surfaced and paused rather than retried into a wall
- Replay tests in CI, so a code change that would break history fails the build
Running it
Temporal Cloud or self-hosted, both of which we have taken to production. We have certified a durable path against Temporal Cloud including a workflow that survived a full application restart and completed from its original history.
- Namespaces, task queues and worker sizing set against the workload rather than the default
- Metrics on schedule-to-start latency, which is the number that tells you workers are short
- History and search attributes used for support, so an operator can answer a customer directly
- Data conversion and encryption where payloads carry anything regulated
What we build
- Durable workflows with complete, replayable execution history
- Existing APIs left in place and called as activities
- Human approval gates that survive a deploy
- Compensation paths designed in, not bolted on after the first incident
Accelerators doing the work.
We use Temporal in our own agent runtime and oversight console. The same durable path is what we bring to a client workflow.
How this is delivered
APIs & durable systems
Long-running operations that survive restarts and partial failure. Temporal under the lakehouse jobs, agent runs and approvals that must not half-complete.
FAQ
Questions we get asked
What problem does Temporal actually solve?
Any process with more than a few steps that must not half-complete. Temporal makes the workflow itself durable, so the process survives the infrastructure under it. The service line is /services/apis.
Do we have to rewrite our services?
No. The usual pattern is a workflow in front of the sequence that already exists. Your APIs become activities. That is how we migrate a job chain without a rewrite.
How is this different from a queue and retries?
A queue gives you delivery. It does not give you the state of a business process. Temporal keeps a complete, replayable history, so you can answer what happened without reconstructing it from logs.
Do you work with Temporal Cloud or self-hosted?
Both. We have certified durable paths against Temporal Cloud, including a workflow that survived a full application restart and completed from its original history. Fabric Tower at /accelerators/fabric-tower uses that path for missions that park at approval.
How can we help?
A technical conversation with a senior engineer. Bring the process that keeps half-completing.