Optional reference · Back to the talk
Optional referenceRevised October 2026

Working Smarter with Agents
Reference.

Supporting procedures and worksheets. Return to the main talk

David Budáč
Optional reference · Back to the talk
2
In practice

Compare current offers using your workload

RecordQuestion
Price and usage limitsWhich work is included, capped or separately billed?
Authentication and automationDoes the intended client and deployment use a supported path?
GovernanceWhich retention, identity and audit controls are configured?
Actual workloadWhat is the cost at your observed volume and success rate?

Verify the current plan pages before purchase. Claude plans

Optional reference · Back to the talk
3
In practice

Verify the configured boundary

  • Identify the provider, region and any cross-region inference routing.
  • Confirm the relevant retention, logging and identity settings.
  • Check model availability, quotas, pricing and your organization’s agreement.

A cloud brand alone does not establish residency or prevent data from leaving a region.

Optional reference · Back to the talk
4
In practice

Evaluate tools against the workflow requirements

  • Herdr — inspect monitoring and coordination behavior.
  • Conductor — inspect workspace and review behavior.
  • Symphony — inspect task lifecycle and tracker support.
  • Sandcastle — inspect its configured environment boundary.

Reference links, not a current product comparison. Check supported versions and integrations.

Optional reference · Back to the talk
5
Publication reference

Package, test, then share

my-plugin/ .claude-plugin/plugin.json skills/invoice-check/SKILL.md 1. Give the manifest a name and version. 2. Test locally: claude --plugin-dir ./my-plugin 3. Review contents and pin the version to distribute. 4. Publish through your team’s approved channel.

Plugin overview · Marketplace setup

Optional reference · Back to the talk
6
In practice

Record what you are comparing

LayerRecord before running
ModelExact identifier, effort and generation settings.
HarnessVersion, prompts, tools, permissions and retry logic.
Routing serviceProvider selection, fallback behavior and data path.
Local runtimeQuantization, hardware, context and concurrency.

A changed harness can change the result even with the same model.

Optional reference · Back to the talk
7
In practice

The independence assumption matters

If each trial succeeds independently with probability p: P(at least one success in k trials) = 1 − (1 − p)^k P(all k trials succeed) = p^k For p = 0.8 and k = 3: 0.992 versus 0.512.

These are mathematical examples, not measured model results. Correlated failures can invalidate the simple model.

Optional reference · Back to the talk
8
In practice

Active parameters are not the RAM requirement

  • Stored weights depend on total parameters, representation and quantization.
  • Inference also needs a KV cache, activations, runtime buffers and overhead.
  • Offloading can move weights to CPU memory or disk, trading memory for transfer time.

Approximate weight storage: parameters × bits per weight ÷ 8. This is a lower-bound estimate, not a fit guarantee.