What a Claude Code rollout engagement includes and what it costs
A Claude Code rollout engagement is a short assessment, a secured setup, a pilot with one team and a plan to scale, usually sold as a fixed-price first phase followed by a monthly fee. The figures on this page are indicative and help you plan a budget. BlackCard quotes each rollout per project after a free 15-minute call. Seats or API usage for Claude are billed separately by Anthropic or your cloud provider.
What the engagement covers
Rollout engagements are assembled from the same building blocks. BlackCard covers each of them and agrees with you in writing which ones your first phase includes.
- Assessment: current tool use, repository inventory, security constraints and a shortlist of pilot teams.
- Secure deployment: choosing between a Claude plan, the API or a cloud provider, then connecting single sign-on and deploying managed settings.
- Repository configuration:
CLAUDE.mdinstruction files, shared permission rules, hooks and a reviewed.mcp.jsonfor the pilot repositories. - CI integration: wiring Claude Code into pull request workflows, for example through Anthropic's GitHub Action.
- Enablement: workshops, pair programming, office hours and an internal champions group.
- Measurement: delivery metrics agreed before the pilot and reviewed on a schedule.
Pricing models you will be offered
| Model | How it works | Suits |
|---|---|---|
| Fixed-price assessment or audit | A few days to two weeks of discovery ending in a written plan | Teams that want a second opinion before committing |
| Fixed-scope pilot or sprint | A defined set of repositories and one team, typically three to eight weeks | First rollouts where scope can be drawn tightly |
| Monthly retainer or program | A named team for a set number of days per month, often with a minimum term | Scaling across many teams after a pilot |
| Per-course or per-cohort training | A set fee for a workshop, a multi-day course or a cohort program | Teams with setup handled that need skills |
| Hourly | Time and materials against an agreed rate | Small, loosely defined support needs |
Indicative price points by type of work
Price follows scope. Where market figures are public, a one-time assessment tends to be listed in the range of about $5,000 to $12,000 and monthly consulting from about $5,000 upward, with enterprise programmes several times that. Hourly offers from smaller providers start at a few tens of dollars. These are indicative reference points, not a market average.
| Type of work | What you pay for | How it is usually priced |
|---|---|---|
| Setup and rollout | Assessment, secure deployment, pilot and enablement | Fixed-price first phase, then a monthly fee |
| Training | A workshop, a multi-day course or a cohort programme | Per course or per cohort |
| Build work | Software, AI tools and agents delivered with Claude Code | Per project after a scoping conversation |
| Programme across many teams | Rollout to many teams with governance and change management | Per programme |
The cost of Claude itself
Agency fees do not include model usage. Anthropic's documentation describes several ways to pay: Claude for Teams and Claude for Enterprise plans with per-seat billing and admin controls, pay-as-you-go API access through the Anthropic Console, or access through Amazon Bedrock, Google Cloud or Microsoft Foundry on your existing cloud contract.
Rates change, so we do not quote them here. Check Anthropic's pricing page for plans and the API page for token rates before you budget.
Automation adds its own line. When Claude Code runs in GitHub Actions, you pay for runner minutes and for the API tokens each run consumes. Anthropic's documentation suggests capping turns and setting workflow timeouts to keep that spend predictable.
What drives the price up or down
- Number of repositories and teams: each one needs its own instruction file and permission review.
- Deployment route: a cloud-provider deployment behind a corporate proxy takes more engineering than self-service seats.
- Security and compliance: regulated industries need documented controls, managed policies and audit evidence.
- Training depth: a single workshop costs far less than a champions program with months of office hours.
- Existing engineering hygiene: weak tests and slow CI limit what an agent can safely do, and fixing them is extra work.
Common questions
Is the Claude subscription included in an agency's fee?
Almost never. You buy seats or API usage directly from Anthropic or through your cloud provider, and the agency bills separately for its time. Confirm this in writing, along with who pays for usage during the pilot.
Can a small team skip the assessment phase?
Often, yes. With a handful of repositories and no special compliance needs, hands-on training plus a few days of configuration is usually enough, and BlackCard scopes small engagements like this. The assessment phase pays off when there are many teams or strict security requirements.