AI-assisted development needs requirements, standards and usage control.
In your team, every bit of rework hurts. When AI goes straight from prompt to code, intent gets lost in the chat history and problems show up late, in review or in production.
Implicit requirements
Requirements left out of your request can lead to errors. Your team may only identify edge cases after the code has been written.
Undocumented standards
Your architecture and your team's conventions are passed on case by case, in every pull request review.
Deferred maintenance
Your tests, documentation and test data become outdated between development cycles because they depend on someone remembering to review them.
Unpredictable cost
Your subscriptions are scattered, with no view of who uses what, how much is consumed or the monthly spend cap.
Specification first, code second.
Your team reviews and approves requirements, design and tasks before implementation. With Kiro, we use native specs; with Claude Code, version-controlled project documents, as a practice in Aquarela's method.
Kiro or Claude Code in your daily work.
Specs and project documents
Kiro offers native specs with requirements, design and sequenced tasks. With Claude Code, your team documents requirements, design and tasks in files versioned alongside the code, following Aquarela's method.
Steering files and CLAUDE.md
Steering files give Kiro persistent knowledge of your project. Claude Code reads CLAUDE.md at the start of every session, with your coding standards, architecture decisions and review checklists.
Agent Hooks and hooks
In Kiro, Agent Hooks run commands or prompts on session events. In Claude Code, hooks run commands defined by your team at specific points in the execution lifecycle to enforce rules and automate tasks.
Model Context Protocol (MCP)
Kiro and Claude Code use this open standard to connect to your team's tools and data sources. Kiro also offers Powers for integrations.
Kiro checkpoints
With checkpoints, you can undo the Kiro agent's changes and return to an earlier state.
Kiro CLI and Claude Code in the terminal
Use Kiro CLI or Claude Code in the terminal to work with your code and run commands in your team's development workflow.
We implement the tools and train your team to develop with AI.
Aquarela leads the work with your team: defines the licensing setup, configures the tools in your workflow, teaches best practices and tracks usage and costs.
Kiro by credits; Claude Code by seats and usage.
Aquarela sizes and supports purchasing based on usage measured during your pilot. Kiro uses monthly credits per developer; Claude Code uses seats on Anthropic's Team or Enterprise plans, with their own usage and billing rules.
- Kiro: team plans include consolidated billing, usage analytics and SSO through AWS IAM Identity Center.
- Kiro: plan credits do not roll over between months; add-on credits can be purchased on paid plans and expire after 12 months.
- Kiro: on Pro, Pro+, Pro Max and Power plans, your content is not used to train models and generated output is covered by intellectual property indemnity, under AWS terms.
- Anthropic Team: for teams of 2 to 150 people, with Standard and Premium seats that can be combined, monthly or annual billing and usage limits. Includes SSO, usage analytics and user and organization spend controls; your content is not used to train models by default.
- Anthropic Enterprise: a minimum of 20 seats, annual seat billing and usage billed at application programming interface (API) rates. Administrators set user and organization spend limits; the plan includes SSO, domain capture, provisioning through System for Cross-domain Identity Management (SCIM), audit logs and usage analytics.
Aquarela guides your team from licensing setup to rollout.
We start with your team's roles, implement Kiro or Claude Code in one repository and lead training with a real feature. You decide on expansion using usage and cost data and consumption policies defined with Aquarela.
Tool and licensing
- We work with you to define Kiro or Claude Code for each role in your team
- We size licenses or seats and help you choose the purchasing model
Tool in your repository
- We configure identity, SSO and your standards in steering files or CLAUDE.md
- We configure hooks for tests and documentation and integrate your pull request workflow
Pilot and training
- We train your team in specifications before code and in reviewing requirements, design and tasks
- We guide a real feature from request to pull request and measure usage and cost
Rollout with governance
- We work with you to adjust licenses or seats for rollout, based on pilot data
- We define consumption policies and spend limits and track your usage and costs
Bring AI-assisted development to your team.
Talk to our team and start with a pilot on a real feature from your backlog. Work with Aquarela to determine whether Kiro or Claude Code fits your workflow.
