PAID BETA PREVIEW / COMING SOON

raza-Developer

Turn a development brief into an approved plan, tested code and a delivery report. This guide previews the workflow. The paid beta is coming soon; the waitlist does not unlock access.

What this connector does

raza-Developer guides a coding-capable AI through a defined development process. You describe the job and approve the plan. The AI prepares the code, checks and delivery report in your workspace.

The result still needs review against your requirements. Timing and deployment readiness depend on the scope, access and checks completed.

A build in five steps

  1. Connect and describe the job. Complete setup, open your project workspace and give one concrete example:
    "Our software team repeats the same setup answers in support. Build an agent that drafts replies from our approved docs, cites the source and sends missing or sensitive answers to a person. It must not send replies without approval."
  2. Answer the follow-up questions. Name the documents, systems, users and actions that require approval.
  3. Read and approve the plan. Check the scope, acceptance criteria and exclusions before code is written. Correct anything that does not match your job.
  4. Review the build evidence. The workflow calls for tests before implementation, agent evaluations and a delivery report. Check failures and incomplete items rather than relying on a finished-looking demo.
  5. Open the delivery report. Follow its run instructions, review remaining setup and decide whether the result is ready for release. Ask: "What was verified, what remains, and how do I run this?"

Before beta access opens

Use an AI tool with MCP support and permission to write project files, such as a suitable Claude Code or Codex setup. A chat window without file access cannot save the project for you.

Join the paid beta waitlist for availability and pricing emails. Setup instructions will be provided when access opens. Meanwhile, you can get Raza SOR free and read the MCP setup guide.

The nine stages of the build guide

These are grouped development stages. The homepage's counted example splits one run into ten individual actions, including human decisions. The two lists use different levels of detail.

  1. UnderstandClarify the job, users and limits.
  2. PlanDescribe the architecture, data, integrations and failure paths.
  3. AgreeApprove the specification and how success will be checked.
  4. Build, tests firstWrite tests, implement the agreed behaviour and keep secrets out of source code.
  5. ScoreEvaluate representative tasks and record unsuccessful cases.
  6. PackagePrepare the agreed deployment package and configuration instructions.
  7. Automate shippingAdd repeatable testing and release steps where required.
  8. ScalePlan and test capacity against expected traffic and infrastructure.
  9. WatchProvide the agreed logs, health checks and operating guidance.

A deployment package is not proof of production readiness. Review the checks, limits and release requirements in the report.

Build capabilities

Agents

  • Single agents: support, intake, research, booking, routing, follow-up
  • Agent teams: multiple agents that hand work to each other (for example: one reads the email, one researches, one drafts the reply)
  • Controls: defined permissions, approval gates and escalation paths, checked during evaluation

Knowledge and data

  • Document retrieval: find relevant approved material and use it to prepare cited answers; test missing information and unsupported claims
  • Databases: store the workflow state and information the job requires
  • Connections to your systems: calendars, CRMs, email, existing company APIs

Interfaces

  • APIs: the "plug socket" that lets your website, app, or WhatsApp talk to the agent
  • Chat interfaces: a chat window for your website or product

Quality and handover

  • Test suites: written before implementation, with results recorded for review
  • Evals: representative cases to check behaviour and rerun after changes
  • A written delivery report: what was built, how to run it, what remains

Deployment and scale

  • Docker packaging: a consistent runtime with documented host and configuration requirements
  • CI/CD pipelines: repeatable tests and releases within agreed approval rules
  • Kubernetes deployments: container management when the deployment scope calls for it
  • Monitoring: health checks and logs for life after launch

How-to guides

How to request a new agent

One paragraph, five ingredients: who you are, the pain, the users, the must-haves, the scale:

"We are [business]. [The painful situation]. I want an agent that [does what], for [who uses it], connected to [your systems/documents]. It must never [red lines]. Expect [number] users/messages per month."

How to build for serious scale

State expected traffic so the plan can define capacity tests:

"…around 8,000 users a month now, 15,000 within 18 months. Include the deployment setup for that scale."

How to connect it to your documents

Point to the material and state the rule:

"Answer ONLY from the documents in [folder/path or system]. If the answer is not there, say so and offer a human handoff, never guess."

How to take a pilot to production

After a pilot delivery:

"The pilot is approved. Now add [the deferred items from the report: e.g. the real API endpoint, the Kubernetes deployment], as a new build on the same plan."

Best practices for a great result

  • Describe the problem, not the technology. "Our team keeps answering the same setup questions" is more useful than a tool list.
  • State the red lines. "It must never change production or email a customer without approval." Check how those rules will be enforced.
  • Give real numbers. Expected users, document count and response-time needs help define the design and tests.
  • Answer its questions specifically. Every answer becomes part of the plan.
  • Read the delivery report. It lists remaining work, required access, known limitations and checks to complete before release.

Availability and workflow questions

Can I use raza-Developer now?+

The paid beta is coming soon. Join the waitlist for availability and pricing. Joining does not provide access or take a payment.

It is asking questions instead of building. Is something wrong?+

No. That is the process. It always understands first, plans second, builds third. If you ask it to skip ahead, it will politely refuse and finish the plan first.

Do I need to know how to code?+

You can describe the job in plain words. Reviewing code, configuring access and deploying the result may need technical help. The delivery report should make those needs clear.

How is this different from asking ChatGPT for code?+

The connector adds an explicit workflow: an approved plan, tests, evaluations, packaging and a delivery report. Review that evidence before putting the system into use.

Where is my project?+

Check the workspace you opened in your coding tool. Ask for the saved project path and delivery report. Confirm file access if no project was created.

Can it work with my existing systems?+

Name the systems in your brief. Available APIs, permissions and credentials determine what can be connected; missing access should be listed in the plan or report.

What if my problem is too big for one sitting?+

Break the scope into reviewable stages. Agree a limited pilot first and record deferred work in the report. Do not assume the whole project will finish in one session.

The first reply took a long time.+

Check whether the tool shows a connection error or a running request. If it fails, confirm the connector setup and access. Include the error message when asking for help.

Glossary

TermPlain meaning
AI agentSoftware that interprets inputs and uses tools to work within defined limits
SpecsWritten requirements and acceptance criteria for the build
TestsAutomatic checks, written before the code, that must pass before delivery
EvalsThe agent's report card: example questions with expected behavior, scored
RAGRetrieving relevant material to inform a generated answer; output still needs checking
APIThe plug socket other software uses to talk to the agent
DockerA container format for a consistent runtime on compatible hosts
CI/CDThe automatic conveyor belt that re-tests and ships every update
KubernetesSoftware that schedules and manages containers across computing resources
Delivery reportThe plain-language note in your folder: what was built, how to run it, what remains

The paid raza-Developer beta is coming soon. Join the waitlist, or explore the free Raza SOR record.

CONTACT / WORKFLOW AUDIT

Want this pointed at your workflow?

Bring the job that should finish without a person typing every instruction. We map the run, count where the humans sit, and decide whether a focused pilot makes sense.

Book a 15-min call →