# How to run a Salesforce implementation from requirements to go-live.

> Running a Salesforce implementation with SyncOnAI 360 means keeping requirements and decisions as pages beside the org, planning sprints and releases on a project board, building with the AI, agents and visual builders inside the org's conventions, and shipping through validated, approved proposals with receipts and rollback up to go-live.

Source: https://synconai360.com/use-cases/implementation

## Key facts

- **4**: Visual builders: Flow, page, component, Apex
- **5**: Specialist agents
- **Every**: Change validated before approval
- **1**: Workspace from requirements to go-live

## Why Salesforce implementations overrun

Requirements drift from what is built, builds drift from the design, and the go-live weekend is when everyone finds out.

- **Requirements lost.** Decisions from workshops never reach the build.
- **Inconsistent builds.** Each builder follows their own patterns.
- **Deploy-day surprises.** Coverage, dependencies and data issues appear at go-live.
- **No undo.** A bad go-live change has no prepared way back.

## How do you keep requirements connected to the build?

Record workshops as transcripts, keep requirements and decisions as pages in the project wiki, and ask the chat to draft from them. Pages are cited when the AI answers, so the build stays tied to what was agreed.

- Workshops transcribed
- Requirements in the wiki
- Cited by the AI

## How do you plan the implementation?

A project has a board, backlog, sprints, roadmap and releases, with reports on one engine. Work items link to the deploys that delivered them, so progress reflects what actually reached the org.

- Sprints and releases
- Work linked to deploys
- Honest reports

## How does the team build consistently?

Org rules set the conventions, and the AI and specialist agents build inside them. Visual builders for Flows, Lightning pages, components and Apex explain before they edit, and the Canvas lens shows the data model as it grows.

- Conventions enforced
- Builders that explain
- Data model on a canvas

## How do you avoid go-live surprises?

Every proposal runs named checks, including dependencies, deploy order and test coverage, and is validated against the target org. Deploy to sandboxes as you go and compare them with production before go-live.

- Coverage and order checked
- Validated against the target
- Environments compared

## How do you run go-live safely?

Approve production proposals with a second admin, inside the policy's windows. Every deploy leaves a receipt with rollback to the recorded previous version, and results can post to Slack as they land.

- Two-person approval
- Receipts and rollback
- Results in Slack

## How do you hand over after go-live?

Requirements, decisions and runbooks stay as pages beside the org, every change has a receipt, and the org stays synced, scored and explained. The support team inherits a documented, measured org rather than a folder.

- Pages beside the org
- Receipts for every change
- A measured org

## How to run an implementation

1. **Capture.** Requirements and decisions as pages.
2. **Plan.** Sprints, releases and the roadmap.
3. **Build.** AI, agents and builders inside conventions.
4. **Prove.** Checks, validation and sandbox deploys.
5. **Go live.** Approved deploys with receipts.

## An implementation, drifting versus connected

| Without | With SyncOnAI 360 |
|---|---|
| Requirements lost after workshops | Requirements cited during the build |
| Each builder's own patterns | Conventions enforced |
| Surprises at go-live | Checks and validation throughout |
| No way back | Rollback from every receipt |

## Implementation checklist

- [ ] Baseline the org with the audit before the build starts.
- [ ] Keep requirements and decisions as pages in the project wiki.
- [ ] Set naming conventions and the preferred automation type as org rules.
- [ ] Plan sprints and releases on the project board.
- [ ] Deploy to sandboxes continuously and compare with production.
- [ ] Read coverage and dependency checks on every proposal.
- [ ] Approve go-live changes with a second admin and keep the receipts.

## Who runs implementations

- **Consultancies.** Deliver implementations with one method.
- **Delivery managers.** Track progress by what reached the org.
- **System integrators.** Run many implementations the same way.

## Frequently asked questions

### Can requirements live in SyncOnAI 360?

Yes, as pages in the project wiki, including transcribed workshops.

### Does it replace our project tool?

It covers boards, sprints, releases and reports; teams that keep Jira can raise findings there.

### Can the AI build from requirements?

It drafts against the org and can cite your pages; every draft is a proposal you review.

### Can we see the data model?

Yes, on the Canvas lens.

### How do we avoid coverage failures at go-live?

Every proposal with Apex runs a coverage check before deploy day.

### Can go-live changes be rolled back?

Yes, from each deploy's receipt; data changes cannot be reversed.

### Can we post go-live results to Slack?

Yes, every deploy result can post to a channel.

### Is this on every plan?

Yes. Every plan includes every feature.

### Can we baseline the org before the build?

Yes. Run the audit and keep the review; later reviews compare against it.

### Can several builders work at once?

Yes. Each builds and proposes; production changes are approved by an admin other than the author.

### Can the client see progress during the build?

Share health and audit reports as read-only links, and report delivery from the project's reports.

### Can AI speed up the build?

Yes. The AI and five specialist agents draft against the org, and every draft is a proposal you review.

### Can we compare the build sandbox with production before go-live?

Yes. Metadata can be compared between any two connected orgs.

### Where do user stories live?

As work items on the project board, with requirements and decisions in the project wiki beside them.