Sales handovers
Bring together opportunity context, commitments and next steps for a colleague to review.
AI, Agentforce & conversational workflows
AI and agents for the work your team does every day.
We help connect AI to Salesforce, Slack and your business processes, from assessing an idea to testing and implementing it.
Start with a useful question
Which task takes more time than it should?
Preparing a handover, finding project context or drafting an update are useful places to start.
The best starting point is a task your team already understands, with a clear way to check the result.
Built around your processes, data and people.
Practical possibilities
Four tasks worth exploring.
These are examples to explore, shaped around your systems and readiness.
Service response drafts
Use approved knowledge and case context to prepare a useful starting point for your service team.
Project updates
Turn permitted project information into a draft of progress, decisions and next actions.
Actions from a conversation
Start a guided workflow through chat, with permissions and review before business records change.
See the possibilities
Ask in Slack. Review the output.
An assistant gathers permitted information and prepares a draft. This sample conversation shows a priorities list, project update and statement of work returned for review.
01 · Priorities
A priorities list
Alex: “@Priorities show today’s priorities”
The agent prepares a short list: review the sales handover, confirm the service owner and prepare for playback.
02 · Project update
A project update
Alex: “@Project prepare the project update”
The agent drafts progress, open decisions and next steps from approved project notes. No update is sent.
03 · Scope
A scope draft
Alex: “@Scope draft a statement of work”
The agent prepares scope, deliverables and open questions. Assumptions and commercials still need review.
04 · Human review
Human review
Alex: “Keep these as drafts. I’ll review them.”
All three outputs wait for human review. Nothing is shared and no business records are changed.
Sample conversation using invented content. Available actions depend on the connections and permissions we implement.
How do the connections work? ↓How the pieces fit
A conversation connected to your business.
A request can bring together the right information, tools and specialist workflows. Salesforce remains the business platform beneath it, with access controls and human review designed into the process.
A request from your team
For example: prepare a sales handover in Slack
Workflow coordination
Understand the task · apply permissions · choose the next step
Permitted tools & business data
Records, approved knowledge and workflows
The appropriate AI capability
Agentforce or an optional model such as ChatGPT or Claude
A result ready for review
Check the draft and approve any permitted action
Salesforce
Your core business records, processes and permissions
MCP setup & connected AI
Connect AI to the information it needs.
An assistant is more useful when it can work with the information your team actually uses. Model Context Protocol (MCP) is an open standard for connecting compatible AI applications to tools and data sources. It can be part of the connection between a conversation and your business systems.
Less copying between systems
Let an assistant retrieve permitted information from connected sources, instead of asking people to paste the same context into each task.
Better-informed drafts
Bring relevant records and approved knowledge into summaries, handovers and updates, with sources your team can check.
Fewer manual handovers
Connect a request to a defined workflow, from gathering information to preparing an output or proposing a permitted action.
A foundation you can extend
Where compatible tools are available, reuse a standard connection approach as you introduce more useful AI workflows.
What setup involves
What we configure and test.
We help identify the right use case, assess suitable connectors and configure the access needed for the agreed workflow.
- Choose the task and systems. Define the outcome, information needed and connector compatibility.
- Set the access boundaries. Configure authentication, limit permissions and separate read-only access from actions that change records.
- Design review and exception handling. Agree what needs human approval, how failures are handled and what activity is recorded.
- Test, document and hand over. Validate behaviour with representative scenarios, including denied access, and agree ownership for ongoing support.
MCP is a connection standard, not a security guarantee. Access controls, approval steps and monitoring need to be implemented and tested. Salesforce, Slack and other systems may use MCP, native connectors or APIs depending on the use case.
A sound foundation
Before an assistant gets access.
Use-case discovery
Start with a practical task across sales, service or operations and define what a useful outcome looks like.
Context & data
Review records, knowledge, data quality, permissions and the information an assistant would need.
Human review & controls
Agree what can be suggested, what needs approval and how actions and exceptions will be recorded.
Choosing the right tool
Choose Agentforce, a custom assistant or conventional automation according to the work and the constraints.
From idea to useful workflow
From evaluation to implementation.
01
Evaluate
Agree the task, value, constraints and readiness.
02
Prototype
Make the proposed workflow tangible using suitable sample information.
03
Validate
Check quality, permissions, review points and exception handling with your team.
04
Implement
Introduce the agreed solution, document it and support adoption.
If conventional automation is the better answer, we will recommend it. Your business does not need a complex agent system to make meaningful progress.
What would you like your systems to do better?
Bring us a task your team would like help with. We will explore the options and the controls it needs.
Talk to Forgemoor