CARISAM Practical AI Workflows

Purpose

AI should reduce repetitive work without weakening Lance's judgment.

The goal is not to make CARISAM an AI company. The goal is to help Lance and Jim move faster with fewer missed details.

Core operating rule

AI may summarize, format, compare, extract, draft, and check.

Lance owns understanding, judgment, vendor trust, customer commitments, and final approval.

Workflow 1: Phone and meeting capture

Trigger

A customer, vendor, architect, contractor, or shipping partner discussion includes project details, pricing context, specifications, timeline, or commitment language.

Process

1. Record the conversation with an approved note-taking tool when appropriate.
2. Save the transcript to the project folder or quote record.
3. Generate a short summary.
4. Extract action items.
5. Extract commitments.
6. Extract missing information.
7. Add follow-up items to the quote tracker.

Output

1. Project summary.
2. Action list.
3. Open questions.
4. Risk notes.
5. Follow-up draft if needed.

Workflow 2: Email chronology reconstruction

Trigger

Lance needs to know what happened across a long email chain or old project.

Process

1. Collect the relevant emails or thread export.
2. Ask the AI to produce a chronology.
3. Identify who said what, when, and what remains unresolved.
4. Extract pricing, product, lead time, shipping, and payment details.
5. Flag contradictions or missing items.

Output

1. Project timeline.
2. Current status.
3. Last known commitment.
4. Open loops.
5. Suggested next message.

Workflow 3: Supplier RFQ draft

Trigger

A new customer request needs supplier pricing.

Process

1. Enter the customer request into the quote tracker.
2. Confirm product description, quantity, unit, destination, and timing.
3. Generate a vendor-specific RFQ.
4. Include required response fields.
5. Add a deadline.
6. Route to Lance for approval before sending.

Output

1. Vendor RFQ draft.
2. Structured response link in the next phase.
3. Follow-up date.

Workflow 4: Vendor response comparison

Trigger

Two or more vendors respond to an RFQ.

Process

1. Extract price, lead time, availability, freight assumptions, expiration, and notes.
2. Normalize the responses into the comparison table.
3. Identify lowest cost, fastest lead time, and most trusted option.
4. Flag missing data.
5. Prepare a decision brief.

Output

1. Vendor comparison table.
2. Decision brief.
3. Risk notes.
4. Draft customer quote data.

Workflow 5: Quote review checklist

Trigger

A customer quote is ready to send.

Process

Check:

1. Customer name.
2. Product descriptions.
3. Quantities.
4. Units.
5. Costs.
6. Sell price.
7. Margin.
8. Lead time.
9. Freight assumptions.
10. Payment terms.
11. Quote expiration.
12. Special notes.
13. Prior customer expectations.

Output

1. Pass or review needed.
2. Missing fields.
3. Possible margin or wording issue.
4. Final review note for Lance.

Workflow 6: Product and vendor research

Trigger

Lance needs a backup supplier, unfamiliar product path, or alternative source.

Process

1. Start with Lance's known vendor context.
2. Search trade associations, industry directories, expo lists, niche publications, and known regional specialists.
3. Avoid relying only on search ranking.
4. Prepare a candidate supplier list.
5. Include reasons to trust or question each option.

Output

1. Candidate vendor list.
2. Contact details.
3. Product fit.
4. Risk notes.
5. Suggested first outreach.

Guardrails

1. Do not send AI-generated customer or vendor messages without review.
2. Do not let AI make final vendor decisions.
3. Do not treat extracted PDF data as final until checked.
4. Do not expose sensitive customer or vendor information to tools that are not approved.
5. Do not replace Lance's relationship judgment with a score.