Building 3PL RFPs and Custom GPTs for Your Operation
This final lesson brings the course full circle. You've learned to use AI for daily writing, document review, exception handling, and KPI summaries. The next step is two-fold: writing a 3PL RFP that gets you good responses, and building reusable Custom GPTs that let your whole team use AI consistently. Both are advanced moves that put you ahead of 95% of logistics managers.
What You'll Learn
- How to write a 3PL RFP that produces comparable, useful responses
- How to evaluate 3PL responses with AI
- How to build Custom GPTs (or Claude Projects) for your team
- The 3 most valuable Custom GPTs every logistics operation should build
Why Most 3PL RFPs Get Bad Responses
3PL RFPs typically fail because the buyer doesn't:
- Define the work concretely (volumes, SKUs, profile, peaks)
- Specify the response format (so quotes can be compared)
- Ask the right questions (vague questions get vague answers)
- Pre-screen the bidder list (avoiding wasted cycles)
AI helps with all four.
Drafting a 3PL RFP With AI
Suppose you're moving DTC fulfillment from in-house to a 3PL.
"Help me write a 3PL RFP for our DTC fulfillment business. Profile: 80,000 parcels/year, ~1.2kg average weight, 3-day Q4 peak at 4x normal volume, 1,400 SKUs (60% apparel, 40% accessories), seasonal returns spike in January, target US-wide 3-day ground delivery. Required services: receiving from 3 import suppliers (FCL containers), put-away and storage, pick-pack-ship for direct-to-consumer orders, returns processing, B2B retailer fulfillment (small portion). Required performance: 99.5% pick accuracy, 24-hour same-day cutoff for ship-out, monthly KPI scorecard. Required pricing format: per-parcel pick & pack, per-pallet receiving, per-pallet-month storage, returns per unit, retailer fulfillment per case. Generate the RFP as: (1) cover letter with timeline (RFP issued, questions due, responses due, finalist presentations, decision), (2) operational requirements section (the scope above, expanded), (3) performance requirements section, (4) pricing format spec (so all bidders quote apples-to-apples), (5) scorecard structure (we'll evaluate on cost, capability, references, location, technology, cultural fit), (6) appendix with our top 20 SKU profile and a representative week of order volume. Plain language. Make it bidder-friendly so we get good responses."
You'll get a usable first draft of a 25-page RFP in 3 minutes. Spend the next hour adjusting for your specifics.
Evaluating 3PL Responses
Once responses come in, AI compresses what would be a week of analysis into an afternoon.
"Below are 4 3PL responses to our DTC RFP. Each has a different pricing model — per-parcel, cost-plus, hybrid, activity-based. Normalize into a comparable annual cost model assuming 80,000 parcels/year, 1.2kg average, with our peak profile. For each bidder: (1) total estimated annual cost, (2) hidden costs (storage minimums, long-tail SKU fees, retrieval charges, exception fees), (3) peak surge pricing exposure, (4) one-time setup vs. ongoing run-rate, (5) flexibility score (can they scale up/down). Then rank by: most cost-effective at our volume, most flexible if our volume changes, best capability fit for our SKU mix. Recommend a finalist short list of 2 with the tradeoff. Responses: \[paste\]."
This single prompt turns 4 confusing rate cards into a defensible recommendation.
Reference Check Conversations
Before signing, you call references. AI helps you prep questions that surface real signal.
"Help me prep for a reference call with the operations lead at one of our finalist 3PL's existing customers. The 3PL is Pinnacle Fulfillment, the reference customer is a similar-size DTC apparel brand. Generate 12 questions in 4 categories: (1) onboarding (was the 90-day ramp on time, what hurt), (2) day-to-day operations (accuracy, response time, exception handling), (3) financial (any surprise charges, billing accuracy, dispute experience), (4) relationship (account manager quality, escalation path, would they re-sign). For each question, also tell me what answer would be a green flag and what would be a yellow/red flag."
Building Custom GPTs (or Claude Projects)
OpenAI's Custom GPTs let you package context, instructions, and reference documents so anyone on your team can use the same well-tuned assistant. Claude has a similar feature called Projects.
The 3 most valuable Custom GPTs for a logistics operation:
1. The "Carrier Communications" GPT
Pre-loaded with: your contracted carriers, your standard SLAs, your tone of voice, examples of past well-written escalations.
Instructions you set:
"You are a logistics communication assistant for \[Company\]. When asked to draft any carrier-facing email, follow these rules: (1) lead with facts, not emotion, (2) include specific dates, PO numbers, and dollar amounts when available, (3) reference the contracted SLA where applicable (see attached contract summary), (4) end with a specific request and timeline, (5) keep under 200 words unless detail is required, (6) sign per the user's instruction. Default tone: firm, professional, collaborative. Never apologize on behalf of the company without user instruction. Always ask for missing facts rather than inventing them."
Anyone on the team types "draft email to Reliable Freight about missed pickup yesterday" and gets a consistent, on-brand draft.
2. The "Exception Triage" GPT
Pre-loaded with: your exception playbook, escalation contacts, common root causes.
Instructions you set:
"You help logistics team members triage exceptions. When given raw information about an exception, organize into: (1) one-line summary, (2) impact severity (low/medium/high/critical), (3) people who must be notified in the next 30/60/120 minutes, (4) facts to collect immediately, (5) what NOT to do or commit to yet, (6) recommended next 3 actions in order. Always lean toward asking for missing facts rather than guessing. Reference the attached exception playbook for standard responses."
3. The "Weekly KPI Memo" GPT
Pre-loaded with: your KPI definitions, last 6 weeks of historical data summary, your VP's preferred memo format.
Instructions you set:
"You produce weekly logistics KPI memos for our VP of Operations in a consistent format. When given the week's data, output: (1) 2-line headline, (2) priority KPI table with current/target/trend, (3) one paragraph on the most important miss or win, (4) 3 actions with owners and dates, (5) one ask of leadership. Maximum 350 words. Match the tone of the example memos attached."
Deploying Custom GPTs to Your Team
The rollout matters as much as the build. A working pattern:
- Build it yourself, use it for 2 weeks, refine the instructions and context
- Pilot with 2 team members for 2 weeks; collect feedback
- Document what it does well and what it doesn't
- Train the wider team in a 30-minute session: when to use it, when not to, how to give it good context
- Review and update monthly as your operation evolves
Custom GPTs are not "set and forget." They are living tools that improve as you tune them.
A Final Thought
The logistics managers who get the most out of AI in 2026 aren't the ones using the most prompts. They're the ones who built 3 great Custom GPTs, integrated them into their team's daily workflow, and freed up 8 hours a week to do the higher-value work AI can't: walk the dock, ride the route, build the carrier relationship, develop the team.
You now have what you need to be one of those managers.
Key Takeaways
- A great 3PL RFP defines work concretely, specifies the response format, asks specific questions, and pre-screens the bidder list — AI helps with all four
- Use AI to normalize 3PL responses with different pricing models into a comparable annual cost view
- Build Custom GPTs (or Claude Projects) for the workflows your team does repeatedly: carrier communications, exception triage, weekly KPI memos
- Roll out Custom GPTs slowly: build, pilot, document, train, review monthly
- The biggest payoff of AI in logistics isn't the time saved on writing — it's the time freed up for the higher-value work AI can't do