How to Run an AGV RFP Without Getting Locked Into One Vendor

An AGV RFP fails when vendors are invited to invent the problem. Each proposal then uses a different fleet size, a different infrastructure list, and a different definition of success. Price stops meaning anything. The first OEM that walked the floor often wins, because their concept is already circulating internally.

A usable RFP starts after independent feasibility, not instead of it. Vendors should bid a defined process, defined constraints, and defined performance — not a blank plant tour. Scoring the returns is a later step, covered in how to evaluate competing AGV proposals. Choosing the firm after that is covered in how to choose the right AGV vendor.

Freeze the Problem Before You Invite Bids

If the moves, the constraints, the peak performance, and the owner-versus-vendor split are not frozen, you are not running an RFP. You are collecting brochures. Vendors will fill the gaps with their own fleet size, their own infrastructure list, and their own definition of done. Independent feasibility is what freezes that package before procurement starts. See what an AGV consulting engagement includes.

Write Requirements, Not a Preferred Vehicle

Name the job. Do not name the SKU. The RFP should say what must be moved, in what environment, at what rate, with what recovery behavior. It should not say “supply twelve of Brand X’s laser forklifts unless you have something similar.”

Vendors can still propose different navigation methods or vehicle types. They must show how their proposal meets the same performance and constraint list. If one bidder needs extra floor work or a larger fleet, that has to appear in their price, not in a later change order. The study does not size the fleet. The bidders do, against the peak rate you stated. Process still comes before vehicle selection. See why process design should come before vehicle selection.

Force Comparable Line Items

Ask every bidder for	Why
Fleet size at stated peak, with utilization assumptions	Stops “fewer vehicles” from hiding lower throughput
Infrastructure they assume the owner will fund	Floors, Wi-Fi, charging power, and doors show up here
Software licenses, annual support, and spare parts	Capital-only quotes understate TCO
Acceptance tests at peak operating conditions	Prevents a quiet-hour site acceptance test (SAT) from closing the job
Exceptions they will not handle	Manual interventions belong in the operating cost

If a vendor refuses to bid in that format, treat that as information. They want the comparison run on their slide template. That is how you get locked in before the PO.

Do Not Let the First Vendor Write the Spec

Many bid packages are written around the first site survey. The path, charger locations, and even the vehicle class are copied from that OEM’s sketch. Competing bidders then look non-compliant for proposing a different, sometimes better, fit.

The fix is sequence. Independent feasibility first. Vendor-neutral requirements second. Then the RFP. Manufacturer visits can inform the study. They should not author the bid package. That is the same problem as going straight to a manufacturer for the design. See warehouse automation consulting vs. going straight to an AGV manufacturer.

What to Put in the Packet

A complete packet is boring on purpose. It should include:

• Process description and volume tables, peak and average
• Pickup and drop points, load types, and required interface heights
• Layout with constraints marked — floors, columns, hazardous area classification, traffic, docks — not a blank CAD file
• Load, interface, and recovery requirements
• Owner-versus-vendor split for floor, power, network, and WMS
• The pricing matrix every bidder must fill in
• Acceptance tests stated as operating conditions, not “successful demo”

If a section is missing, bidders will invent it. That is how one RFP turns into four unrelated proposals. “Non-compliant” should mean a bidder refused the same problem, not that they refused the first OEM’s sketch.

Answer bidder questions in writing, once, to every firm on the list. A private clarification recreates the first-vendor problem. If a question reveals a hole in the packet, patch that section and reissue it rather than letting one integrator bid a different scope.

A Practical Sequence

1. Finish independent feasibility and freeze the process in scope
2. Write vendor-neutral requirements and a pricing matrix
3. Issue the RFP to a short list that can actually execute in your environment
4. Hold one round of technical questions, answered in writing to all bidders
5. Score the returns on the same problem — the method is in how to evaluate competing AGV proposals
6. Keep an owner’s engineer through clarification and award

The short list should include firms that have worked in similar plants, not only firms that presented well. Chemical, mixed-traffic, and brownfield sites are not interchangeable with a dry warehouse reference. Review after award is the owner’s engineer role.

Key Takeaways

• An AGV RFP only works if every vendor bids the same problem
• Write requirements and constraints before you invite proposals
• Force comparable line items for fleet, infrastructure, software, and acceptance
• Do not let the first OEM survey become the specification
• Independent feasibility should precede the RFP, not clean up a messy bid after the fact

Frequently Asked Questions

How many vendors should we invite?

Three to five that can actually execute in your environment is enough. A dozen brochure responses will not make a weak spec comparable.

Should we share our feasibility study with bidders?

Share the requirements, volumes, constraints, and performance targets. You do not need to share every internal recommendation. You do need every bidder working from the same facts.

Can we run an AGV RFP without a consultant?

Yes, if the internal team can write a vendor-neutral spec and hold every bidder to it. Many plants cannot, because the first vendor already shaped the problem. That is when independent help belongs before the RFP, not after the award.

What if one vendor is clearly cheaper?

Check what they excluded. A low vehicle price with owner-funded floor, network, and extra labor during exceptions is not a low bid.

Review our feasibility study scope for the work that should sit in front of an RFP.