Autodesk
Modernize the backbone where sales, pricing, and fulfillment finally had to agree with each other.
Autodesk had outgrown its ERP — aging, heavily customized, a brake on digitization. Apollo was the ground-up replacement, and I owned the experience layer end to end for its Configure-Price-Quote and quote-to-order flows: both the customer-facing path and the dense internal platform behind it.
One owner for two experiences.
A quoting-and-fulfillment system has two sides that have to reconcile, and I picked them both up mid-handoff from an outside vendor. Part of the early job was archaeology — surfacing inherited assumptions baked so deep nobody could fully see them, before building on top of them.
Wants to be simple, guided, forgiving — a path a customer follows without a manual open in the other window.
Unavoidably dense — where export controls, pricing authority, and contract structure actually get enforced.
This is where the money moves.
A confusing pricing control or an ambiguous quote-to-order handoff isn’t a cosmetic problem — it’s a revenue and compliance issue, feeding pricing, revenue recognition, and export compliance downstream. An over-simplified screen isn’t kindness — in this system, it’s risk.
CPQ a real person could actually move through.
Configure, Price, Quote — and quote-to-order — that a salesperson or ops user could complete without a manual in the other window, while respecting the deep business rules an enterprise system can’t wish away.
Pricing authority
Who can set and change what, and where authority lives.
Availability
Controls on what can be offered, to whom, where.
Export control
Permissions and capabilities that keep the company compliant.
User roles
Who can do what as a deal moves toward a contract.
SFDC boundary
A clean handoff to Salesforce and the systems around it.
I made myself the connective tissue.
I set up and facilitated the design cadences and became the central point of contact across every group. Unglamorous — and exactly what large programs live or die on.
Expose the dense machinery — on purpose.
I designed and contributed across the whole journey — eight flows that had to read as one continuous path from landing to order:
On the finalize screen the guardrails are visible by design — pricing authority within limit, export control cleared, two-reviewer approval, and the linked Salesforce opportunity — because in this system, hiding them is the risk.
Compose the offer — with the price in view the whole time.
The core of CPQ: build an offer from attribute-based parts — products, services, programs, discounts — while a live summary keeps the number honest. Multi-year discount and Premium uplift apply automatically per the pricing rules, so a rep never has to leave to do the math.
Where a quote becomes an order — simple, guided, forgiving.
The same deal the internal CPQ was enforcing, shown to the customer as one clean transaction: billing, VAT, payment, and a running order summary. The dense machinery stays on the internal side; the customer sees a path they can finish without a manual open in the other window.
From seats you renew to tokens you spend.
Apollo had to enable Flex — a consumption model where customers buy credits and draw them down as they use products. That changes what a quote even is, and the flows had to say so.
Buy credits
Tokens purchased up front — with future start dates — instead of a fixed seat count.
Draw them down
Access tied to a live balance — it stops at zero, resumes on purchase, with no renewal.
See what’s left
Admins track credits purchased, used, and expiring — and plan the next buy.
Design impact
Quoting & access had to express consumption, not just a seat count.
Consumption as a single source of truth.
Flex turned quoting into an ongoing balance, so reporting had to make consumption legible: tokens purchased, used, and expired, plus deferred revenue and retention — the numbers finance needs to hold together as a quote stops being a one-time seat count.
Which complexity to hide, and which to make the user look at.
Let the system make the decision. Keep the path forgiving. Hide the machinery.
Force a conscious choice. Expose the control. Because getting it wrong moves money.
Facilitating alignment was facilitating this trade-off — over and over, until it held together as one product instead of two.
The highest-leverage skill was alignment, not the artifact.
Being the deliberate center of communication prevented more rework than any single screen I drew. Respect the inheritance before adding to it; and remember that ERP UX is deciding which complexity to hide and which to make the user look at on purpose.
The screens shown are clean recreations for portfolio use with sample data; program metrics (adoption, cycle-time, error-rate) aren’t quoted here.