Founder of Weft. Helping SaaS teams improve customer-request handoffs across support, product and engineering. Assessments & practical workflow blueprints.
Joined October 2026
- Tweets75
- Following1
- Followers1
- Likes0
Pinned Tweet
I’m building Weft to help SaaS teams improve customer-request handoffs across support, product and engineering.
One workflow: assessment, owners, process maps and a practical rollout plan.
Explore the illustrative blueprint: withweft.app/blueprint.html
Chris describes reserving Fridays for experiments and small requests. At Weft, we'd add an end-of-day decision: ship, stop, or seek capacity for more work. An unfinished experiment shouldn't silently become Monday's commitment. #SaaS #ProductManagement #ProductOps
My co-founder, @marcbaumbach, has a habit of shipping customer requests the same day. Sometimes, even while they're still chatting with support.
It's a pretty magical experience for customers.
It often happens on a Friday, so it became a running joke. We’d ask, “Is this going to be a Free-for-All Friday?” whenever something popped up.
Well, we recently made Free-for-All Friday (FFAF) a real thing for the whole company!
Monday through Thursday, we focus on major initiatives. Fridays are reserved for things that are harder to prioritize, experimenting with new ideas, knocking out customer requests, and unplanned work added to our plate by partners.
As we say, shavings make a pile.
We're only a few weeks in, but it's been awesome to see the team embrace it. My favorite random project was a Mac menu bar app that displays some company metrics, because every founder loves staring at their revenue. :)
FFAF gives us room to build even more momentum without distracting our big initiatives... What's been on your mind for too long? Today's a great day to check it off. 🤘
Karri describes people rotating between projects. Open customer commitments need a successor when a project ends. At Weft, we'd record those commitments and the accepting owner in the handover blueprint. #SaaS #ProductOps #CustomerSuccess
No huge changes. Engineers primary concern is technical, designer primary concern is experience, PM primary concern is clarity and business outcome. Anyone can write code if they want to prototype or fix something.
We been always operating in a way that design+eng lead projects, and own the product work, quality and output of the projects. The team has access to customers and customer feedback we have in the system. We run internal betas and betas with customers. The team learns from the feedback.
PMs operate higher level and support when needed. There is less of that classical PM/design/engineering division. We also haven't had product area teams. People work in Product or Infra or Mobile. People in the product team would just be rotated from project to project.
We are changing some of that now because there is more distinct areas we're creating in product and that need more focused way of working on it.
A fast support acknowledgment can hide a slow product decision. In Weft’s fictional blueprint, decision delay runs from receipt to the first recorded product decision. Track that separately from response time. Illustrative, not results. #SaaS #ProductOps #CustomerSupport
ALT Measurement table from Weft’s fictional sample blueprint. Decision delay is time from receipt to the first recorded product decision; the proposed pilot check measures a baseline and compares similar requests. Other rows cover evidence completeness, follow-up coverage, manual coordination time and outcome evidence. Illustrative plan, not measured client results.
Before adding a feedback register, check whether your existing tools can hold it. In Weft’s fictional blueprint, Notion is a pilot choice; Jira may absorb the register later. Test fields, access and update ownership first. #SaaS #ProductOps #WorkflowAutomation
ALT Table from Weft’s fictional sample blueprint, with columns System, Proposed role and Verify before implementation. Notion holds the initial pilot’s evidence and decision register; its checks include owner, field schema, link visibility, update discipline, and whether Jira can absorb the register later. Other rows cover Intercom/Slack, Jira, GitHub, internal release panel, PostHog and optional automation. Illustrative plan, not client results.
Correction: the image shows the fictional sample introduction, not the technology table described in its alt text. The Notion/Jira register guidance is under Technology here: withweft.app/blueprint#techn…. The sample is illustrative, not client results.
Weft is now on Flowelva! We help PE teams plan clearer reporting workflows, follow-ups and handoffs.🎉
Explore: flowelva.com/providers/weft
#privateequity #fundadministration #automationflow
Useful routing example. At Weft, I’d also check each queue’s capacity: arrivals, time per exception and oldest waiting item. Agree when overload triggers reassignment before promising a response time. #CustomerSupport #WorkflowAutomation #SaaS
Automate the routine work, send everything unusual to one expert, and you've quietly hired a universal exception handler.
Nobody put that role on the org chart. It's part of your AI operating model anyway.
Illustrative example: an order workflow handles the routine cases and sends the rest to Priya (fictional). Pricing questions, damaged shipments, missing customer records. One inbox, one person.
Better design separates the exceptions:
- Pricing calls go to whoever owns pricing
- Data defects go to a named data owner
- Customer remedies go to the service team
Priya still helps with the strange cases. She stops being the permanent destination.
Track repeat exceptions so someone removes the cause. Give each route a response expectation and an escalation path.
Before you add another assistant, map the work that arrives after it gives up.
Where does that work land in your org today?
A support handoff can break after a field or permission changes. I run Weft: our fictional blueprint calls for error reviews after those changes, not only at launch. We’d define the owner and recheck in the plan. #CustomerSupport #WorkflowAutomation #SaaS
ALT Screenshot of Weft’s fictional sample blueprint, showing Fields to record, Integration safeguards, and Cost and effort assumptions. The safeguards assign a maintainer, alert them on failed transfers, and call for recurring error reviews weekly and after permission or field changes. This is an illustrative planning example, not implemented client results.
When a promised fix shrinks in scope, support needs the new boundary. I run Weft: our fictional blueprint routes scope changes back to product. We’d define who updates the customer too. Implementation is separate. #CustomerSupport #ProductManagement #SaaS
ALT Excerpt from Weft’s fictional sample blueprint, step 04: Implement with context. Product approves a scoped need; engineering uses the specification, source links, acceptance criteria and linked issues. The action includes escalating scope changes to product. Original evidence must remain accessible and required tests recorded. Illustrative planning example, not client results.
Correction: the attached crop shows the sample introduction, not step 04 as the alt text says. The scope-change instruction is under “04 — Implement with context” here: withweft.app/blueprint#instr…. The sample is fictional.
Haipi reports AI-to-human handoffs happening too late. I run Weft: we’d map triggers and test redacted cases against staffed capacity, then document the fallback in a blueprint. Implementation separately scoped.
#CustomerSupport #CustomerExperience #SaaS
Pony | Weft retweeted
Consider a ticket marked “closed” while the customer’s problem remains unresolved.
Support has handed it to product. Product has handed it to engineering. Each handoff is complete.
The customer is still asking what happens next.
At Weft, I’d add decision limits beside each exception owner: what can they fix, what needs approval, and when must they escalate? A named owner who can only forward the ticket still leaves the customer waiting.
#SaaS #WorkflowAutomation #CustomerSuccess
A customer used the feature. Did it solve their original problem?
Weft’s fictional blueprint separates adoption from resolution: solved, partial, unresolved or unknown. Missing evidence stays unknown—not a success claim.
#SaaS #CustomerSuccess #ProductManagement
At Weft, I’d also check who inherits the work: faster AI triage can leave support reviewing more false escalations. Count review and correction effort across the receiving teams, not just time saved at intake. #SaaS #AIAdoption #EnterpriseAI
“We deployed AI” is not an outcome. Measure whether the workflow became faster, more accurate and easier to use—and track the exceptions and human effort it still requires. #AIAdoption #EnterpriseAI
Repo-derived docs still need the context outside the code: negotiated exceptions, promised dates and escalation agreements. At Weft, I’d keep these beside the technical rules, with an owner to resolve conflicts. #SaaS #CustomerSupport #Documentation
Docs go stale, so people ask the engineer who built the feature. That's why we built Live Docs in @DevplanAI:
Step 1: Connect your repo from @github, @gitlab or @bitbucket.
Step 2: Devplan writes a wiki page and diagram for every feature.
Step 3: Open one to see what it does and the rules behind it.
They update when the code does.
Useful workflow. One addition from Weft: make feedback merges reversible. Keep original records and a merge reason so a reviewer can split a mistaken theme without losing who asked for what. #SaaS #ProductManagement #CustomerFeedback
your customers are already telling you what to build all day.
sales calls, support tickets, Slack, reviews, surveys.
all of that feedback gets scattered across different tools before product ever sees the full picture.
here's the product feedback system i'd build:
1. send every customer-facing source into one feedback inbox. sales calls, support tickets, shared Slack channels, surveys, reviews + community posts all land there with the source link, date, customer/account, segment + the customer's exact words.
2. split conversations into individual findings. one sales call can contain a bug, two feature requests + a completely separate complaint. save each one separately and keep the transcript excerpt or timestamp attached so anyone can jump back to the original conversation.
3. keep the customer's wording next to the summary. save what they asked for, why they wanted it + what they were trying to accomplish. that context becomes important when five customers ask for slightly different versions of the same thing.
4. merge requests that describe the same customer problem even when the wording is different. keep every customer, quote + source attached to the shared theme so one issue can show the full amount of demand behind it.
5. tag each theme by product area, request type + the job the customer is trying to get done. this makes it much easier to see that 20 different requests are all coming from the same broken part of the product.
6. attach account data. customer count, customer tier, ARR, open opportunity + deals that are blocked by the request. now you can see the difference between something 40 free users mentioned and something affecting several of your largest accounts.
7. pull in product usage for the same customers. if people keep complaining about onboarding, look at what those accounts actually do during onboarding. if they ask for a feature, check whether they're using the surrounding part of the product at all.
8. rank the themes using demand, revenue exposure, strategic fit, effort + confidence in the evidence. keep the raw customer evidence underneath the score so someone can challenge the ranking without digging through five different tools.
9. keep one human triage step before anything reaches the roadmap. AI can extract requests, group duplicates + surface patterns, but someone should still review the evidence and decide whether the theme deserves discovery, gets parked or maps to work that's already planned.
10. when a theme moves forward, link the supporting feedback directly to the project or spec. engineering should be able to see who asked, what they were trying to do + the exact conversations that caused the work to exist.
11. sync the status back to the feedback system. when something moves from UNDER REVIEW → PLANNED → SHIPPED, the sales/support teams that logged those requests should know automatically.
12. close the loop with the customers who asked for it. you already know exactly which accounts requested the feature, so when it ships you have the list of people who should hear about it first.
every Friday, your product should be able to answer:
what kept coming up?
which customers are affected?
how much revenue is attached?
what does their product usage show?
which themes need discovery next?
once it can, you're set.
Useful read for SaaS teams: a handoff should preserve not just the decision, but the hypothesis behind it and what evidence would overturn it. At Weft, I’d make that part of the workflow review. #SaaS #ProductManagement #ProductStrategy
A large customer’s request still needs a product-fit review. Weft’s fictional blueprint weighs impact, urgency, evidence and effort, then records the reason for the decision. A reviewer should understand it without reconstructing a Slack conversation. #ProductManagement
A drafted customer update still needs review. In Weft’s fictional blueprint, the support lead checks the message, sends through the existing channel, and records sender, time and reply. Never promise unsupported functionality. #CustomerSuccess