Refounding: When product strategy gets a second life
Some startups are treating a mid-life reset as a real strategy, not a failure. Recent coverage from TechCrunch highlights companies like Airtable, Handshake, and Opendoor choosing to "refound" - shifting models and culture while signaling fresh intent.
Airtable framed its June update as a revival. Instead of stacking more AI features on top, leadership called for a rethink of the platform's direction. Handshake went the cultural route: bringing back high-urgency office rhythms, five days a week, with hours aimed at goals over comfort. The message is clear: product innovation and cultural standards move together.
Why this matters for product development
Refounding is a forcing function. It squeezes out slow bets, resets the mission, and gives product teams cover to ship what actually matters. AI is the catalyst here - not as a shiny add-on, but as a reason to simplify, automate, and re-architect around new user value and unit economics.
If you lead product, this is a rare chance to clear the slate, re-rank priorities, and rebuild momentum without the baggage of "how we've always done it."
A practical refounding playbook for product leaders
1) Reframe the product thesis
- Rewrite the product promise in one sentence. What pain do you erase, and for whom?
- Define the AI "job" in the workflow: speed, accuracy, cost, or new capability. Pick one.
- Kill features that don't serve the core job. Momentum beats breadth.
2) Ship AI that earns its keep
- Attach every AI feature to a measurable outcome: cycle time, activation, retention, or margin.
- Instrument model cost per task. Ship only if it improves unit economics or user outcomes.
- Add human-in-the-loop where stakes are high. Keep review UX fast and obvious.
3) Reset the culture to match the cadence
- Set weekly release rituals. Fewer meetings, tighter specs, more demos.
- Co-locate for hard pushes if it helps. Handshake chose five days; your bar might differ - just be explicit.
- Reward outcomes over output. Celebrate solved customer problems, not shipped tickets.
4) Rework operating metrics
- Core KPIs: time-to-value for new users, weekly active builders/creators, model-assisted task completion rate, and model cost per successful task.
- Quality guardrails: false positive/negative rates, escalation rate to human review, and customer trust signals.
- Portfolio view: prune or double down every 30 days based on KPI movement - not sentiment.
5) Build the AI foundation, not just features
- Data: clean event tracking, feature stores where needed, clear ownership for prompts and evaluations.
- Evaluation: offline test sets, golden datasets, and live A/B with feature flags.
- Governance: privacy, safety, and audit trails. Use recognized frameworks like the NIST AI Risk Management Framework.
30/60/90 plan to make it real
Days 0-30: Decide
- Clarify the product thesis and AI job. Write the one-pager that becomes the team's north star.
- Cut 20-30% of features or projects that don't serve the thesis.
- Pick two AI bets: one efficiency win, one user-facing capability.
Days 31-60: Ship
- Stand up evaluation, analytics, and cost tracking for AI features.
- Release weekly. Document learnings in public inside the company.
- Reset rituals: demos every Friday, decisions documented, goals visible.
Days 61-90: Scale or stop
- Double down on what moved core metrics. Kill the rest.
- Harden governance and reliability. Add fallbacks, rate limits, and escalation paths.
- Publish a simple "product contract" to customers: where AI helps, where humans decide, how feedback changes the product.
What the Airtable and Handshake moves signal
- Language matters. Calling it a "revival" or "refounding" gives teams permission to rethink without abandoning commitments.
- Culture is a feature. Handshake's in-office push pairs with higher pace - a clear bet on speed and clarity.
- Model over features. Airtable's stance suggests strategy first, AI second - features follow the model, not the other way around.
Common failure modes
- Shipping AI as decoration: no metric moves, costs creep, users get confused.
- Process whiplash: new rituals without purpose. Tie cadence to decisions and releases.
- Shadow AI: teams using unvetted tools. Provide sanctioned options and clear data rules.
If your team needs upskilling
If your roadmap leans into AI and you need structured training for product roles, these resources can help:
- AI courses by job role - pick tracks that map to product, data, and engineering.
- Popular AI certifications - useful for setting team standards and expectations.
Bottom line
Refounding isn't a slogan. It's a reset of product truth, operating cadence, and culture. Done well, it cuts noise, speeds learning, and makes AI earn its place in the product. That's the work.
Your membership also unlocks:
