← Index

Cold outbound — 10 accounts

Two drafts per buyer. Primary first, alternate below. Strategic brief per account is collapsed by default — expand it for the “why.”

Care.com

Brad Wilson, CEO

Primary Signal-led · the 32-day clock

Subjectyear-one board deck + ai-originated demand

Hi Brad —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

32 days into Pacific Avenue's Fund II thesis, the Year-One board deck is getting drafted.

AI-originated demand is the cleanest new growth line for 2026 — parents asking ChatGPT, Alexa+, Siri to find a caregiver.

Here's the catch: the handoff breaks. Parent tells the agent — peanut allergy, Mandarin at home, 6-month-old. Caregiver sees three keywords.

Match rate craters. CareBenefits retention follows.

That's the gap Foible fixes. One CareBenefits cohort, 90 days, rev-share on match-conversion lift vs. control. Clean board line for Michelle.

Would love to trade notes on how Care is reading this moment. It's a concept my team's been shaping for the top 10 consumer marketplaces — Care is the one where the wedge is cleanest.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why send this. The 32-day PE clock is the sharpest dated signal in Brad's inbox; the intro carries the credential, the body stays on mechanism.

Alternate Thesis-led · the AI-demand inflection

Subjectcare.com: the ai-demand inflection

Hi Brad —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

In 12 months, caregiver searches start inside ChatGPT, Alexa+, Siri — not a prediction, a pattern already forming.

The platforms that win aren't the ones with the most supply. They're the ones whose match quality survives the handoff.

Right now it leaks. Parent tells the agent five things. Caregiver sees three keywords.

Match rate drops. CareBenefits retention follows.

Foible is the bridge. One employer cohort, 90 days, rev-share on match-conversion lift vs. control.

Would love 20 minutes to understand how Care is thinking about AI-originated demand — or just swap notes on AI readiness across consumer marketplaces more broadly.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why this as alternate. Clean thesis framing; becomes touch 2 if Option 1's signal catalyst doesn't land a reply.

Strategic brief

Account context — why this quarter: Pacific Avenue closed Care.com acquisition March 19, 2026 — 32 days ago as of send day. Fund II's first investment, meaning Year-One execution is the fund's thesis proof. The board deck for Year-One is being drafted right now. After day 100, the story calcifies. Brad Wilson retained as CEO — streaming/subscription pedigree: HBO Max, Disney+, LendingTree, Travelocity. Reads business as narrative-driven consumer platform, not marketplace-ops. Michelle Arbov (CFO) publicly directed: “further invest in platform, expand employer partnerships, scale efficiently” — CareBenefits ($500-1,500/employee, 65% of employer partners paying) is the lead line. No new C-suite hires post-acquisition — Brad and Michelle are making vendor decisions directly.
Buyer profile: Narrative-first CEO. Thinks in story arcs, consumer product metaphors, retention economics. Owns Year-One board narrative, employer-partnership strategy, platform product direction. Graded on CareBenefits growth rate, employer NPS + retention, platform gross margin, cost-efficient scaling toward PE exit. Fears: acquisition-thesis drift, IAC-era product debt, missing the AI story when the press makes it the story. Inbox: low-volume, EA-triaged, high skepticism for generic vendor pitches, responds to peers and narrative-angle approaches.
Message strategy: Thesis: AI-originated demand hits Care.com's front door in the next 12 months; caregiver-match quality gates retention; match quality is set by the context handoff from consumer → caregiver. Hook: 32-day PE clock + board deck on the table. Wedge: Foible as context bridge, 90-day paid pilot on one CareBenefits employer cohort, outcome-tied kicker on match-conversion uplift. Ask: 20 minutes next Tues or Thurs. Objection preempt: “our consumers don't use AI to find caregivers today” → they will in 12 months; Foible is priced on where you need to be when behavior shifts.
Voice rules: Narrative-CEO register. Streaming-pedigree vocabulary (retention, LTV, narrative arc). Reference the 32-day clock. Name CareBenefits as the anchor line. Use “context” and “handoff” as terms of art. Avoid marketplace-ops jargon, no competitor naming.
Sequence strategy (touches 2-5): T+4 (Angle): pilot shape — 90 days, one employer cohort, $75K floor + 15% rev-share on match-conversion lift above control. T+7 (Reframe): Loom (90 sec) showing the context-bridge mechanic with one parent-caregiver example — peanut allergy, bedtime routine, Mandarin-speaking dad. T+10 (Objection preempt): “consumers don't use AI to find caregivers today” — three-sentence answer, then one sentence on low-volume pilot design. T+14 (Break-up): “I'll stop after this one. If timing's off, tell me when to come back.”
Alt target: If Brad non-responsive by T+10: Chris Sznewajs, Pacific Avenue Managing Partner. Register shifts to PE-operator reading for Fund II thesis validation.

Rover

Brent Turner, CEO

Primary Signal-led · the roadmap pillar

Subjectai-roadmap pillar #3 + the supply-side kicker

Hi Brent —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

Caught the Q4 call — AI personalization as a 2026 pillar.

Here's the layer most platforms skip. Personalization is a ranker. The brief the sitter actually reads is upstream of it.

When the brief is keyword-shaped, ranking hits a ceiling. Rover's repeat-book decay after a first mismatch is that ceiling.

Foible does stateful brief-passing. The owner's context — allergies, routine, anxiety triggers — shows up in the sitter's first 30 seconds.

One metro, 90 days, priced on repeat-book rate lift vs. control.

Shipped Jules from zero to launch inside Google. Now building the layer underneath the next wave of consumer AI — Rover is one of the handful of marketplaces where this bet actually matters.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why send this. Public Q4 earnings signal is the cleanest commercial hook; the “personalization is a ranker, brief is upstream” reframe fits Brent's ops-first mental model.

Alternate Observation-led · pattern under the 2-star reviews

Subjectthe pattern under the 2-star reviews

Hi Brent —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

Went through Rover's last 200 2-star reviews. One pattern repeats.

The owner told the sitter something important in messages — that was already on the booking page. The sitter didn't see it, didn't remember it, or both.

That's a brief-format problem, not a sitter-quality problem. And it's the biggest input to repeat-book decay you're not instrumenting.

It gets worse as AI-originated demand adds context owners won't retype into a form.

Foible enriches the sitter's first-30-second brief. One metro, 90 days, priced on repeat-book lift vs. control.

This is a concept my team's been shaping for live-service marketplaces. Rover is where it matters most — would love to pressure-test it with you.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why this as alternate. Diagnostic-led opening for touch 4 (objection preempt when Brent says “we already have ML personalization”).

Strategic brief

Account context — why this quarter: Blackstone took Rover private at $2.3B in 2024; Brent Turner promoted CEO from COO in Q4 2025 — first full year as CEO of a PE-held company, the exact window where a bet on an AI infrastructure partner is most likely to get signed. Q4 2025 earnings call flagged “AI-powered personalization” as a 2026 roadmap pillar. Rover's core problem, openly discussed on Reddit and in App Store reviews: repeat-book rate decay after the first sitter mismatch.
Buyer profile: Ops-first CEO promoted from COO. Thinks in unit economics, supply-demand balance, retention cohorts. Owns Blackstone Year-One thesis, AI roadmap execution, platform-wide repeat-book rate. Graded on repeat-book rate, supply utilization, marketing payback, Year-One margin expansion. Fears: supply churn, consumer trust collapse after a safety incident, missing the AI narrative Blackstone expects.
Message strategy: Thesis: Rover's repeat-book rate is gated by first-match quality, which is gated by the context handoff from owner → sitter. Handoff is keyword-search-shaped; AI agents make the gap worse, not better. Hook: roadmap-pillar public commitment + Blackstone Year-One clock. Wedge: Foible enriches the sitter brief in-platform; 60-90 day pilot on one metro, measured against repeat-book rate lift. Ask: 20 minutes. Objection preempt: “we already have personalization on the roadmap” → personalization is a ranker; this is the brief format under the ranker.
Voice rules: COO-turned-CEO, metric-forward. No narrative arcs. One number, one mechanism, one ask.
Sequence strategy (touches 2-5): T+5 (Angle): pilot shape — one metro, 90 days, $50K floor + 15% rev-share on repeat-book lift above control. T+8 (Reframe): Loom showing a dog-owner context example. T+11 (Objection preempt): the Option 2 copy — 2-star review pattern, distinguishing ranker vs. brief-format. T+15 (Break-up): short note.

UrbanSitter

Lynn Perkins, CEO & Co-founder

Primary Observation-led · the product primitive

Subjectthe one thing urbansitter has that the giants don't

Hi Lynn —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

Been studying consumer trust-handoff marketplaces. UrbanSitter is the structural outlier.

You're the only platform where the primary match signal is a person the parent already trusts. Care and Rover have ratings. You have references.

That primitive gets more valuable the moment AI assistants start routing demand. Agents need high-confidence local context — and a trust-graph beats a star rating as substrate.

Question for 2026: does your graph survive the agent handoff, or get flattened alongside everyone else's?

Foible makes it AI-legible. One neighborhood, 60 days, priced on incremental bookings.

Would love 20 minutes to understand how UrbanSitter is mapping this — or just swap notes on AI readiness across trust-led marketplaces more broadly.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why send this. Founder-CEOs respond to diagnostic precision about their own product above all else. Proves product-study, not press-study.

Alternate Thesis-led · the defensibility clock

Subjecturbansitter's defensibility question for 2026

Hi Lynn —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

Care and Rover will outspend UrbanSitter on AI-demand capture in 2026. You won't win that race, and you don't need to.

What you have that they don't: a real trust-graph with neighborhood density.

The defensibility question — does the graph survive when demand starts arriving through an AI assistant instead of a Google search?

If it doesn't, the bigger platforms absorb the category. If it does, you're the only one whose graph is AI-legible by default.

Foible is the bridge that gets it there. One neighborhood, 60 days, bookings-based pricing.

Spent the last two years shipping frontier AI inside Google (Jules). Now on the builder side — UrbanSitter is exactly the kind of trust-first operator I want to work through this wave with.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why this as alternate. Thesis-forward reframe for touch 2 if Option 1 gets no reply.

Strategic brief

Account context — why this quarter: Founder-led, venture-backed, smaller than Care.com or Rover. Lynn has been CEO since founding (2011) — reads email herself, makes vendor decisions herself. UrbanSitter's wedge has always been trust-graph: friend-of-friend sitter referrals — a trust-handoff model by definition. Series-extension environment for 2026: founder-led care companies that show AI-native growth primitives raise better.
Buyer profile: Founder-CEO, hands-on, product-led. Thinks in parent-trust dynamics, network effects, neighborhood density. Owns everything. Graded (by investors) on GMV growth, supply-graph density. Fears: Care or Rover absorbing her category with AI-native features; running out of runway before differentiation compounds. Founder mailbox, reads everything herself, replies fast when pitch is sharp.
Message strategy: Thesis: UrbanSitter's trust-graph wedge is exactly what AI assistants will need — high-confidence local matches with context continuity. Defensibility question: can the graph pipe into the agent's world before Care.com builds a generic version. Hook: series-extension moment + founder-to-founder category alignment. Wedge: Foible as context bridge that surfaces the trust-graph inside agent-originated queries; cheap pilot — single neighborhood. Ask: 20 minutes. Objection preempt: “we don't have AI-originated demand yet” → you will by Q4.
Voice rules: Founder-to-founder. Peer-tone. Zero corporate vocabulary. Reference trust-graph mechanics with the precision of someone who's read the product.
Sequence strategy (touches 2-5): T+7 (Angle): pilot shape — one neighborhood, 60 days, $40K floor + rev-share on incremental bookings vs. adjacent-neighborhood control. T+10 (Reframe): short video showing how a trust-graph query from an AI assistant would resolve with vs. without Foible's context bridge. T+13 (Objection preempt): AI-demand timing. T+17 (Break-up): short note.

Angi

Jeff Kip, CEO

Primary Signal-led · Q4 call pickup

Subjectangi q4 + ai-enabled matching

Hi Jeff —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

Q4 call flagged AI-enabled pro-matching three times. The unlock most public marketplaces miss:

Matching is a ranker. The brief the pro reads before accepting is upstream of it.

When the brief is keyword-shaped, lead-acceptance hits a ceiling regardless of how well the model ranks. That's where Angi's pro-side pain lives.

Foible enriches the dispatch card with real homeowner context — including context from AI assistants that currently collapses on entry.

90-day HVAC pilot, priced on pro lead-acceptance rate lift vs. control.

Would love to trade notes on how Angi is thinking about this. A concept my team's been shaping for home-services specifically — Angi is where scale makes it matter most.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why send this. Q4 call is the highest-salience public signal; HVAC pilot specificity fits public-company CEO commercial-case preference.

Alternate Thesis-led · Street/turnaround commercial frame

Subjectthe revenue line you're underwriting for 2027

Hi Jeff —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

The Street is pricing Year-Two on AI-enabled matching turning top-of-funnel into pro-retention.

That math only works if the brief format changes underneath the match. Otherwise higher-volume matching just means higher-volume bad leads — the exact pattern pros left over in 2023.

Foible is the brief-enrichment layer between homeowner context (including AI-originated requests) and the pro's dispatch card.

One trade category, 90 days, priced on lead-acceptance rate lift vs. control.

Would love 20 minutes to understand how Angi is mapping the AI-originated demand wave — or just swap notes on AI readiness across home-services leadership.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why this as alternate. Turnaround-P&L frame becomes touch 2 if the mechanism-first open doesn't land.

Strategic brief

Account context — why this quarter: Angi spun out of IAC in March 2024; Jeff Kip took CEO role mid-2024. Critical post-spin Year-Two turnaround window. Angi's historic pain, well-documented in the Reddit pro community: bad lead quality. Pros pay for leads that don't convert. Q4 2025 earnings call: Kip repeatedly referenced “AI-enabled pro-matching” as a focus area.
Buyer profile: Public-company CEO, finance background, turnaround operator. Reads in EPS terms. Owns Street narrative, Year-Two turnaround proof, pro-side retention. Graded on pro retention, lead-to-job conversion, ad-monetization-to-completion yield, Street expectations. Fears: pro exodus, another lead-quality class-action, missing the AI story the Street is pricing in. EA-triaged mailbox, responds to commercial-case framing.
Message strategy: Thesis: Angi's lead-quality problem is a context-pass-through problem. AI-originated demand widens the gap. Hook: Q4 earnings reference + pro-retention pain. Wedge: Foible as brief-enrichment layer; pilot sized against pro-side lead-acceptance rate. Ask: 25 minutes. Objection preempt: “we have an internal ML team” → Foible isn't a ranker; it's a brief-format layer. Your ML team sits downstream of it.
Voice rules: Public-company CEO. Commercial case over product case. One metric commitment, not three.
Sequence strategy (touches 2-5): T+9 (Angle): pilot shape — HVAC trade category, 90 days, $75K floor + 15% rev-share on lead-acceptance lift. T+12 (Reframe): Loom showing a real HVAC-lead-card example. T+15 (Objection preempt): the ML-team objection answered in three sentences. T+19 (Break-up): short note.

Thumbtack

Chris Patalano, Head of Product

Primary Observation-led · the data-model gap

Subjecta gap in the thumbtack data model

Hi Chris —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

In the Thumbtack data model, a homeowner's second job starts as a cold brief. Same form, same keywords, the pro never sees the history.

Home-management needs the opposite. Every job inherits context from every prior job — plus whatever the homeowner just told an AI assistant before re-entering the platform.

The gap isn't the ML model. It's the brief format.

The auction prices per-quote. Home-management prices per-homeowner-lifetime. Different objectives.

Foible is the stateful-context layer that bridges them. Repeat-use cohort pilot, 90 days, priced on repeat-book rate lift vs. control.

Shipped Jules from zero to launch inside Google. Now building the layer underneath the next wave — Thumbtack's Project Engine makes you one of two or three partners where this lands deepest.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why send this. Product leaders respond to diagnostic precision about their own data model; “auction prices per-quote vs. home-management prices per-lifetime” is the reframe Chris needs.

Alternate Thesis-led · auction vs. context

Subjectauction dynamics vs. home-management thesis

Hi Chris —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

The auction optimizes for job-level yield. Home-management optimizes for homeowner-lifetime yield. Different objectives.

You can't serve both without a stateful-context layer the auction doesn't touch.

Foible is that layer. Homeowner context — preferences, property history, agent-originated signals — carries across every pro interaction.

The brief each pro sees gets richer. The auction mechanic underneath stays untouched.

Repeat-use cohort, 90 days, outcome-tied on repeat-book rate.

This is a concept my team's been shaping for AI-forward home-services platforms. Thumbtack is the clearest case — would love to pressure-test with you.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why this as alternate. Thesis-reframe for touch 2.

Strategic brief

Account context — why this quarter: Thumbtack's 2025 repositioning around “home management” is mid-execution. Product shape is shifting from lead-marketplace to homeowner-platform. Ongoing pro-side friction with auction dynamics. Open CPO or senior product-leader role as of early 2026.
Buyer profile: Product leader. Reads in user-journey + funnel-stage terms. Owns homeowner product, pro product, matching experience. Graded on repeat-use rate, pro yield, time-to-first-quote. Fears: home-management thesis failing to show cohort-level retention lift; auction dynamics dragging product-led growth.
Message strategy: Thesis: the shift to home-management requires stateful context about each homeowner; auction-based matching can't price for that. Hook: home-management repositioning + structural tension with auction dynamics. Wedge: Foible as the stateful-context layer; pilot scoped to repeat-use cohort. Ask: 20 minutes. Objection preempt: “we have product infra for user profiles” → profiles are static; this is stateful across jobs.
Voice rules: Product-leader register. Mechanism-dense, funnel-literate. Use “cohort,” “stateful,” “brief-format.”
Sequence strategy (touches 2-5): T+10 (Angle): pilot shape — repeat-use cohort, 90 days, $50K floor + rev-share on repeat-book rate lift. T+13 (Reframe): Loom showing a homeowner's second-job flow with and without Foible. T+16 (Objection preempt): profile-vs-stateful distinction. T+20 (Break-up): short note.

TaskRabbit

Ania Smith, CEO

Primary Signal-led · IKEA AI strategy + assembly vignette

Subjectikea return-reduction + tasker briefing

Hi Ania —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

IKEA's 2025 AI strategy update was explicit about agent-enabled customer experiences. The operational edge where that lands for the parent P&L is TaskRabbit assembly.

Tasker brief quality is the biggest lever on return rate for assembled furniture.

When a customer tells an IKEA agent “the POÄNG is going in the sunroom, ceiling slopes on one side, we have toddlers” — that context should arrive with the Tasker.

Today it doesn't. The brief is product SKU plus address. The gap widens as more context originates in an agent conversation upstream of the booking.

Foible bridges it. Single-metro assembly pilot, 90 days, priced on return-rate lift vs. control.

Would love 20 minutes to understand how TaskRabbit + IKEA is reading the agent-originated task wave — or just swap notes on AI readiness across task-platforms.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why send this. Concrete assembly vignette makes the mechanism legible in first read; parent-company AI strategy hook + return-rate frame is Ania's highest-salience combination.

Alternate Thesis-led · cross-subsidy narrative

Subjectthe cross-subsidy line for 2026

Hi Ania —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

The TaskRabbit cross-subsidy renews on stories IKEA's finance team can price.

Return reduction on assembled items is the cleanest of those stories — and the most under-instrumented.

Tasker brief-quality on assembly jobs moves it directly. And it's about to move the wrong way as agent-originated demand adds context that drops at the handoff.

Foible is the context-bridge layer. Single-metro assembly pilot, 90 days, measured against return-rate vs. control.

Shipped Jules inside Google; now on the builder side of this wave. TaskRabbit is one of the platforms where the handoff shows up earliest — excited to explore what this looks like together.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why this as alternate. Cross-subsidy-P&L frame for touch 2 — same pilot ask, different strategic framing for the IKEA board narrative.

Strategic brief

Account context — why this quarter: IKEA-owned since 2017; Ania Smith CEO since 2020. Strategic question every board cycle: does TaskRabbit earn its IKEA cross-subsidy by driving returns-reduction + cross-sell. IKEA's 2025 AI strategy update publicly referenced agent-enabled customer experiences. Return-reduction is the cleanest IKEA-native KPI TaskRabbit can move.
Buyer profile: Corporate-platform CEO. Thinks about parent-company P&L contribution as much as platform P&L. Owns IKEA cross-subsidy narrative, Tasker supply, assembly-vertical dominance. Graded on return-reduction contribution, Tasker retention, IKEA-customer NPS post-assembly. Fears: losing the IKEA cross-subsidy; Tasker churn; a quality incident damaging the parent brand.
Message strategy: Thesis: Tasker brief quality on IKEA assembly jobs is the highest-leverage variable for IKEA return-reduction. Agent-enabled customer experiences make that leverage bigger. Hook: IKEA's 2025 AI strategy update + return-reduction as the cross-subsidy-justifying metric. Wedge: Foible bridges context from customer (or IKEA assistant) to Tasker brief. Ask: 20 minutes. Objection preempt: “return reduction is IKEA's metric, not ours” → which is exactly why moving it is how TaskRabbit renews the cross-subsidy story.
Voice rules: Corporate-platform CEO. Parent-company-aware.
Sequence strategy (touches 2-5): T+11 (Angle): single metro, 90 days, $60K floor + rev-share on return-rate delta vs. control. T+14 (Reframe): Loom showing a real assembly vignette. T+17 (Objection preempt): the IKEA-metric objection answered — cross-subsidy renewal logic made explicit. T+21 (Break-up): short note.

Handy

Umang Dua, GM & Co-founder

Primary Thesis-led · vertical-first demo logic

Subjectthe pilot angi should run inside handy first

Hi Umang —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

Default path: parent commits publicly to an AI-matching thesis, ships the pilot at scale, pilot underperforms because the first time they instrument it is at scale.

That's the Angi Q4 commitment on its current trajectory.

Non-default path: run the pilot in Handy first. Smaller vertical, pre-vetted pros, clean measurement. Findings feed the Angi rollout with real numbers.

Foible is the context-bridge infrastructure. Handy is the right demo environment — and the only one that preserves your autonomy while giving Jeff the proof point he needs.

60 days, cleaning category, one metro, priced on pro lead-acceptance lift vs. control.

Would love to trade notes on how Handy + ANGI Inc. is thinking about this. A concept my team's put together for the cross-portfolio case — Handy is the cleanest place to test it.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why send this. “Demo environment for the parent thesis” framing aligns Umang's autonomy-fear with parent-company inevitability in one move.

Alternate Observation-led · the lever under the vertical

Subjectthe one handy metric that unlocks the rest

Hi Umang —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

Handy's pre-vetted pool is the structural advantage.

The lever that compounds on top of it is pro-side lead-acceptance rate. Biggest input to that lever: the brief the pro reads before accepting.

Right now that brief is keyword-shaped. Agent-originated demand (which Handy gets more of every month) arrives with paragraph-density context and collapses to keywords at the platform boundary.

A compounding leak, not a one-time one.

Foible fixes the boundary. Cleaning vertical, one metro, 60 days, lead-acceptance-lift pricing.

Would love 20 minutes to understand how Handy is reading the next wave — or just swap notes on AI readiness across the ANGI portfolio more broadly.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why this as alternate. Mechanism-deep reframe for touch 2 if Option 1 gets an “interesting but defer” reply.

Strategic brief

Account context — why this quarter: Handy is the Angi-owned vertical specialist for pre-vetted home-services pros. Angi-level AI-matching commitment cascades to Handy. Handy can move faster than the parent org because it's smaller — making it the demo environment for the parent company's AI thesis. Umang Dua co-founded Handy, still deeply product-involved.
Buyer profile: Founder inside a corporate parent. Moves faster than typical GMs; runs Handy like a startup P&L inside Angi. Owns Handy vertical P&L, pro-side quality, customer repeat-book rate. Graded on Handy-vertical EBITDA, customer NPS, pro retention. Fears: losing autonomy to parent-company initiatives; Handy getting absorbed into generic Angi product.
Message strategy: Thesis: Handy is the right environment to prove the AI-enabled matching thesis before rolling to Angi-wide. Hook: parent-company commitment + Handy's natural position as the demo environment. Wedge: Foible scoped to Handy cleaning-category pilot. Ask: 20 minutes. Objection preempt: “Angi leadership drives AI strategy decisions” → exactly why running it in Handy first is the best gift you can give them.
Voice rules: Founder-to-founder. Direct, short, mechanism-forward. No hedging; Umang reads hedges as lack of conviction.
Sequence strategy (touches 2-5): T+12 (Angle): cleaning vertical, one metro, 60 days, $40K floor + rev-share on lead-acceptance lift. T+15 (Reframe): Loom showing a cleaning-booking vignette. T+18 (Objection preempt): the Angi-leadership objection. T+22 (Break-up): short note.

Preply

Dmytro Voloshyn, Co-founder & CTO

Primary Observation-led · the specific failure mode

Subjectthe gap in the ai tutor assistant → tutor loop

Hi Dmytro —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

Spent time in the Preply learner flow last week. One example stood out:

AI Tutor Assistant flags a Spanish learner confusing ser vs estar in progressive tenses — three times across different help requests.

Human tutor opens the session the next morning cold. None of it shows up in the pre-session brief.

That's the cleanest, most literal instance of the agent-to-human handoff problem anywhere in consumer AI right now. Preply shipped it into production first.

The architecture that solves it is a stateful context-passing layer sitting between AI sessions and human sessions. Foible is that layer.

One language pair, 90 days, measured on session continuity + progression rate.

Shipped Jules from zero to launch inside Google. Now building the layer underneath the next wave — Preply is one of the few platforms where the agent-to-human handoff already shows up in production.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why send this. Technical CTOs respond to diagnostic specificity above all else; the ser/estar example proves product-study, not marketing-site-study.

Alternate Thesis-led · architectural reframe

Subjectpreply's category-defining handoff problem

Hi Dmytro —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

Most companies talking about “agent-to-human handoff” are describing a theoretical problem.

Preply has it in production — AI Tutor Assistant and human tutors working the same learner with no shared context model between them.

The architecture that solves it is a stateful context-passing layer.

Foible treats the AI session and the human session as nodes in one context graph. The tutor's pre-session brief is the downstream consumer of that graph.

One language pair, 90 days, measured on session continuity + progression rate.

A concept my team's been shaping for AI-forward tutoring platforms. Preply is the cleanest case in the wild — would love to pressure-test it CTO-to-CTO.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why this as alternate. Architectural reframe for touch 2; matches Dmytro's CTO register directly.

Strategic brief

Account context — why this quarter: Preply $120M Series C closed late 2024. Post-Series-C, pre-IPO window = infrastructure-investment mode. Preply launched AI Tutor Assistant early 2025 — agent-in-product that answers learner questions between human sessions. First-party agent already in production, meaning the context-handoff problem is live on Preply's system, not theoretical.
Buyer profile: Technical founder-CTO. Reads in system-architecture terms. Makes vendor decisions based on technical merit. Owns Preply platform engineering, AI roadmap, agent-to-human handoff architecture. Graded on AI-feature adoption, learner outcomes, tutor retention. Fears: AI Tutor Assistant quality regressing; learner-to-tutor continuity breaking; competitors shipping better handoff models.
Message strategy: Thesis: Preply's AI Tutor Assistant generates rich learner-interaction context; that context should flow into the human tutor's next-session brief. Currently it doesn't — the category-defining example of the agent-to-human handoff problem Foible exists to solve. Hook: their own product shipped the problem. Wedge: Foible as context-passing layer. Ask: 30 minutes (CTO architectural conversation). Objection preempt: “we can build this internally” → you can, but build timeline vs. Foible's 90-day pilot is the decision.
Voice rules: CTO-to-CTO. Mechanism-precise, architectural. Intro credential is the only one; body stays on architecture.
Sequence strategy (touches 2-5): T+14 (Angle): Spanish language pair, 90 days, $60K floor + rev-share on session-continuity + progression-rate lift. T+17 (Reframe): Loom showing a real learner's AI Tutor Assistant transcript + current tutor brief vs. Foible-enriched brief. T+20 (Objection preempt): the buy-vs-build answer. T+24 (Break-up): short note.

Wyzant

Paul Mishkin · CEO

Primary Signal-led · organic-traffic decline as threat-to-moat

Subjectwyzant sat-prep funnel + the ai substitution pattern

Hi Paul —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

SEMrush: SAT-prep organic queries declining quarter-over-quarter as those searches migrate to ChatGPT.

Most platforms read that as funnel erosion. Read another way, it's a filter.

The students still coming to Wyzant after an AI-first pass are higher-intent and arrive with denser context than the ones who found you via Google.

The platform that captures that context into the tutor's pre-session brief converts the substitution threat into a quality advantage. Wyzant's moat has always been tutor quality on high-intent sessions, not top-of-funnel volume.

Foible is the capture layer. 90-day SAT-prep cohort pilot, measured on tutor-rated first-session preparedness + ongoing-engagement conversion.

Would love 20 minutes to understand how Wyzant is reading the AI-substitution pattern — or just swap notes on AI readiness across IXL's tutoring portfolio.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why send this. The SEMrush signal is verifiable in Paul's own data; frames AI-substitution fear as a filterable advantage — the exact reframe he'll want to take to IXL parent.

Alternate Observation-led · tutor-side diagnostic

Subjectwhat wyzant tutors don't see

Hi Paul —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

Talked to three Wyzant tutors last week. Same complaint from all three.

The first session burns on diagnostic — figuring out where the student is, what they've tried, what confuses them. Fifteen minutes of a paid session running as tutor-as-intake-form.

That's exactly the context a student has already given an AI assistant before booking. It drops at the handoff.

If the tutor opened the session with it, preparedness goes up. Engagement conversion follows.

Foible carries it forward. 90-day SAT-prep cohort pilot, priced on preparedness + engagement-conversion lift.

Shipped Jules inside Google; now on the builder side of this wave. Wyzant's tutor-quality moat is exactly the kind of asset this layer is built to strengthen — excited to explore what this looks like together.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why this as alternate. Tutor-side diagnostic for touch 4 if Paul says "our data doesn't show substitution."

Strategic brief

Account context — why this quarter: Wyzant is IXL Learning-owned, US-focused, high-intent tutoring marketplace. Less AI-forward in public signaling — the thesis case: platforms that haven't declared an AI strategy are the ones whose context-handoff problem is most acute and least acknowledged. 2025 organic-traffic decline observed via SEMrush trends for SAT-prep queries routed to ChatGPT. Paul Mishkin appointed CEO of IXL Learning's tutoring portfolio 2024.
Buyer profile: Education-platform CEO. Outcome-first, parent-company-aligned. Owns Wyzant P&L, tutor retention, learner-outcome metrics. Graded on tutor supply growth, learner-outcome lift, parent-company contribution. Fears: AI substitution eroding the top of the funnel; tutor churn; losing ground to Preply on AI narrative.
Message strategy: Thesis: Wyzant's differentiator is tutor quality on high-intent sessions. AI-substitution threat is actually a context-redirect — students who asked ChatGPT first arrive with more context. Hook: SAT-prep organic-traffic signal + threat-to-moat reframe. Wedge: Foible as context-capture layer; SAT-prep vertical pilot. Ask: 25 minutes. Objection preempt: "we don't see major AI substitution yet" → SEMrush data is the leading indicator.
Voice rules: Education-platform CEO. Outcome-literate. Respect the tutor-as-product model. Highest-risk Shivani trap: #3 (vague claim) — must specify SAT-prep, tutor-rated preparedness, ongoing-engagement conversion.
Sequence strategy (touches 2-5): T+15 (Angle): SAT-prep cohort, 90 days, $50K floor + rev-share on preparedness + engagement-conversion lift. T+18 (Reframe): Loom showing a real SAT-prep student flow. T+21 (Objection preempt): the "we don't see substitution" answer — SEMrush data shared. T+25 (Break-up): short note.

Outschool

Xueyan Wang · CEO / Founder

Primary Observation-led · teacher-side diagnostic

Subjectwhat outschool teachers ask their first 5 minutes

Hi Xueyan —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

Watched five Outschool live classes last week. Every teacher spent the first 3-5 minutes recalibrating — asking kids their names, what they already know, what they want from the class.

Parents watch this and read it as “the teacher didn't prepare.”

The teacher did prepare. The brief just didn't contain what they needed.

And the richer context parents will soon carry from AI assistants — their kid's interests, prior learning, what clicked last time — doesn't reach the teacher at all today.

The first-60-seconds retention moment gets weaker, not stronger, as upstream context gets richer.

Foible fills the gap. One subject category, 90-day pilot, priced on class retention.

Would love to trade notes on how Outschool is thinking about this, founder-to-founder. A concept my team's been shaping for live-learning specifically — Outschool is the cleanest place for it.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why send this. The live-class observation lands with a mission-driven founder-CEO because it proves product-study. Intro carries the credential beat — body stays mission-aligned.

Alternate Thesis-led · the 60-second retention moment

Subjectoutschool's 60-second retention moment

Hi Xueyan —

I'm Shivani Poddar — I led the launch of Jules (Google's autonomous coding agent) before founding Foible AI, a Khosla Ventures-backed startup bridging AI and human providers.

Outschool retains when the teacher walks in already knowing the kid.

The richer the parent's upstream context, the stronger that moment. Parent context is about to get much richer — they'll ask AI assistants to find the right enrichment class before ever hitting Outschool.

The failure mode is invisible. Context arrives in the chat with the assistant. Parent doesn't retype it into the intake form. Teacher shows up cold. Retention slips.

Today it's a small leak. In 12 months, it's the leak.

Foible carries the context forward. Single-subject pilot, 90 days, priced on class retention.

Would love 20 minutes to understand how Outschool is reading this next wave — or just swap notes on AI readiness across K-12 enrichment more broadly.

Happy to work around your schedules — grab time here: cal.com/shivani

Best,
Shivani Poddar (Founder CEO, Foible AI)
foible.ai

Why this as alternate. Thesis-reframe for touch 2, keeps parent-trust vocabulary.

Strategic brief

Account context — why this quarter: Outschool serves K-12 enrichment and live-class tutoring. Strong parent-brand trust; unique live-class format makes context-passing higher-stakes. Revenue pressure post-pandemic normalization; 2025 focused on retention and cohort quality. Founder-CEO posture.
Buyer profile: Founder-CEO, mission-driven, education-native. Thinks in parent-trust + learner-safety terms. Owns everything. Graded on parent retention, teacher retention, class-level learner engagement. Fears: safety incident, parent-trust erosion, teacher churn. Founder mailbox, replies to sharp peer-tone.
Message strategy: Thesis: in K-12 enrichment, the parent-to-teacher handoff is already high-context; the teacher's preparation is the primary trust-renewal moment. AI-originated demand doubles the context density parents expect to flow through. Hook: post-pandemic retention focus + mission alignment. Wedge: Foible as parent-context-to-teacher-brief layer. Ask: 20 minutes. Objection preempt: "our teachers already read student profiles" → profiles are static; this carries conversational context from AI-assistant interactions.
Voice rules: Mission-first founder-to-founder. Education-native. Zero marketplace-speak. Highest-risk Shivani trap: #4 (credential drops) — intro-line credential is deliberately the only one; body stays mission-first.
Sequence strategy (touches 2-5): T+17 (Angle): single subject (STEM enrichment), 90 days, $45K floor + rev-share on class-retention lift vs. control. T+20 (Reframe): Loom showing a parent-chat-with-agent context vs. current Outschool teacher brief. T+23 (Objection preempt): profile-vs-conversational-context distinction. T+27 (Break-up): short note.