XMarts room

Summer 2024

Handled

AI teammate for finance and operations teams at consumer brands

Visit website

What they do

AI teammate for finance and operations at consumer brands

How close to us

5, very close

Learns a brand's data and rules, catches issues, does recurring finance work with approvals; close to W7 for SMEs.

Wedges

  • W7 Find and fix margin leaks direct
  • W5 One layer for one type of small firm adjacent
L1 Company brain (knows the organisation) 2, clear
L2 Agent team (runs agents, tools, approvals) 2, clear
L3 Does and learns (finishes, checks, measures, updates the brain) 2, clear

Who buys

Firm size
sme
Buyer type
business
Role
finance or operations lead at consumer brands (CFO, head of ops, fractional ops)
UK
not known
Their own description (from YC)

Consumer brands run finance and operations across ERPs, commerce platforms, 3PLs, accounting systems, email, and spreadsheets. Because these systems do not share context or complete work, teams still spend hours assembling reports, reconciling numbers, entering orders, monitoring inventory, processing invoices, and chasing exceptions. Problems are often discovered only after they have cost the business time, margin, or revenue.

Handled is an AI teammate for finance and operations. It connects to the systems and unstructured information a brand already uses and learns its data, rules, and workflows. Handled answers questions about the business, proactively catches issues, and takes on recurring manual work across financial reporting, inventory, procurement, wholesale order-to-cash, and AP/AR. It works across systems and brings a person in when judgment or approval is required.

Handled gives lean teams more operating capacity without adding another tool to manage, more outsourced support, or more headcount.

Full teardown

Handled (YC Summer 2024): teardown vs XMarts

In one line: Handled is an AI teammate for finance and operations at consumer goods brands: it reads their systems and inboxes, answers questions, flags problems, and does recurring work such as reports, orders, stock and invoices, with approval rules [site] [yc]. Threat: 2, because it is the same buyer type (finance or ops lead at a small firm) and the same margin and money-leak problem as W7, but it is 3 people, sells to one industry only, and shows no confirmed funding for Handled itself and no formal case study [yc] [site]. It is the closest YC rival we have seen to W7, so expect a partner to name it.

1. Fact file

Item Finding Source
What they say Headline: "The AI teammate made for the messy reality of CPG." YC line: "AI teammate for finance and operations teams at consumer brands." [site] [yc]
Founded, team, funding Founded 2023, YC S24, team of 3, New York City. Funding for Handled itself: none found. Its earlier name Zimi got a seven-figure Fearless Fund investment (TechCrunch, 11 Oct 2024); a secondary page puts it at a $2M seed (UNVERIFIED). [yc] [press]
Founders Audrey Djiya, CEO. She ran operations, international expansion and go-to-market at e-commerce firms including Peloton, launched fintech products at Duplo (YC W23), has a Stanford MBA, and was named to Forbes 30 Under 30 for Retail and E-commerce [yc] [press]. A profile names Peter Nsaka as her Zimi co-founder [press]; he is not on the YC page or the site. What YC likely saw: a founder who lived consumer-brand operations at scale (inference). [yc] [press]
Customers Logos on the site: Hormbles Chormbles, Stemregen, Delta, Bembien, The Shirt, Baobab, Boxraw. Two short quotes: a CFO (reporting, cost savings) and a fractional head of ops (wholesale orders in odd formats). No formal case study. [site]
Price Not public: "Book a demo" only. Job ads: none listed. [site] [yc]
Path Zimi (2023 to 2024) became Handled. TechCrunch and Afrotech (Oct 2024) describe Zimi as logistics (inventory, storage, border logistics) for businesses in emerging markets such as Africa selling abroad. Since then: an AI operations teammate for US consumer brands and 3PLs. No dated rename post found. [press]

Other facts: the site claims 170+ integrations, 3 to 4 days to set up, month-end close cut from 12 days to 1, weekly reports from 2 hours to 2 minutes, "10+ hours" a week saved per team, and more than $150k a year of billing errors and refund leaks caught. These are what they claim, not checked [site]. Audrey Djiya pitched at Groceryshop on 23 Sep 2026 in a cost-savings session [press]. The demo page quote is signed "Placeholder Name", so at least one testimonial is still a placeholder [site].

2. Their proposition, reverse engineered

Part What they do
Who buys Finance and ops leads at small and mid-sized consumer goods brands; logos are brand names, not big groups [site]. Separate sign-in for brands and for 3PLs (warehouse partners) [site].
The pain Systems do not share context, so teams spend hours on reports, reconciling numbers, entering orders and chasing exceptions; problems are found after they cost margin [yc].
The promise More operating capacity without more tools, outsourcing or headcount [yc]. Catch issues early, run recurring work [site].
How it works Connects to ERP, store, marketplace, 3PL, accounting, email, spreadsheet and Slack data and learns the brand's data, rules and workflows [yc] [site]. Rules decide what runs alone (for example small purchase orders) and what waits for sign-off. Every action is logged [site].
The output Answers to business questions; reports (for example margin by channel); reconciled stock; processed wholesale orders; purchase orders; invoice and refund-leak checks [site].
Proof style Time cut (12 days to 1, 2 hours to 2 minutes) and money caught ($150k+ a year), all they claim. No named before/after case [site].
Who sells it Demo booking with the founder's email as the only contact, so founder-led sales (inference) [site]. No job ads [yc].
Price model Not public [site].
The story CPG back-end work is messy, spread over many systems; a teammate that keeps working in your own tools beats another dashboard [site] [yc].

3. Positioning

  • Category they claim: AI teammate for finance and operations at consumer brands [yc].
  • Category a YC partner would put them in: vertical AI for back-office and finance ops in CPG (inference).
  • What they want to be known for: the ops and finance teammate that works in the messy mix of CPG systems and emails.
  • Where they sit on the map: layers L1/L2/L3 = 2/2/2, buyer = sme, business.

4. Three layers vs ours

Layer Them (score + evidence) Us (Marton's logs show)
L1 Company brain 2. "Learns its data, rules, and workflows" from systems plus email, sheets and Slack, and reads from it in later work [yc] [site]. Not sold as the memory itself, and no public sign of what is stored or how it is kept fresh. A vault of about 1,400 pages, 112 call transcripts and about 8,700 customer quotes, searchable by meaning.
L2 Agent orchestration 2. Several workflows (orders, stock, purchasing, AP/AR) with approval rules and a hand-off to a person [site] [yc]. Whether it is one agent or a team is not public (UNVERIFIED). About 5,600 jobs run by agents from three makers, each checked by a different AI, then automatic tests.
L3 Execution and learning 2. Does the work, logs actions, and claims measured gains [site]. No public sign that checked results change the next run, so not 3 (inference). Every failure gets a record and an automatic check (about 160). Offer feedback lines fell from 44 to 18 as rules were added. Only about 15% of jobs reached the main code, and merges were unreliable in October.

Scores rest on marketing pages only; no docs or demo were seen (inference). If the product is deeper than it reads, the scores go up, not down.

5. Same and different

Topic Same or different Them Us
Core belief Same A teammate that learns the firm's data and rules, runs the work, asks a human at the right moment [yc] [site]. Same shape (brief).
Owner of the result Same They do the work and claim a number [site]. The brief's thesis is do the work and own the number.
Buyer size Same Small and mid-sized brands [site]. SMEs and scale-ups (brief).
Kind of work Different Finance and ops tasks in consumer goods: orders, stock, close, AP/AR [site]. W7 is margin leaks for commercial or finance leaders in any SME; no calls yet (brief).
Industry Different One industry: CPG, with 170+ connectors built for it [site]. Not tied to one industry yet (brief).
Country Different US-first: New York, a US trade show, US retailers and platforms [yc] [press]. UK mid-sized firms in view (brief).
How it learns Different "Learns" the data, rules and workflows [yc]. Marton's logs show rules added from failures and feedback, on his own work.
Proof Different Time and money claims, no named case [site]. No customer number yet.
Delivery Different Software with 3 to 4 days setup, one founder-led sales motion (inference) [site]. The system plus the two founders.
Price Different Not public [site]. Not set.

6. Wedge overlap

Wedge Fit Why
W7 Find and fix margin leaks for commercial or finance leaders at SMEs and scale-ups direct Same buyer type and the same job. Their demo shows contribution margin by channel with a 3.1 point drop on Amazon, and they claim $150k+ a year of billing and refund leaks caught [site]. They cover it for consumer brands only.
W5 One layer for one type of small firm adjacent Handled is one vertical (CPG) with an ops layer built for it [site]. It is the proof that a vertical teammate can get YC backing; it also shows the crowded form of W5.

7. Whitespace (where they stop)

  • Buyer gap: CPG only, US first [site] [yc]. Not for services firms, software firms, B2B suppliers outside consumer goods, or UK mid-sized firms. To own it we need 3 to 5 named buyers outside CPG, or in the UK, saying they have the margin-leak pain.
  • Number gap: The money numbers are claims with no named before/after case [site]. To own it we need one workflow run for 4+ weeks with a number a finance person would sign, and the method written down.
  • Layer gap (weaker, inference): No public sign that checked results rewrite rules for the next run, only that it "learns" the business [yc]. To own it we need a log of rule changes tied to checked runs, with rework falling.
  • Price and speed gaps: Not open. The setup claim is 3 to 4 days [site], so a speed claim from us would have to beat that. Price is unknown.

8. For the YC application

  • Proven today: Marton's logs show his system already runs agents that do work and have it checked by a different AI: about 5,600 jobs, with about 160 incident records that each must name an automatic check. It runs on Marton's own business only; no customer has used it.
  • Draft answer to "how are you different from Handled?": DRAFT. "Handled does finance and ops for consumer brands, one industry. We are building the same kind of teammate for margin leaks in any small firm, and our system already runs checked agent work every day on our own business." Founders say it their own way.
  • Future claim (needs proof): "We find the leak, fix it with approvals, and show the number moved, in a UK firm Handled does not serve." Needs one customer workflow run for 4+ weeks with a before/after number, plus 3 to 5 outside-CPG or UK buyers who confirm the pain.
  • What we understand that they don't: needs proof: a margin fix should be measured and fed back into the rules, so the second run is better than the first, and the method should not depend on one industry's connectors.
  • Copyability (M3): medium. Adding a feedback loop on checked results is a few months for a 3-person team with live customers (inference). Leaving CPG would pull them from the connector set that is their edge, and from the buyer they know [site]. Not hard, so M3 can not score above 3 against Handled alone.
  • Why now (M4): None found. Do not use "agents can now do the work": that is general AI progress.
  • Do not claim: that nobody serves margin leaks at SMEs (Handled and CPG finance rivals do, see section 10). That we have a customer or a margin result. That Handled does not learn (UNVERIFIED either way).

9. Ideas worth taking

  1. Rules that set what runs alone and what waits for sign-off, with every action logged [site]. It makes a buyer trust the agent.
  2. A concrete demo number: margin by channel with the drop named (for example Amazon down 3.1 points) [site]. A buyer sees the leak in one screen.
  3. Start from the buyer's own mess (inbox, sheets, odd order formats) rather than clean data [site].

10. Unknowns

  • Funding for Handled after the rename, and whether a co-founder is still active (Peter Nsaka appears in a profile only). Check Crunchbase or PitchBook, and YC's page for edits.
  • Real prices and contract size. Ask a brand that has seen a quote.
  • How much of the product is real versus the marketing page: no docs, demo video or changelog were found. Ask for a demo as a buyer, or find a customer to talk to.
  • Whether they sell outside the US. Look at customer logos (Boxraw, Baobab) for UK or European brands.
  • Nearby rivals in the same CPG niche to check next: Confido (finance operations for CPG, $20M across Series A and seed, Aug 2025) [press] and Jampack AI (wholesale operations for CPG, seed, Feb 2026; amount UNVERIFIED) [press]. They show this vertical is already crowded.

Sources

Technical detail
Slug
handled
Lens hash
179096c479
Pass
deep
Stale
no