XMarts room

Fall 2026

Covera

AI-native OS for insurance brokers

Visit website

What they do

AI operating system for independent insurance agencies, inside their inbox and system

How close to us

5, very close

One layer sold to many small firms of one type (independent P&C agencies), with reviewed agent workflows and logs.

Wedges

  • W5 One layer for one type of small firm direct
  • W3 Company memory from every call, for consultancies adjacent
  • W4 Fit call to tailored proposal in 10 minutes 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) 1, light

Who buys

Firm size
sme
Buyer type
business
Role
agency owner or principal of an independent P&C insurance agency
UK
no
Their own description (from YC)

Covera is the AI-native operating system for independent P&C insurance agencies. It lives inside the agency's AMS and inbox and runs agentic AI workflows across the whole client journey, from the first email to the next renewal. Producers quote first and win more business; CSRs and account managers stop retyping data.

Submissions, for producers: every new request becomes a ready file, one click keys every carrier portal for a first quote during the call, quotes come back side by side, and binding updates the AMS with no double entry.

Servicing, for CSRs and account managers: answers about any account with their source, COIs and follow-ups drafted, renewal reminders 90, 60 and 30 days out with remarketing when premiums jump, and coverage gaps across the book turned into cross-sell leads.

Agencies using Covera go from lead to quote in 5 minutes, save 10 hours a week per user and win 20% more deals against competitors. Every action is reviewed by the agency team and logged for E&O. Works with Applied Epic, AMS360, QQCatalyst, HawkSoft, EZLynx, Momentum, Gmail and Outlook.

Built by the founders of Simio (conversational AI, acquired in 2024).

Full teardown

Covera (YC Fall 2026): teardown vs XMarts

In one line: Covera sells AI workflows that sit inside the inbox and agency system of small independent insurance agencies in the US and do the quoting and servicing paperwork, with a human approving every step [site] [yc]. Threat: 2, because it is the cleanest live example of our W5 shape (one layer, one type of small firm), but it is two people, one year old, and the only named customer is in France [yc] [site].

1. Fact file

Item Finding Source
What they say "Quote first, win more business." YC line: "AI-native OS for insurance brokers". [site] [yc]
Founded, team, funding Founded 2026, YC Fall 2026, 2 people, San Francisco. Funding: a third-party tracker lists $125K, which matches the standard YC cheque (inference). No other round found. [yc] [press]
Founders Tiphaine Venet (CEO): ex-VC at Serena, HEC Paris and Polytechnique. Thibaut Regerat (CTO): co-founder and CTO of Simio. Both built Simio, a WhatsApp commerce AI that was sold in 2024. A third-party listing says the buyer was Alcmeon (inference that it is the same Simio; the listing is thin). What YC likely saw: a repeat team with an exit and a technical co-founder (inference). [yc] [blog] [press]
Customers They claim "Trusted by 50+ independent agencies". No US customer is named. One case study: Hello Risk, a French commercial broker, 155 tasks and about 10 hours a week saved after two weeks (they claim). The YC launch post gives no customer count, revenue or growth. [site] [blog]
Price Not public. Only "Book a demo". The one job ad asks the new hire to build "pricing" from scratch, so a price list probably does not exist yet (inference). [site] [jobs]
Path No rename found. Site has a French version and French-market integrations, but the YC post and the job ad say the US is the focus. [site] [blog] [jobs]

2. Their proposition, reverse engineered

Part What they do
Who buys Principals and operations leaders of independent property and casualty agencies, "of every size" [jobs]. The site lists commercial, personal and independent agencies [site].
The pain Re-typing data into carrier portals and the agency system. Slow first quote. Renewals that slip. Coverage gaps nobody sells into [site] [yc]. No buyer quotes on the site.
The promise "Lead to quote in 5 minutes", "save 10 hours a week per user", "win 20% more deals" (they claim) [yc] [site]. The home page also says "12× ROI" after four weeks (they claim); their own calculator page does not show it [site].
How it works Lives inside the agency system (Applied Epic, AMS360, HawkSoft, EZLynx, Momentum and others) and Gmail or Outlook. Five named workflows: document collection, carrier quoting (prefills portal screens), quote comparison, past-due premium follow-up, account rounding (cross-sell) [site] [yc]. Names and policy numbers are aliased before the AI sees them; data stays in the US on a dedicated instance [site].
The output Ready files, prefilled carrier portals, a side-by-side quote grid, a client proposal, draft emails and reminders, cross-sell leads. Emails are saved as drafts and portals are filled but not submitted [blog] [site].
Proof style Time saved and tasks handled (155 tasks, 10 hours a week). No money result, no named US agency [site].
Who sells it Founders now, plus a first sales hire: Founding GTM / Account Executive, $60K to $100K, builds the playbook from scratch [jobs]. Demo-led.
Price model Not public. Likely per user, because the claims are "per user" (inference).
The story "Automate the transactions, not the relationships." The agent keeps the client; the AI takes the typing [blog]. Human approval on every action, with a log for E&O (errors and omissions) cover [site] [yc].

3. Positioning

  • Category they claim: AI-native operating system for insurance brokers [yc] [site].
  • Category a YC partner would put them in: vertical AI for insurance agencies, back office automation (inference).
  • What they want to be known for: the first quote during the call, and a safe paper trail.
  • Where they sit on the map: L1/L2/L3 = 2/2/1, buyer = sme, business, US.

4. Three layers vs ours

Layer Them (score + evidence) Us (Marton's logs show)
L1 Company brain 2. One sourced risk profile per file, reused by quoting, renewals and cross-sell. "Ask anything about any account", answers carry their source. Reads the agency system and mailbox. No claim that it learns agency habits over time (inference this is borderline 1 to 2) [site] [yc] [docs] A vault of about 1,400 pages, 112 call transcripts and about 8,700 customer quotes, searchable by meaning.
L2 Agent orchestration 2. Five workflows in sequence (intake, quote, compare, bind, renew) with a human approval at each [site] [yc]. No claim of one agent routing others. About 5,600 jobs run by agents from three makers, each checked by a different AI, then automatic tests.
L3 Execution and learning 1. Does the work (fills portals, drafts, updates the system). The check is the agency's human, plus an audit log [site] [blog]. No public claim that results change the next run. 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.

5. Same and different

Topic Same or different Them Us
Core shape Same One layer of AI work sold to many firms of one type [yc] [site]. W5 in the brief is the same shape.
Human in the loop Same Human approves every action and it is logged [site]. Marton's logs show cross-engine review and Marton's approval on decisions.
Buyer size Same Small agencies, owner signs [jobs]. Target SMEs and scale-ups (brief).
Industry Different Insurance only [site]. Not set. Accountancy is the brief's example for W5.
Country Different US focus; data kept in the US [jobs] [site]. UK firms (brief).
Owner of the result Different The agency's staff; Covera saves hours [site]. Aim is the result with a number (brief). Not proven for a customer.
How it improves Different Not public. Benchmarks are static: about 1 minute intake, 2 to quote [blog]. Marton's logs show rules added from failures and feedback.
Proof Different 50+ agencies (claim), one French case study [site]. No customer yet.
Delivery Different Software that plugs into the agency system, then a demo and POC [jobs]. The system plus the two founders.
Price Unknown Not public [site]. Not set.

6. Wedge overlap

Wedge Fit Why
W5 One layer for one type of small firm direct Same problem and buyer type: an operating layer for one kind of small firm, built once, sold to many. Covera does it for insurance agencies [yc] [site].
W4 Fit call to proposal in about 10 minutes adjacent Their quote comparison turns carrier quotes into a client proposal [site]. Different buyer (brokers) and a different input (quotes, not a sales call).
W3 Company memory from every call adjacent "Ask anything about any account" with sources is a small account memory [site]. It reads files and email, not calls (UNVERIFIED that calls are not used).

7. Whitespace (where they stop)

  • Layer gap: no public claim that checked results change the next run [site] [blog]. To own it we need a rule-change log tied to checked runs, with rework falling.
  • Buyer gap: US insurance only; data and pricing built for the US; the one case study is French [site] [jobs]. A UK mid-sized firm in another trade is not their buyer. To own it we need named UK firms saying they have the pain.
  • Number gap: hours and tasks, no money result and no named US agency. "12× ROI" has no shown basis [site]. To own it we need one before/after number a finance person would sign.
  • Price gap: not public, so we can not say. Their sales role pays $60K to $100K, which hints at a small-deal model (inference) [jobs]. Skip this lens until a price is seen.
  • Speed gap: not whitespace. They claim results in two weeks and a quote in 5 minutes [site]. Do not claim speed against them.

8. For the YC application

  • Proven today: Marton's logs show his system runs agents that do work and have it checked by a different AI: about 5,600 jobs, with about 160 failure records that each added an automatic check. We have no customer and no insurance work. Covera has a live product with a claimed 50+ agencies [site] (claim).
  • Draft answer to "how are you different from Covera?": DRAFT. "Covera is the same idea in insurance: one layer for one type of firm. We would build it for a different type of firm in the UK, and our agents check their own work before a person sees it." Founders say it their own way.
  • Future claim (needs proof): "We run one workflow for a UK firm and own the number." Needs one customer workflow run for 4+ weeks with a before/after number.
  • What we understand that they don't: a layer for one kind of firm should learn from checked results, not just save typing time. needs proof on a customer; Marton's logs show it on his own work (offer feedback lines fell from 44 to 18 as rules were added).
  • Copyability (M3): medium. They could add learning from approvals to their product in months, since they already log every action for E&O [site]. But it would not move them out of insurance or the US. The edge we would claim against them is mainly a different industry and country, which they do not chase (inference). Not hard, so M3 can not score above 3 against Covera alone.
  • Why now (M4): None found. Their YC post says YC will speed up US growth, which is not a market change [blog].
  • Do not claim: that we beat Covera in insurance (no work there). Any speed advantage (they claim two weeks and 5 minutes). That their 50+ agencies are false (we only know it is unproven). That we have a customer result.

9. Ideas worth taking

  1. Make every action reviewable and logged. A buyer in a regulated trade pays for the paper trail, and it also makes the buyer trust a first pilot [site].
  2. Plug into the tools the firm already uses (agency system, Gmail, Outlook). No new screen to learn [site] [blog].
  3. Publish free tools and checklists as lead magnets: a time calculator, a quote comparison page, ten how-to posts. They bring the owner in before a demo [site] [blog].

10. Unknowns

  • Whether "50+ agencies" are paying customers or pilots. The YC post gives no count; a two-person team founded in 2026 would be fast to reach 50 (inference). Look at their case studies page again in two weeks and ask in broker forums.
  • Price. Ask an agency owner who has seen a quote.
  • How many US agencies are live. The only named case is French.
  • Whether they plan other trades or the UK. The job ad says US focus.
  • Real funding. $125K is from one third-party tracker.

Sources

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