XMarts room

Summer 2025

Mimos

The growth OS for plaintiff-side law firms

Visit website

What they do

agents run ads, intake CRM and spend-to-case tracking for injury law firms

How close to us

5, very close

one growth layer sold to one type of small firm, with spend-to-outcome loop; strong W5 match

Wedges

  • W5 One layer for one type of small firm direct
  • W3 Company memory from every call, for consultancies adjacent
  • W7 Find and fix margin leaks adjacent
  • W8 The other 90% of an AI skill adjacent
L1 Company brain (knows the organisation) 1, light
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
plaintiff law firm owner or managing partner; intake or marketing lead; agencies and multi-firm operators as partners
UK
no
Their own description (from YC)

Plaintiff-side law firms compete in some of the most expensive search ad markets to acquire high-intent leads, but the path from click, call, or form fill to signed retainer is still fragmented. Disconnected intake workflows, slow response times, and disconnected spend-to-case data makes it hard to know what is actually working.

Mimos automates their growth and intake playbook:
1. Generate Qualified Leads: Our agents manage Google Ads, Local Services Ads, Google Business Profile, and other high-intent acquisition channels.
2. Convert Faster: Every lead flows into a CRM built for intake, triage, follow-up, and handoff from inquiry to signed retainer.
3. Close the Loop: We connect spend, lead source, intake actions, and signed-case outcomes so firms know what to scale, pause, or fix.

The result: more qualified leads, less wasted spend, and more signed cases.

Full teardown

Mimos (YC Summer 2025): teardown vs XMarts

In one line: Mimos sells supervised AI agents to plaintiff-side (injury) law firms, to answer and follow up on inquiries, tie ad spend to signed cases, and build case timelines [site] [yc]. Threat: 2, because it is the shape of our W5 (one layer for one type of small firm), but it is two people, names no customer, and has changed its product three times in 14 months [yc] [site].

1. Fact file

Item Finding Source
What they say Site, Oct 2026: "Find, sign, and develop better cases." YC line (still live): "The growth OS for plaintiff-side law firms". [site] [yc]
Founded, team, funding Founded 2025, YC S25, New York City, team of 2. Funding: UNVERIFIED. A third-party directory in search results says a $500K seed from YC and XFactor Ventures; we could not open it. [yc]
Founders Rohit Sirosh (CEO): led architecture of Azure's high-throughput storage at Microsoft, worked with OpenAI on training workloads. Michael (CTO; surname not on the YC page): Alexa AI integration, petabyte-scale data at Coinbase. They say they have built together "since sixth grade". What YC likely saw: two technical co-founders with a long working history (inference). [yc]
Customers None named anywhere. The site shows a "Backed by" label with no names. Partner page has no named partners. The only traction claim is from Aug 2025 and is about the old product: "5 paying customers" in a month, in finance, proptech and wealth (they claim). [site] [yc]
Price Not public. No pricing page. No open job ads, so no hint from sales hiring. The careers page asks people to email the CEO. [site] [yc]
Path Aug 2025: answer engine optimization (AEO) for regulated firms (YC launch post, 15 Aug 2025). Nov 2025: "Get a ChatGPT App for Your Business" (build and host apps for gyms, salons, clinics, law firms). Apr 2026: "Adaptive Legal Intelligence", a self-improving system for law firms. Oct 2026: "Find, sign, and develop better cases", supervised agents. The pivot to law is somewhere between Nov 2025 and Apr 2026; the Oct wording appears between Jun and Oct 2026 (from archive captures). [yc] [site]

2. Their proposition, reverse engineered

Part What they do
Who buys Plaintiff-side law firms [site] [yc]. Agencies and "MSOs" (operators of several firms) as channel partners [site].
The pain "Too much value gets lost between the first inquiry and the signed case." Calls, forms, documents, ad spend and outcomes sit in separate systems [site]. Plaintiff firms bid in very expensive search ad markets and cannot tell what spend became a signed case [yc].
The promise Spend goes behind cases that sign; good inquiries get answered and followed up; case facts get gathered. "Optimize for signed cases, not raw leads." [site]
How it works Three steps: the firm defines rules and approval points; agents gather context and prepare work, and only carry out approved steps; the team reviews and approves [site]. Five product areas under Find, Sign, Develop: Growth Optimization, Intake Operations, Revenue Recovery, Case Development, Client Information Gathering [site].
The output Signed-case attribution and campaign recommendations the team approves; answered and routed calls with follow-up drafts; a list of declined inquiries worth a second look, with the reason; a source-linked case timeline and a missing-facts list [site].
Proof style Sample dashboards only: "$42k spend, 31 qualified, 7 retained" and "42 calls reviewed". The site makes no performance claim and no customer claim [site]. Apr 2026 page showed "3x faster qualification" and "22% down since launch" on sample screens; not marked as customer data [site].
Who sells it Founder-led demo calls ("Book a Demo"), plus agency and MSO partners [site]. No sales hires listed [yc].
Price model Not public. Revenue share for partners: not mentioned [site].
The story Enemy: disconnected tools and lost inquiries. Belief: agents do the repeat work, "people keep the judgment calls" [site]. In Apr 2026 the enemy was agencies and CRMs ("Not an agency. Not a CRM. A system that learns.") [site].

3. Positioning

  • Category they claim: "growth OS" (YC) [yc]; "supervised work" for find, sign and develop cases (site) [site].
  • Category a YC partner would put them in: vertical AI for plaintiff law firms: intake and lead-to-retainer ops (inference).
  • What they want to be known for: closing the gap between ad spend and signed cases, with agents the firm can supervise.
  • Where they sit on the map: layers L1/L2/L3 = 1/2/2, buyer = sme, business.

4. Three layers vs ours

Layer Them (score + evidence) Us (from the system inventory, "Marton's logs show")
L1 Company brain 1. Firm-set rules and policies steer the agents, and each case gets a source-linked record [site]. No claim of a lasting firm memory that other work reads. In Apr 2026 they did claim policies that improve from attorney edits; that claim is gone from the Oct 2026 site [site]. A vault of about 1,400 pages, 112 call transcripts and about 8,700 customer quotes, searchable by meaning.
L2 Agent orchestration 2. Several agents across five workflows, with owners, permissions and approval points for messages, campaign changes and decisions [site]. About 5,600 jobs run by agents from three makers, each checked by a different AI, then automatic tests.
L3 Execution and learning 2. The agents do the work (answer calls, draft follow-ups, build timelines) and measure it against signed cases; qualified and retained outcomes go back to Google [site]. No public claim that results change the next run (that was an Apr 2026 claim) [site]. Every failure gets a record and an automatic check (about 160 records). Offer feedback lines fell from 44 to 18 as 22 lint 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 belief Same Agents do the repeat work, people keep judgment, nothing runs without approval [site]. Same shape: Marton's logs show an approval-gated, reviewed work loop.
Learning from feedback Same (their Apr 2026 claim) "Captures feedback from every interaction and proposes improvements, with your approval." [site] Dropped from the Oct 2026 site [site]. Marton's logs show rules added from failures and feedback, approved by Marton.
Buyer Different US plaintiff law firms, one sector [site] [yc]. Target: SMEs and scale-ups, UK (brief). Sector not chosen.
Kind of work Different Ads, intake, follow-up, case facts, for one profession [site]. A general agent team that has run Marton's own business (code, content, offers).
Owner of the result Same The firm approves each action and owns the decision [site]. Same, by design (brief).
How they capture Different Calls, forms, email, Outlook, Google Ads, case files [site]. Calls into a vault, plus jobs and checks (Marton's logs show).
Proof Different Sample screens, no customer number [site]. No customer number either. Marton's logs show internal numbers only.
Delivery Same Software plus founder-led onboarding; "30 min to onboard" was claimed in Apr 2026 [site]. The system plus the two founders.
Size Different Two people, no named customers [yc]. Two founders; the system has run for about 5 months (Marton's logs show).
Price Unknown Not public [site]. Not set yet.

6. Wedge overlap

Wedge Fit Why
W5 One layer for one type of small firm direct This is the pattern itself: one operating layer, sold to one type of small firm (injury law), with a spend-to-outcome loop [site] [yc]. If we pick a firm type, they show YC funded the shape. They do not cover any other firm type.
W7 Find and fix margin leaks adjacent "Signed-case attribution" finds wasted ad spend and recommends what to scale or pause, with approval [site]. Same job (find the leak, approve the fix, measure), different buyer (a law firm marketing owner, not a finance lead).
W3 Company memory from every call adjacent They review call transcripts for declined inquiries and build timelines from calls and email [site]. It is call memory for one case or lead, not a firm-wide memory for a services firm.
W8 The other 90% of an AI skill adjacent Their Apr 2026 page described the upkeep loop: attorneys edit, the system spots patterns, policies update after approval [site]. Gone from the Oct 2026 site [site].

7. Whitespace (where they stop)

  • Layer gap: no firm memory beyond rules and per-case records, and no public claim that results change the next run any more [site]. To own it we need a rule-change log tied to checked runs, with rework falling.
  • Buyer gap: US injury law only. No UK or other sector, no accountants, consultancies or agencies [site] [yc]. To own it we need a UK mid-sized firm of one type who says they have the pain.
  • Number gap: no customer number in public, only sample screens [site]. To own it we need one workflow run for 4+ weeks with a before/after number a firm would sign.
  • Price gap: not found. They publish no price, so we can not say they miss a small buyer (UNVERIFIED).
  • Speed gap: not found. They claimed "30 min to onboard" in Apr 2026 [site]; no proof of time to first value.

8. For the YC application

  • Proven today: Marton's logs show his system already runs agents that do work and check it, across three AI makers: about 5,600 jobs, each checked by a different AI, and about 160 failure records that each added an automatic check. Mimos's public site shows sample screens and names no customer [site].
  • Draft answer to "how are you different from Mimos?": DRAFT. "Mimos sells one workflow to injury law firms in the US. We are picking one firm type in the UK, and our system already learns from checked work every day on our own business." Founders say it their own way. Only true once the firm type is chosen.
  • Future claim (needs proof): "Our agents run one workflow for a UK mid-sized firm and the number moves." Needs one customer workflow run for 4+ weeks with a before/after number.
  • What we understand that they don't: a rule log tied to checked results beats a rules page the firm writes once. needs proof on a customer; Marton's logs show it on his own work (offer feedback lines fell from 44 to 18).
  • Copyability (M3): easy. They already advertised the learning loop in Apr 2026 [site]. It would be a feature for them, it fits their buyer, and they have a two-person team that ships fast (pivoted three times) [site] (inference). They would not chase our UK or non-law buyers, but that is a buyer gap, not a hard edge. So M3 can not score above 3 against Mimos.
  • Why now (M4): None found in this teardown. Do not use "agents can now do the work": that is general AI progress.
  • Do not claim: "no one does supervised agents for small firms" (Mimos does, for law). "Self-improving" as a headline (Mimos used it in Apr 2026 [site] and dropped it). Anything about customers or revenue.

9. Ideas worth taking

  1. Start at the money event, not the activity: they connect spend to a signed case, and send the result back to the ad platform [site]. We should pick our "signed case" (the one number) before we pick the workflow.
  2. A declined-inquiry review: "every decline gets a second look, with the reason" [site]. A simple way to show hidden loss in a first demo.
  3. Define, Run, Review as the three-step pitch: the firm sets rules and approval points first [site]. Easy for a cautious buyer to say yes to.

10. Unknowns

  • Whether any firm pays today. Look for a customer, case study or review after Oct 2026; none found. Ask a plaintiff-firm owner if they have seen a quote.
  • Funding. One directory says $500K from YC and XFactor Ventures; we could not open it. Check Crunchbase or a YC deal note.
  • Why they dropped the "learning loop" wording by Oct 2026: a buyer reaction, or a safer pitch for lawyers (inference only). Look at their demo or ask a buyer.
  • Whether they plan UK or other firm types. Not on the site.
  • Exact date of each pivot. The archive shows captures on 30 Nov 2025, 13 Apr 2026 and 5 Oct 2026 only.

Sources

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