Acme Flow

2 of 2 contested points cite retrieved evidence; 6 of 6 sources were used.

What the evidence contests

Claims the deck makes that retrieved evidence pushes back on. A solid left edge means a source you can open; a dashed one means the analysis disagrees but cannot show you why.

Where the deck disagrees with itself

Arithmetic over the deck's own numbers — no outside source involved, so no outside source can be wrong.

Not checkable from what the deck states: price × customers vs revenue (needs a price (ACV/ARPU), a customer count, and a revenue figure, which the deck does not state plainly); LTV/CAC as stated vs computed (needs both an LTV and a CAC as dollar figures, which the deck does not state plainly).

What the deck leaves out

Present in the market evidence, absent from the deck.

What could not be checked

Neither confirmed nor refuted by the evidence retrieved. These are research tasks, not marks against the company — an analysis must not turn its own gaps into a verdict.

What to do next

1. Reframe market slide around the $3-5B serviceable slice
2. Add a slide addressing Power Automate directly
3. Disclose retention and CAC payback
4. Ask the founder: What is net revenue retention on the first 11 customers?
5. Ask the founder: How many deals were competitive against Power Automate, and what happened?
6. Ask the founder: Is inference cost inside the 78% margin?
7. Verify or refute “SAM: $6B (mid-market North America)” — the research found nothing either way.
8. Verify or refute “$1.2M pipeline” — the research found nothing either way.

Summary

Acme Flow is pitching into a market that genuinely exists and is genuinely growing, which is more than can be said for most agentic-workflow decks. The traction is the strongest part of the story: $340k ARR with four consecutive months of 18% growth is specific, recent, and a number an investor can verify quickly.

The market slide is where the deck and the evidence part company. The $47B figure is a 2030 projection from a vendor-sponsored roll-up that bundles legacy RPA seats, iPaaS, and agent platforms into one number. Independent 2026 estimates put the whole category at $18-24B, and the slice a mid-market agentic tool can actually serve at $3-5B. That is roughly a tenfold gap. Notably, the company's own $400M SOM survives this correction intact, which means the overstatement buys them nothing and costs them credibility with anyone who checks.

The more serious omission is competitive. The deck names Zapier and Make and stops there. The market evidence points squarely at Microsoft Power Automate, which arrives bundled inside E5 licenses that this exact buyer already pays for, and at UiPath, which is moving its enterprise base toward agents. An investor who has seen five of these decks this quarter will ask about bundling in the first ten minutes, and the deck has no answer prepared.

What the deck understates is its own wedge. Approval gates are not a feature footnote — they map onto a compliance requirement that mid-market finance ops already have budget for, and reliability is one of the three axes buyers in this category actually decide on. That argument is stronger than the TAM argument the deck leads with.

On balance: the market is real, the traction is real, the framing is inflated, and the competitive picture is incomplete. Two disclosures would resolve most of the uncertainty — net revenue retention on the first cohort, and the outcome of any deal contested against Power Automate.

The advisor's read — judgment, not evidence

Everything above this line is audited against the run's evidence. This section is one analyst's opinion, written after reading it — allowed to reason beyond the record, required to say when it does.

My read: the machine above is honest about a deck that is itself fairly honest — the numbers reconcile, the growth tension is the founders under-promising rather than over-claiming, and the real risk lives in the silences. I suspect the omitted competitors are omitted because the founders have not yet lost a deal to them, which is a seed-stage answer, not a bad one.

What I'd do: take the meeting; price off the plan's implied growth, not the headline.

The bet: governed approvals become the system of record for mid-market automation before the platforms bundle it.

What would change my mind: one lost-to-Power-Automate reference call, or churn in the design-partner cohort.

What this adds up to, for this lens

LEAN NO · confidence: low

Confidence basis: Traction is verifiable; retention and CAC are not disclosed.

A verdict is one reader's reading of the findings above, through one lens. The findings are the durable part; this line is not.

Scorecard

Per-dimension, each with its reasoning. There is deliberately no headline total: a weighted average of seven subjective scores is the one figure here that cannot be traced to a source.

Claim-by-claim audit

Alignment

Risks

Who does what

Annex A — What the market evidence shows

Incumbents

What could not be verified

• No public retention benchmarks for agentic workflow tools.

Annex B — What the deck claims

Market structure

Saturation

Consolidation: Two acquisitions of seed-stage tools by iPaaS incumbents in the last 18 months.

Entrant flow has slowed while acquisitions have picked up, which usually marks the turn from land-grab to consolidation. The compliance-gated mid-market slice is still thinly served.

Is this a product or a feature?

Precedents

What would keep this a standalone market

• Compliance and audit depth that a bundled feature will not reach
• Connector coverage outside the Microsoft estate
• A buyer who deliberately avoids consolidating on one vendor

Open source, and what it predicts

What commercial products still provide once open source arrives

This is what decides the outcome. Parity only matters to the extent that what is left can be cheaply reproduced by a platform vendor that already owns the customer.

Adjacent markets

References

6 sources retrieved and screened: 6 cited, 0 consulted without being cited, 0 dropped by the security screen. Every source is listed, so absence of evidence is as visible as its presence.

Cited in this analysis

Input integrity screen