◆ The self-audit · Q2 2026 · 34 competitors reviewed

We applied our own methodology to ourselves.

USD 1,500 / 5-day works today. Tomorrow it's structurally squeezed: above by USD 200/month Pro AI seats (ChatGPT Pro, Claude Max, Manus Extended — each USD 2,400/year, 1.6× one Vantage report), below by Singapore senior-analyst day rates of SGD 1,500–3,000. The defensible re-pitch isn't "we research your competitors." It's methodology install + SG-licensed accountability + PSG eligibility. The first cheque is foot-in-door; the moat monetises in the quarterly refresh.

Read the audit → Pricing thesis

34
Competitors reviewed
4
Buckets of threat
3
Real moats (not 9)
$1,500USD start
Foot-in-door, not the price
§ I — The verdict, said plainly

The price is squeezed from both sides.

USD 1,500 sits between the ceiling set by AI-native subscriptions (USD 2,400/year buys "unlimited" Deep Research) and the floor set by Singapore senior analysts (SGD 1,500–3,000/day means a 5-day engagement starts at SGD 7,500). On raw research volume we lose to the subs in 12 months. On primary fieldwork we lose to the boutiques today. The question this audit answers is: what's left, and is it defensible.

Ceiling — AI-native sub
USD 2,400/yr
ChatGPT Pro / Claude Max / Manus Extended each at USD 200/mo. 1.6× one Vantage report buys a year of "unlimited" research.
Floor — SG senior analyst
SGD 1.5–3k/day
A SG-based fractional senior analyst at 5 days runs SGD 7.5–15k. Vantage at USD 1,500 is below the floor — until the senior analyst rate disclosure becomes public.
PSG-effective price
USD 750
If PSG/IMDA pre-approved at 50% subsidy. Below a single ChatGPT Pro annual seat. Structural 2× advantage AI subs cannot replicate in 12 months.
Window before the squeeze closes
~12 mo
Perplexity, Claude and ChatGPT will index regional press by H1 2027. The SEA-source-weighting moat is real now, decaying fast.
§ II — The threats (not the friends)

Every alternative is squeezing us.

The earlier version of this page framed competitors as "wrong for businesses your size." That was the seller's voice. The audit's voice is different: each of these is a real threat to USD 1,500 / 5-day Vantage. Threat scores are from competitors.json → top_five[].

Threat 5/5 · ceiling

AI-native subscriptions

ChatGPT Pro USD 200/mo with 250 Deep Research runs. Claude Max USD 200/mo. Manus Extended USD 200/mo with 40k credits and SG nexus. A literate buyer extracts 70–80% of one Vantage artefact per annual seat — for the cost of 1.6 Vantage reports they get a year of unlimited research.

USD 2,400 / yr each
Threat 5/5 · OSS replication

CrewAI · LangGraph · Claude Skills

Free OSS frameworks plus Anthropic's marketplace (4,200+ Claude Skills, 770+ MCP servers, 2,500+ marketplaces as of Q2 2026). Anyone with one engineer can fork an equivalent template in a weekend. The 9-agent pipeline is not the moat.

~USD 50 in tokens / refresh
Threat 4/5 · same buyer, same week

Asia Insight · Kadence SG

30-year-old MRSS-accredited Singapore boutiques. SGD 5–25k typical project, real local fieldwork. Same procurement category, same approval threshold, same delivery window. They don't need to win on price — they win on "we sat in the room."

SGD 5–25k project
Threat 3/5 · brand premium

Frost & Sullivan custom

USD 15–25k branded report. The CFO signs without questions. Brand recognition does the work that methodology rigor cannot do for a risk-averse mid-market buyer. We don't beat F&S on logo; we have to win on something else entirely.

USD 15–25k
§ III — The four moats that survive

It's not "better LLM output." That moat doesn't exist.

Across the 34-competitor sweep, only four defensibility lines hold up under the squeeze. Each has a falsifier — what kills it if we don't address it.

Moat 01 · regulatory
PSG / IMDA pre-approval
Agentic AI is not on the IMDA pre-approved list as of May 2026. PSG-eligible delivery cuts the SG SME effective price to ~USD 750 — below a single ChatGPT Pro annual seat. Structural 2× advantage that AI subs cannot replicate in 12 months because regulatory pre-approval is process-bound, not feature-bound.
What kills it: letting the IMDA submission slip past Q3 2026, or letting the AI subs land on the list first.
Moat 02 · accountability
SG-licensed Pte Ltd on the cheque
Buyers want a name to email when the analysis is wrong. ChatGPT can't be on the hook. Claude can't either. A licensed Singapore entity, with a UEN, that signs an NDA and answers a phone is the only structure that lets a procurement team buy "AI-amplified research" without their compliance team breaking the deal.
What kills it: outsourcing the analyst signature, ghosting buyers post-delivery, or refusing to sign a rework clause.
Moat 03 · methodology
Reproducibility — buyer owns the rerun
The 600-line Agent 1 prompt, the FIELD-DICTIONARY, the JSON-handoff contract: after one engagement the buyer owns the data shape and can rerun the methodology themselves. SaaS doesn't give that. AI-native subs give black-box outputs you cannot audit. The methodology install is the artefact that compounds quarter-over-quarter.
What kills it: shipping the deliverable without shipping the process. Hand over the JSON, the dictionary, the rerun protocol — or this moat doesn't exist.
Moat 04 · source corpus
SEA-weighted ≥60% HQ quota
Agent 1 enforces a ≥60% SEA-HQ quota on the competitor corpus. AI-native subs are US-news-biased — Perplexity, ChatGPT and Claude index western press first, regional press late. The SEA-source-weighting gap is real in May 2026 and decisive for buyers selling into Singapore, Hong Kong, Malaysia, Thailand, Philippines, Vietnam, Indonesia.
What kills it: Perplexity Max / Claude Research closing the SEA index gap by H1 2027. ~12-month half-life. Monetise the moat now or lose it.
"Stop selling 'we research your competitors.' Start selling the methodology install with SG-licensed quarterly refresh. USD 1,500 is foot-in-door. The moat monetises in the second and third refresh."
— The repositioning, in one line
§ IV — The repositioning

Sell the process, not the deliverable.

The first USD 1,500 isn't the price. It's the cost of getting the buyer to install the methodology. The price card has to reflect that — or every $200/mo AI subscription erodes our deliverable pitch by Q4 2026.

01

The deliverable pitch (legacy)

"Buy 30 competitor profiles + TAM/SAM/SOM + whitespace atlas + design audit for USD 1,500." This is what the previous showcase sold. Audit verdict: ChatGPT Pro at USD 2,400/year produces 70–80% of this output. Position is structurally indefensible by Q4 2026.

02

The methodology pitch (recommended)

"Install the FIELD-DICTIONARY + 9-agent specs + JSON handoff in your stack, with a SG-licensed analyst on signature for accountability and quarterly refresh on retainer." The buyer keeps the rerun protocol. AI subs cannot offer signed accountability. SaaS cannot offer the data shape.

03

The compounding tier

USD 1,500 one-shot is foot-in-door. The recurring price (Quarterly Refresh on the pricing thesis) is where the moat monetises — quarter 2 and quarter 3 are where the buyer realises they've outsourced the rerun protocol they don't want to maintain themselves. Lift the recurring tier price, hold the entry-tier price.

04

The PSG callout, prominently

SG SME effective price USD 750 (post-50% PSG) is below one ChatGPT Pro annual seat. This is the single decisive number on the price card for the SG SME founder persona. Bury it under the table and we lose the structural advantage that makes the rest of the pitch coherent.

Read the audit data,
not just the verdict.

Every claim above is sourced in the working dataset: 34 competitor profiles, 5 personas with NBA arithmetic, the squeeze table by bucket, and the moat falsifiers in full. The admin is the working tool. The PDF is the board pack. Both ship together.

Open the admin → Pricing thesis