Find the dates you're mispricing — before they cost you.
Razor watches forward demand for every date you sell, flags the ones that need a price decision today, and recommends a guarded next room price — with the market evidence attached. It shows its work; you approve every change.
Live demo on your own market · free trial when you're ready · one hotel or a whole group.
No price is ever published automatically. Hotel approval stays final.
Eight things Razor does — and shows.
One workflow, from the first signal to a decision on the record. Every capability below is either shipped, or clearly marked as preview.
Demand read
Forward demand for every stay date, in five bands from Very low to Very high.
Smart Compset
A governed set of your five most comparable competitors — chosen, not typed.
Rate position
Where each mapped room sits against the competitor median, per date and site.
Guarded price
A recommended next room price, clamped so it steps toward the market.
Opportunity flags
Dates separated into upside and rate risk, ranked by estimated value.
Cited Assistant
Ask “why?” and get an answer that names its sources and links back.
AI team
Specialist agents that watch the signals and prepare the work for you.
Daily briefing
The morning read delivered to the channel your team already uses — WhatsApp today.
Revenue teams don't lack dashboards. They lack a shorter list.
You already have a wall of rates and reports. What you don't have is the one thing a busy revenue manager actually needs each morning: which stay dates moved, which of them deserve a price decision today, and the reason to act — before a manual review would have surfaced it.
Razor makes the stay date the unit of work, not the report. It finds where forward demand and competitive rates have shifted, tells you where your mapped rooms sit in that market, and proposes a single guarded next price with the evidence attached. Nothing is hidden behind a score you can't inspect.
Show me the reason before you move the rate.— The one question Razor is built to answer, on every date
One date, four filed sections.
This is how Razor presents a single pricing decision — the signal, the position, the recommendation and the evidence, each one inspectable. Follow the case file below.
Something moved on 21 March.
Forward demand for this date reads Very high — Razor's read across the market it monitors. Beside it sit the signals it observed: search interest up 74% on a normal week, conversion at 10.6% of viewers, and where each competitor is priced.
You're 27% under the market.
Against your four observable competitors, the median for a comparable Standard King is RM 330. You're holding RM 240 — well below the middle, with demand pushing the other way.
A guarded next price — RM 276.
Closing the whole gap in one move would be reckless, so Razor doesn't. The unclamped target is near RM 313; the per-cycle guard caps the move at +15% — so this cycle proposes RM 276, with the rest flagged as upside still available.
The reasoning stays attached.
Every recommendation carries its receipts — observability, source freshness, what was excluded and why, and the window it was read over. No confidence score: a number we couldn't defend would be worse than none.
Demand you can see, not guess.
Razor reads forward demand for every date you sell and lays it out as a calendar you can scan in seconds — five bands, one glance, the dates that deserve a look pushed to the front.
- Observed, not surveyed. The signals beside the read — search interest, conversion, competitor moves — are each shown as their own number, never blended into a score you can't see.
- Context, kept separate. Events and holidays sit beside the read for judgement — explicitly not inputs to it.
- Honest gaps. Too few sessions to read a date? The metric greys out and names the gap instead of inventing one.
Your five closest competitors — chosen, not typed.
A hand-picked competitor list drifts the moment the market moves. Razor governs the set for you: it scores every candidate on comparability and keeps only the ones that genuinely belong.
- Weighted on comparability — location, star rating, price and substitution behaviour, ranked and capped at five.
- Thin, not padded. If fewer than three qualify, Razor tells you the set is thin rather than filling it with the wrong hotels.
- Excludes what can't count. Sold-out, restricted and off-market rooms drop out of the stats, then rejoin with their history intact.
One guarded number, and the guards that made it.
Razor recommends a single next room price — and shows you every rail it had to clear. It moves your position toward the market without lurching, and it stops at the guard rather than pretending it didn't exist.
- Position-preserving. It holds where you sit in your compset and steps with demand, never leaping the whole gap in one cycle.
- Guardrails you can read. A floor, an event-adjusted ceiling, a per-cycle move cap and a ±2% minimum — each one named on the card.
- Reprice is a human action. Behind explicit approval, with an audit trail and rollback. Never automatic — that's a house rule, not a setting.
Illustrative view. The deeper Pricing workspace — what-if and a forward rate plan — is in preview and clearly marked as such inside the product.
Ask “why?” and get an answer that cites itself.
Razor's Assistant sits on top of the same screens, not beside them. Every answer names how many competitors were observable, which source was excluded and why, and links straight back to the view that proves it.
- Continuity of context. The hotel, date and room carry into every reply — nothing is only knowable in chat.
- Sessions and skills. Each thread keeps its own scope; written procedures run against your observed data, including a CSV you upload.
- Capabilities, not colleagues. No fake presence or personas — answers are attributed to Demand, Rates, Pricing and Coverage.
You're at RM 240. The median for the same room and date is RM 330, taken from 4 competitors we could observe — one was excluded as sold out. The per-cycle guard caps this move at RM 276, with further upside flagged.
Illustrative conversation — the live Assistant is shown in a demo.
Specialists that prepare the work — two live, two in preview.
Each agent does one part of the job on your own data. You meet them all in the Assistant. Two are working today; two carry a preview mark because the models behind them aren't built yet, and a preview page that reads as shippable is worse than no page.
Reads every incoming signal — Flow search data, rate feeds, event and holiday calendars, sold-out counts — finds the dates worth a decision, and packages the context. The Scout finds the opportunity and never prices it.
Turns the Scout's findings into a recommendation, with the reasoning attached — a specific rate suggestion, its guardrails and its evidence. This is the voice you meet in the Assistant.
Reads what rivals do, not just what they charge — who undercuts, who holds, the likely cause and a recommended response — delivered as a notification, not another page. The behavioural classifier isn't built yet.
Plans months at a time — seasonal positioning, promotional timing — and drafts the rate card that goes with it. It waits on a PMS connector, so we don't show it working when it can't.
2 of 4 invited to your team today. Which channel each message takes is set in Notifications — there is no fifth “orchestrator” agent.
The morning read, on the channel you already use.
Razor sends the daily briefing to your team's chosen delivery channel — one that's already part of the working day, not another dashboard to remember. You set where it goes in Settings.
- On your channel, not ours. The briefing lands where your team already works — WhatsApp shown here as one example.
- Reply, and it answers back. Where the channel supports it, ask a follow-up in the thread and get the same cited answer — not a link to log in later.
- Bilingual by default. English and Traditional Chinese — the same briefing, the market's own languages.
Example channel — WhatsApp
Example channel · two-way Q&A runs where the channel supports it · nothing is repriced from chat
Built to be questioned.
Five house rules sit under everything Razor says. They can't be switched off by a skill, a setting or a busy morning — they're the reason you can trust the number.
Answer only from observed data
If Razor didn't see it, Razor doesn't claim it. No modelled figures dressed as fact.
Name every exclusion
Sold out, restricted, stale — each one is shown and left out of the maths, never quietly averaged in.
Keep your figures separate from ours
Your rates and our reads never blur together. You always know which number came from where.
Observation, classification, recommendation — kept distinct
What happened, what it means, and what to do are three separate statements, not one blurred verdict.
Never state a revenue outcome, never apply a price automatically
No guaranteed uplift, no projected booking curve, no autopilot. The decision — and the upside — stays yours.
No confidence score
We show sources, exclusions and coverage instead of a number we couldn't defend if you challenged it.
Written for the desk, read across the team.
Razor speaks the vocabulary a revenue manager already uses — competitor median, search interest, opportunity, reprice — so it slots in beside the tools they know, and the briefing carries the same read out to the people who don't open the app.
Revenue managers
Start the day on the short list of dates that moved, inspect the reasoning, and reprice with a guardrail behind every move.
Owners & commercial leads
Stay across every property through the daily briefing — the same evidence, in plain language, without opening the app.
Analysts & front office
An analyst can draft a suggestion but not publish it; front office sees today's numbers, with commercial figures kept separate.
Built for a group — controlled per hotel.
Run one hotel or a whole cluster from one account — how many is set by your plan. Access is two separate dials, so a person's job and the hotels they touch never get tangled.
- Role × scope, kept apart. A role is a set of permissions; a scope is which hotels — all of them, a region, or specific properties.
- Precise switches. View rates vs commercial figures, draft vs publish a rate, edit guardrails or the compset, manage users or connectors — each is its own permission.
- Publishing a rate is always logged. Who changed what, on which property, sits in the audit trail — the owner holds billing and users.
The questions a buyer actually asks.
Does Razor publish prices for me?
No. Every reprice is an explicit human action, behind approval, with an audit trail and rollback. Never applying a price automatically is a house rule, not a toggle — and there is no autopilot to switch on.
Will it guarantee more revenue?
No, and it won't pretend to. Razor states no revenue outcome, no uplift percentage and draws no projected booking curve — there's no model honest enough to draw one from. You get evidence and a guarded recommendation; the judgement is yours.
Do I need to connect my PMS?
No. Razor reads public rates and Flow's own funnel behaviour, so it works without a PMS. That also means it deliberately doesn't show ADR, occupancy, RevPAR or booking pace — figures a PMS would own, that we won't fake.
How is the suggested price kept safe?
It holds your position in the compset and moves with demand, then clamps: a floor, an event-adjusted ceiling, a per-cycle move cap and a ±2% minimum. When a guard binds, Razor shows the capped price and flags the upside — never a silent truncation.
What's live today, and what's preview?
Live: the demand read, Smart Compset, rate position, the guarded recommendation and reprice, the cited Assistant, the Scout and Advisor agents, and the daily briefing (WhatsApp today). Preview, and marked as such: the deeper Pricing workspace (what-if, forward plan) and the Watchdog and Strategist agents. We'd rather show you the seam than blur it.
Can I try it before I commit?
Yes. We start with a live demo on your own market — your competitors, your dates — then a free trial so your team can use it in the daily run before deciding. Plans scale by how many hotels you price; we'll size that with you.
One hotel or a group?
Either. Run a single property or a whole cluster from one account — how many is set by your plan. Each member gets one role and one scope (all hotels, a region, or specific properties), so a cluster manager, a single-hotel revenue manager and the front desk each see and change exactly what they should.
Which stay types and markets?
Overnight stays only — no hourly or flexible types. Launch market is Malaysia, with fixtures in RM and full English and Traditional Chinese.
See it on your own market — then try it free.
We'll run a live demo on your hotels and your dates, read one decision end to end, then set you up with a free trial so your team can use it in the daily run before you commit.
- A live demo on your own competitors and dates — the real screens, no slide deck.
- A free trial afterwards; plans scale by how many hotels you price.
- Straight talk on what's live today and what's still preview.
Noted — thank you.
Your request is with our team — someone from Flow will be in touch to set a time.