Run the stablecoin ↔ KES exchange from one cockpit.
Sargo OS shows your team what is stuck, what is slowing settlement, where liquidity is thin, and what to do next. Every action stays human-approved.
Why operators can trust it.
You can judge Sargo OS the same way operators do: funds stay in on-chain escrow, every recommendation carries evidence, every choice is logged, and the system is scored on outcomes.
On-chain escrow
Funds sit in escrow until both counterparties confirm. Sargo OS watches and recommends; it can't move the money.
Every decision is recorded
Accept, override, or defer — each choice is logged with a receipt. The operator can always explain what was done and why.
Measured against outcomes
The system replays decisions against real outcomes to score accuracy, calibration error, and whether it earned more autonomy.
Human controls every action
Sargo OS recommends and explains. The operator decides. That separation stays visible throughout the product.
Review-gated learning
Knowledge doesn't quietly rewrite itself. Updates require satisfying explicit invariant checks and proof.
Honest about every state
When a module is idle or an upstream rail is off, the cockpit says so — instead of hiding it behind missing numbers.
See the queue. Rank the next move. Record the outcome.
Most dashboards stop at reporting. Sargo OS recommends the next action, keeps the operator in control, and measures whether that call helped.
See the live state
Trades, rails, rates, escrow, and market context enter as the operating picture.
Analyze the bottleneck
Pricing, risk, liquidity, and growth checks run together, so the team sees the bottleneck in one place.
Rank the next move
The cockpit turns output into ranked operator actions with evidence and consequence.
Record the decision
Operator accepts, overrides, or defers. Every choice leaves a receipt.
Measure the result
Decisions replay into calibration so the system earns trust instead of claiming it.
What the system measures.
Sargo OS keeps operators focused on throughput, escrow risk, trust, liquidity, and trader activity — then makes the state of each signal explicit.
Weekly Active Traders — distinct traders active in the trailing seven days. One number that says whether the exchange is alive.
Revenue & unit economics
liveTake rate, fees, revenue per trade.
Volume & throughput
liveVolume, queue depth, rail mix.
Escrow value-at-risk
liveMoney stuck in aging or disputed escrow.
Trust & disputes
liveOpen disputes, collusion, concentration.
Retention & engagement
liveGrowth cohorts, new traders, UX friction.
Response-time & SLA
liveCreated-to-completed timing, aging trades.
On every refresh.
Pricing, liquidity, risk, growth, routing, and response-time all execute against the production feed.
Even when idle.
Modules stay on the page even when their upstream stream is empty.
No fake data.
Rails that aren't connected stay explicitly off. Nothing claims a live rail unless it's connected.
One cockpit. Seven views of the job.
The team moves from queue triage to risk, liquidity, growth, and review without bouncing between tools or rebuilding context.
Operate
Keep the exchange running — clear the queue, move money, fix what's stuck.
What matters now
One ranked queue, one brief, one exposure view.
- Ranked action queue
- Escrow exposure
Queue + SLA
Stuck escrow, settlement routing, inline SLA.
- Stuck escrow aging
- Operator SLA tracking
Cases, not flags
Fraud, collusion, and disputes in one queue.
- Velocity & collusion
- Compliance posture
Grow
Expand supply, recruit merchants, and find the next corridor that unlocks volume.
Supply health
Which corridor is thin and what improves fill.
- Corridor gap detection
- Fill-time estimates
Recruit & retain
Trader activity, merchant scoring, rewards.
- Weekly active traders
- Reward effectiveness
Where next
Geographic and rail gaps that block growth.
- Geographic gaps
- Rail expansion targets
Prove
Measure whether the system is getting better and earn automation through outcomes.
Automation readiness
Is the system getting better? Earn trust through outcomes.
- Override rate tracking
- Engine health surfaced
Was the OS right?
Replays decisions against real outcomes.
- Accuracy scoring
- Automation tier gates
Review-gated knowledge
Versioned knowledge that doesn't quietly rewrite itself.
- Entity graph
- Invariant-checked updates
Six levers. One optimizer. Just-in-time liquidity.
Sargo OS doesn't just observe liquidity — it actively shapes it. A reward optimizer watches fill times, gaps, and demand in real time and recommends the smallest incentive that moves supply. Merchants earn across six dimensions, not one.
Speed rewards
Fill an order under 2 minutes, earn a Blitz bonus. Under 5 minutes, a Standard bonus. The fastest UX wins the most.
Volume milestones
Bronze at 500 cUSD. Silver at 2,000. Gold at 10,000. Tiered swaps that reward merchants for scaling with the network.
Surge coverage
When a liquidity gap meets a demand spike, the optimizer activates boosted rewards to pull float back into the corridor — automatically.
Reliability streaks
Ten dispute-free swaps in a row earn a consistency bonus. Clean execution compounds.
Rail coverage
Merchants who offer bank rails (not just M-Pesa) earn a coverage bonus — diversifying settlement beyond mobile money.
Monthly loyalty
Stay active for 30 days, earn a retention bonus. The optimizer watches for declining merchant recency and fires before churn.
The optimizer doesn't fire blindly.
Every recommendation carries an expected effect, an estimated cost in cUSD, and an urgency level. Surge only activates when a real demand spike meets a real gap — not on a hunch. Volume boosts require a trailing baseline, not a single uptick. Operators accept, boost, reduce, or end each program.
Merchant tiers
Every merchant is scored and ranked: Bronze → Silver → Gold → Platinum. Tiers unlock exclusive programs, higher reward ceilings, and priority routing.
Measured effectiveness
Every accepted program is measured before and after. Did it move fill rate? Did it lift volume? Programs that don't earn their cost are flagged for sunset.
Program ledger
Every reward decision is recorded in a durable ledger. Cohort analysis, WAT decomposition, and growth accounting run on the same store — no spreadsheets.
Every surface captured. Every signal extracted.
Sargo OS listens to merchant comms, community channels, and internal discourse — strips PII, classifies sensitivity, and feeds safe signals into the furnace. Dispute risk, support burden, demand signals, and sentiment all become intelligence.
Seven channels, one pipeline
In-platform escrow chat, Telegram merchant ops, Telegram community, Discord team, WhatsApp, Twitter, and the agent-internal bus — all flow through one redaction pipeline.
PII stripped before storage
Phone numbers, emails, wallets, bank details, M-Pesa codes, IDs, and amounts are redacted or bucketed. The furnace only ever sees safe content. Raw stays encrypted at rest.
Eight signal types extracted
Sentiment, support burden, dispute risk, demand signals, competitor mentions, community health, merchant health, team alignment — from raw text to structured intelligence.
Read-only by design
The bot captures messages, it never sends them. Only whitelisted chat IDs are ingested. Telegram user IDs are SHA-256 hashed before storage — not reversible.
Furnace feedback loop
Safe signals become furnace observations, canon entities, and learning reasons. Merchant frustration today becomes a retention recommendation tomorrow.
Proactive merchant outreach
When float drops, the system drafts a top-up nudge. When a tier is earned, a promotion message. When churn risk rises, a comeback bonus. All operator-approved before send.
Capture is scoped and consented.
The Telegram adapter only captures from explicitly configured chat IDs. No wildcard ingestion. Messages expire after 90 days. Every captured message carries its channel, actor (anonymized), and sensitivity classification — auditable end to end.
Before you open the cockpit.
Who is Sargo OS for?
What does it change day to day?
Does it trade or move funds automatically?
Why should I trust the recommendations?
What happens to messages and dispute text?
Where do the funds actually sit?
See the operator view the team uses live.
Inspect the queue, risk, liquidity, and decision surface exactly as the operator sees them.
Open the live cockpit →