Agentic AI co-pilots, designed for private banking.
We help private banks identify, prototype, deploy and govern AI agents that work alongside their teams — securely, transparently and on-premise when required.
We help private banks identify, prototype, deploy and govern AI agents that work alongside their teams — securely, transparently and on-premise when required.
SpeciCRED 智慧助手套件。
七個生產級智慧體已在我們面向私人銀行的 AI Agents Control Center 中執行 — 在統一控制台中完成配置、監控與審計。即時更新、完整審計軌跡,在關鍵節點保留人在迴路。
套件中的助手數量
7
已生產就緒
套件成功率
95.7%
覆蓋全部智慧體
決策時間節省
−65%
信貸委員會準備
審計軌跡覆蓋率
100%
每個動作均可追溯
- 關鍵
信貸檔案分析助手
對每份信貸檔案進行健康檢查
對每份信貸檔案執行完整健康檢查 — 必備檔案、組合構成、客戶歷史、風險評分與情境分析 — 並向信貸委員會輸出建議的審批傾向。
- 觸發條件
- 新檔案提交檔案更新委員會前複核
- 執行動作
- 核對必備檔案對組合與歷史評分執行情境分析建議審批傾向
96.4% 成功率38 秒 平均 - 關鍵
違約分析助手
流動性分析與行動建議
對每起違約執行流動性分析並說明情況,建議最合適的行動 — 追加保證金、追加資本注入、組合再平衡、風險接受或轉交法律 — 並附上支撐文件。
- 觸發條件
- LTV 違約契約違約流動性閾值觸發
- 執行動作
- 分析流動性建議追加保證金建議資本注入建議再平衡轉交法律
94.2% 成功率28 秒 平均人在迴路 - 高
市場風險分析助手
可即時啟用的壓力測試
在信貸風險官給定主題後,構建結合每位客戶真實資料的壓力測試場景,並以可在 SpeciCRED 中即時啟用的形態交付。
- 觸發條件
- 風險官請求市場狀態切換週期性壓力測試
- 執行動作
- 構建情景注入客戶資料校驗校準交付可執行包
97.1% 成功率90 秒 平均 - 高
條款書分析助手
放貸建議 — 是、否、多少
讀取任意條款書,並對照您的信貸政策、觀察期與現有頭寸進行核對 — 然後給出明確的放貸建議(是 / 否 / 多少)與支撐理由。
- 觸發條件
- 收到新條款書放貸請求再評估週期
- 執行動作
- 解析產品描述核對觀察期核對信貸政策對照現有頭寸建議放貸金額
95.8% 成功率22 秒 平均 - 常規
信貸運營副駕
對簡單檔案進行批次處理
處理 80% 的簡單、非戰略性檔案 — CHF 2M 以下授信、Pool 4 客戶、零售型客戶 — 通過批次分析資料並起草可直接傳送的備註與檔案。
- 觸發條件
- 授信低於 CHF 2MPool 4 客戶零售型客戶
- 執行動作
- 執行批次分析起草備註預填文件進入傳送佇列
92.3% 成功率6 秒 平均 - 常規
報告與儀表盤生成器
基於 Trajan 的按需 SSRS 報告
在 Trajan 聊天機器人的基礎設施之上,按需從全部既有資料生成 SSRS 報告與儀表盤,並交付給分析師與決策者。
- 觸發條件
- 即席請求計劃執行委員會準備
- 執行動作
- 查詢資料生成 SSRS 報告構建儀表盤分發資料包
99.4% 成功率18 秒 平均 - 高
衍生品策略助手
期權策略,雙腿分析
管理 Accu/Deco、TARF、Strangle 等期權策略 — 同時分析兩條腿的結構,並建議增加或減少風險敞口。
- 觸發條件
- 新期權結構策略再平衡風險閾值漂移
- 執行動作
- 拆解雙腿分析 Accu/Deco分析 TARF分析 Strangle建議風險調整
93.5% 成功率14 秒 平均
AI Agents Control Center
在一個控制台內完成每個智慧體的配置、監控與審計。
即時更新、可匯出審計日誌、精細化的 RBAC 與待稽核佇列,讓運營團隊與第二道防線始終保持掌控。
The SpeciVIM Co-pilot Suite.
Eight production-grade compliance agents already running inside our private banking clients' AI Agents Control Center — from KYC onboarding to SAR drafting. Configured, monitored and audited from a single pane of glass, deployed on-premise, with human-in-the-loop where regulation demands it.
Co-pilots in suite
8
Production-ready
Suite success rate
96.7%
Across all agents
False positive reduction
−68%
On AML & sanctions alerts
Audit trail coverage
100%
Every action traceable
- High
KYC Onboarding Co-pilot
Straight-through client onboarding
Ingests onboarding files, extracts identity and beneficial-owner data, and produces a risk-based score ready for compliance officer review.
- Triggers
- New client fileUpdated KYC packRe-onboarding cycle
- Actions
- Extract entitiesScore risk profilePre-fill KYC formRoute to officer
97.4% success32.1s avg - Critical
Source of Wealth Analyzer
On-prem AI for unstructured documents
Securely structures Source of Wealth and Source of Funds evidence — tax returns, sale contracts, statements — with full provenance and citations.
- Triggers
- SoW document uploadPeriodic SoW refreshRM evidence request
- Actions
- Classify documentExtract amounts & datesCross-check narrativeFlag inconsistencies
94.2% success18.7s avgHuman-in-loop - Critical
Sanctions & PEP Screening
Continuous list re-screening
Re-screens the entire client and counterparty book against updated sanctions, PEP and watchlist feeds — with intelligent name-matching to cut false positives.
- Triggers
- List update publishedNew counterparty addedDaily refresh 04:00 UTC
- Actions
- Match candidatesFilter false positivesOpen caseEscalate true hits
99.6% success4.1s avg - High
Adverse Media Watchdog
Real-time negative news monitoring
Continuously scans global media in multiple languages for adverse coverage of clients and connected parties — with relevance scoring and source citations.
- Triggers
- New article ingestedDaily client sweepManual lookup
- Actions
- Score relevanceTranslate snippetAttach to client fileNotify compliance
92.8% success6.3s avg - Critical
AML Transaction Monitor
Behavior-based anomaly detection
Watches transaction patterns against each client's expected profile and flags deviations — outpacing static rule engines with far fewer false positives.
- Triggers
- Transaction postedProfile drift detectedThreshold exceeded
- Actions
- Score anomalyCluster related flowsOpen alertNotify investigator
96.1% success3.2s avg - Normal
Periodic Review Orchestrator
Hands-off review scheduling
Schedules and pre-fills periodic client reviews based on risk class, surfacing changes since the last cycle so officers focus only on what matters.
- Triggers
- Review dueRisk reclassificationMaterial change detected
- Actions
- Pull updated dataPre-fill review fileAssign reviewerTrack completion
98.9% success22.5s avg - High
Regulatory Reporting Compiler
Audit-ready reports in minutes
Compiles MiFID II, LSFin, FATCA / CRS, GoAML and SRD II reports directly from your operational data — with validations and template-aware formatting.
- Triggers
- Reporting period closeRegulator requestData refresh
- Actions
- Aggregate dataRun validationsRender templateFile via channel
100% success240s avg - High
SAR / STR Drafting Assistant
Defensible drafts, with human approval
Drafts Suspicious Activity / Transaction Reports with cited evidence and a recommended narrative. Every output is queued for compliance officer review before filing.
- Triggers
- Investigation case openedEscalation from AML MonitorMLRO request
- Actions
- Draft narrativeCite supporting flowsAttach evidenceRoute to MLRO
94.3% success28.6s avgHuman-in-loop
AI Agents Control Center
Configure, monitor and audit every compliance agent — from a single console.
Live updates, exportable audit log, granular RBAC and a pending-review queue keep your MLRO and your second line of defense in control at all times — with no client data ever leaving the bank.
範圍測算器
為您的 SpeciCRED 人工智慧專案定規模。
告訴我們您的銀行畫像，我們將推薦合適的合作層級、部署方式和需要規劃的治理等級。不展示價格——具體報價由 Talent 團隊在洞察會議後私下提供。
How it works
From discovery to bank-wide rollout, in four steps.
Built around the way banks adopt technology — piloted privately, validated against real metrics, then scaled with the change management Talent consultants are known for.
Step 01
Discovery
An Talent consultant runs a scoping workshop with your business owners — identifying the right co-pilot, the data we will plug into and the success metrics.
Step 02
Pilot
We deploy on-premise inside your environment and connect to your systems through MCP servers (Model Context Protocol) — the same standard used to integrate SpeciCRED, SpeciVIM and your core banking — shipping a working co-pilot in 6–10 weeks under human-in-the-loop validation.
Step 03
Scale
Once measured value is confirmed, we extend across departments, add co-pilots, hand over to your IT and structure long-term governance with the AI Lab.
Step 04
Run
Quarterly model and prompt reviews, regulatory monitoring, FINMA evidence pack and an embedded Talent engagement manager keeping the platform aligned with your strategy.
Outcomes
Measured impact, role by role.
Indicative results observed across our pilot programs in private banks. We measure the same KPIs in your environment before any commitment to scale.
- Compliance officers
−60%
time per Source-of-Wealth file
KYC Co-pilot structures unstructured documents, drafts the SoW narrative and pre-fills SpeciVIM workflows under reviewer validation.
- Credit officers
3×
faster Lombard credit memos
Credit Co-pilot synthesizes positions, collateral and policy on top of SpeciCRED data — drafting memos that an analyst only needs to review.
- Relationship managers
+20%
client meetings prepared in advance
RM Co-pilot pre-briefs every meeting, drafts follow-ups and surfaces commercial opportunities grounded in the bank's data.
- Operations & IT
−40%
L1/L2 ticket triage effort
Operations Co-pilot routes, summarizes and resolves common middle and back-office tickets with full audit trail.
Pricing
Engagement models built for banks.
Pricing is structured around scope, deployment posture and SLA — quoted privately after a discovery workshop with our Talent team. All tiers are denominated in CHF or EUR and include the relevant FINMA evidence pack.
Pilot
Validate value in one workflow.
Banks running a focused proof of value with one business owner.
On request
90-day pilot agreement
- 1 co-pilot, 1 workflow
- Up to 25 active users
- Plugged into SpeciCRED, SpeciVIM or Mobile Banking
- Embedded Talent consultant during the pilot
- Hosted LLM, on-premise inference
- Best-effort support, business hours
Scale
Roll out across a department.
Banks expanding to multiple co-pilots and 100+ users after a successful pilot.
On request
Annual contract, billed yearly
- Up to 3 active co-pilots
- Up to 250 active users
- Custom integrations to your core banking systems
- Dedicated AI Lab squad + Talent engagement manager
- Customer-hosted LLM optional (BYOLLM)
- 99.9% SLA, business hours with on-call
- Quarterly model & prompt review
Enterprise
A bank-wide co-pilot platform.
Tier-1 private banks running co-pilots as a regulated, cross-functional capability.
On request
Multi-year framework agreement
- Unlimited co-pilots and users
- Air-gapped, fully on-premise deployment
- BYOLLM with customer-hosted models
- Co-pilot factory — build your own under our governance
- Dedicated AI Lab squad + named Talent Partner
- 99.95% SLA, 24/7, dedicated TAM
- Comprehensive FINMA & audit evidence pack
Indicative engagement models. Final pricing is tailored after a scoping workshop with our Talent team and aligned with your procurement process.
Compare engagement levels
Built for procurement, audit and IT security.
We support DDQs, RFPs and custom MSAs at every level. Each row below is detailed in our private commercial pack on request.
| Capability | Pilot | Scale | Enterprise |
|---|---|---|---|
| Capacity & scope | |||
| Active co-pilots | 1 | Up to 3 | Unlimited |
| End users | Up to 25 | Up to 250 | Unlimited |
| Use cases covered | 1 workflow | Departmental | Bank-wide |
| Integrations | |||
| SpeciCRED, SpeciVIM, Mobile Banking | Included | Included | Included |
| Core banking (Avaloq, OLYMPIC, Temenos, FNZ, Finnova) | Not included | Included | Included |
| Custom internal systems (CRM, ECM, ERP) | Not included | Up to 2 | Unlimited |
| Bring Your Own LLM (BYOLLM) | Hosted only | Optional | Customer-hosted |
| Deployment & security | |||
| On-premise / private cloud | Included | Included | Included |
| Air-gapped deployment | Not included | On request | Included |
| Data residency control | Standard | Customer | Customer |
| Penetration test cadence | Pilot scope | Annual | Quarterly |
| FINMA & audit evidence pack | Standard | Extended | Comprehensive |
| Service & support | |||
| Talent consultant | Embedded during pilot | Engagement manager | Named Partner |
| AI Lab squad | Shared | Dedicated | Dedicated + named TAM |
| SLA | Best-effort | 99.9% | 99.95% + 24/7 |
| Quarterly review | Not included | Included | Monthly |
| Co-pilot factory | Not included | Not included | Included |
| Commercial | |||
| Term | 90 days | Annual | Multi-year |
| Billing | One-off + success fee | Annual subscription | Multi-year subscription |
| Procurement support | Standard | DDQ + RFP | DDQ + RFP + custom MSA |
Need a side-by-side ROI model for your CFO? We share it under NDA after a discovery call.
Request the commercial pack →Security & compliance
Banking-grade controls, by design.
The SpeciTec AI Lab inherits the security posture of our platform — the same one already trusted by Swiss private banks for credit and compliance.
On-premise by default
Inference runs inside your environment — private cloud, virtual private network or fully air-gapped data centers.
Bring Your Own LLM
Run your chosen models hosted by you. We integrate with Azure OpenAI, AWS Bedrock or fully self-hosted open-weights models.
Full auditability
Every prompt, retrieval and action is logged with the same audit posture as our platform — ready for FINMA, internal audit and external auditors.
RBAC & data zoning
Role-based access aligned to your existing entitlements. Co-pilots only see what the requesting user is already allowed to see.
Data residency
Customer-controlled data residency, with deployments hosted in Switzerland, the EU or your jurisdiction of choice.
Banking-grade SDLC
Penetration testing, SAST/DAST, model evaluation and red-team exercises baked into every release of the AI Lab.
FAQ
Frequently asked questions.
Quick answers to the questions we hear most often. Need more detail? Our team is one message away.
Let's identify the agent that moves your needle.
A 30-minute discovery call to share what we are seeing across the industry and explore where AI can deliver measurable value for your bank.
