Spitch Agent Assist
Spitch Agent Assist is a real-time AI companion for contact centre agents, delivering next-best-action guidance, instant knowledge retrieval, live coaching, automatic call summaries and compliance prompts inside a unified agent desktop. It is sold to enterprises by quotation.
What is Spitch Agent Assist?
Spitch Agent Assist is the agent-facing module of the Spitch Collaborative Agentic AI Platform, a Swiss suite for contact centres that also includes Virtual Assistant, Speech Analytics, Voice Biometrics, Knowledge Agent, Chat Platform, Agent Training, Quality Management and a Customer Contact AI Hub. Where most of those modules face the customer, Agent Assist faces the person answering the call, and the vendor's stated intent is to augment human talent rather than replace it.
The product page organises the module into five numbered capabilities. Next-Best Action reads the content of the live interaction and pushes step-by-step checklists alongside personalised suggestions on what to say, what to do and which prompt is most likely to solve a problem or sell a solution. Real-Time Knowledge Provisioning proactively recommends the material relevant to the case, cutting the time agents spend hunting for answers. Call Summary produces summaries while the conversation is running, then an automatic write-up afterwards for storage in the CRM or ticketing system. Real-Time Coaching supplies recommended phrases, phrases to avoid and live transcription, aimed at getting new hires up to speed faster and reducing agent attrition. Real-Time Compliance combines the first two so agents stay on script on regulated topics, and redacts personally identifiable information where required.
All of it is surfaced through what Spitch calls a unified agent desktop, a set of on-screen widgets sitting beside the tools the agent already uses. Underneath sits the wider platform: large language models and retrieval-augmented generation for knowledge, conventional NLU and NLP, and sentiment analysis. That platform is deliberately open, able to run Spitch's own models, third-party engines such as OpenAI or Amazon Bedrock, or models built by the customer's own AI team.
Deployment options span cloud SaaS, a Spitch private cloud, the customer's private cloud, on-premises and hybrid. Adapters already exist for Genesys, Microsoft Teams, Luware, Zendesk and Salesforce. Spitch publishes four headline gains for the module — a 20% cut in average handling time, 25% better compliance adherence, 15% higher customer satisfaction and 40% fewer knowledge searches — although no methodology accompanies them. The vendor, SPitch AG of Zurich, employs around one hundred people and claims more than 150 enterprise customers.
What it does
- Suggest the next best action and the words to use, live during the interaction
- Surface the relevant knowledge article without the agent having to search for it
- Summarise the conversation automatically and file it into the CRM or ticketing system
- Coach agents in real time with recommended phrases, phrases to avoid and live transcription
- Hold agents to the script on compliance topics and redact personally identifiable information
- Identify customer intent automatically and route the case to the right person
- Flag cross-sell and up-sell opportunities while the conversation is still open
When to use Spitch Agent Assist / When not to
A quick filter to help you decide if Spitch Agent Assist is the right fit.
When to use Spitch Agent Assist
- Enterprise contact centres in banking, insurance, healthcare, the public sector, utilities, retail, telecoms and transport that already measure average handling time and script compliance
- Customer service directors who need new hires productive quickly and want to slow front-line attrition
- Quality and compliance teams that must evidence script adherence and mask personal data while a conversation is still running
- Organisations bound by data residency rules, since the module can run as SaaS, in a Spitch or customer private cloud, on-premises or hybrid
- Contact centres already running Genesys, Microsoft Teams, Luware, Zendesk or Salesforce, for which adapters already exist
When not to use Spitch Agent Assist
- Small businesses and standalone support teams: everything is sold by quotation through a project cycle staffed with solution architects and deployment engineers
- Buyers who need a published price, since no rate card, bracket or entry-level tier appears anywhere on the site
- Anyone wanting to try before buying, as neither a permanent free plan nor a free trial is advertised and the only entry point is a sales demo
- Freelancers and individual agents, since there is no mobile app, no browser extension and no self-service sign-up
- Developers expecting to integrate unaided, because no public API documentation is published even though public APIs are claimed
How to use Spitch Agent Assist
A typical end-to-end flow, from setup to results.
- Request a product tour or book a demo from the Spitch site, since there is no self-service sign-up
- Scope a first use case with the Spitch team, which typically recommends starting narrow before extending
- Agree commercial terms through a Proposal issued by Spitch and signed by you, or through an Online Order
- Choose a deployment model: cloud SaaS, Spitch private cloud, your own private cloud, on-premises or hybrid
- Connect the platform to your contact centre stack and back-end systems using the existing adapters and connectors
- Decide which language models to run, whether Spitch's own, a third-party engine or your team's in-house models
- Build and tune the assistance scenarios with the low-code visual tools and pre-built components
- Embed the Agent Assist widgets into the agent desktop so prompts, knowledge and transcription appear during calls
- Wire the automatic call summaries into your CRM or ticketing system
- Run and iterate with the assigned success team, using the contractual 9x5 support and its three escalation levels
Pros & Cons
Pros
- Five real-time capabilities are named and described individually rather than promised in the abstract
- Deployment choice is genuinely broad: SaaS, Spitch private cloud, customer private cloud, on-premises or hybrid, which makes data residency negotiable
- Serious privacy governance for a vendor outside the EU: a named Data Protection Officer, a named Article 27 EU representative with contact details, and a Data Processing Agreement built into the contract rather than offered on request
- The platform is open to third-party language models such as OpenAI or Amazon Bedrock, and to the customer's own models, which limits lock-in
- ISO 27001:2022 certification held by the vendor, issued by Swiss auditor Attesta AG on 21 November 2024 and valid for three years
- Ready-made adapters for Genesys, Microsoft Teams, Luware, Zendesk and Salesforce shorten integration work
- An established vendor with around one hundred staff, five offices and named, documented customer references
Cons
- No price is published anywhere: there is no pricing page at all, and the pricing URL redirects to the site's 404 page
- No public API documentation exists, even though the vendor's own FAQ claims public APIs and MCP support
- Neither a permanent free plan nor a free trial is offered, so the only way in is a commercial demo
- No list of processing languages is published, which is a notable gap for a product built on speech recognition
- The headline gains of 20% lower handling time, 25% better compliance, 15% higher satisfaction and 40% fewer knowledge searches come without methodology, scope or period
- Contractual support is 9x5 Central European Time as standard, with no documented 24/7 option
- No documented way for a customer to exclude its data from model training, while the contract lets the vendor exploit usage metadata
Pricing & Plans
No price is published. Spitch operates no public rate card for Agent Assist: the site carries no pricing page, the sitemap contains no pricing URL, and requests to the pricing path are redirected to the 404 page. Neither a permanent free plan nor a free trial is advertised. The product is sold by quotation: under the Master Agreement, charges are set in a Proposal issued by Spitch and signed by the customer, or in an Online Order, with payment due within ten days of the invoice date. The licence covers a number of daily hours specified in that Proposal, and any customisation or additional service is billed separately under a distinct agreement. Spitch describes a phased commercial approach, beginning with a narrow use case at minimal cost before extending. Prospective buyers should therefore expect a negotiated enterprise contract rather than a published subscription.
Data, GDPR & hosting
A consolidated view of how Spitch Agent Assist handles your data.
GDPR overview
Implementation is concrete and unusually well documented for a vendor established outside the European Union. The Data Policy, effective 17 April 2025, names a Data Protection Officer, Dr. Alexey Popov, reachable at info@spitch.ai, and separately designates an EU representative under Article 27 GDPR, Reiner Schmidt, with his own email address and telephone number. A Data Processing Agreement is not an option but one of the documents constituting the Master Agreement, explicitly ranked in its contractual hierarchy. Data subject rights covered include access, portability, erasure and rectification, with a thirty-day statutory response period that may be extended in writing. International transfers rely on the European Commission's standard contractual clauses, a prior risk assessment and additional technical measures. The company pages claim compliance with both the GDPR and the EU AI Act. Note that the policy itself is written mainly from a Swiss data protection standpoint.
Who owns the data?
Under the Master Agreement, SPitch AG keeps sole and exclusive ownership of the Services and of every copyright, trademark and other intellectual property right attached to them, including anything it develops or modifies. What the customer processes through the platform is defined separately as Customer Data. A third category, Corporate Metadata, covers usage statistics such as call identifiers, transaction counts and types, and the agreement expressly allows SPitch AG to exploit that metadata to improve the Services and widen its partner network. Customer Data may be disclosed to Spitch affiliates, or to a reselling partner, provided they carry equivalent confidentiality duties. Personal data is governed by the separate Data Processing Agreement.
Reuse rights
The Master Agreement grants a non-exclusive, non-transferable and non-sublicensable licence to use the ordered Services within the customer's own business, for the agreement period and for the number of daily hours set out in the Proposal or Online Order. It transfers no ownership and no right to resell or sublicense, and Spitch reserves every right not expressly granted. Customers therefore work with their own Customer Data freely inside the platform but cannot reuse the software itself, and any customisation falls under a separate agreement. In the opposite direction, the customer permits Spitch to use its name or logotype as a general reference and to state publicly that it belongs to the Spitch network. Operationally, interaction data drives real-time prompts, knowledge retrieval and call summaries written into the CRM or ticketing system, with personally identifiable information redacted where required. On the website side, the Data Policy enumerates around forty third-party services and six categories of recipients, and permits transfers outside Switzerland, worldwide, under the European Commission's standard contractual clauses.
Data retention & training
Hosting summary
For the standard managed service, the Master Agreement is explicit: the licence covers use of the Services in a Spitch-designed cloud instance in Italy. That is the only hosting country the vendor names. It is not the only option, however. Spitch also offers a Spitch private cloud, the customer's own private cloud, on-premises installation and hybrid arrangements, and states that these exist to align with the customer's security and data governance policies; under those models the hosting jurisdiction becomes the customer's own and should be settled in the contract. Jurisdictionally the picture is mixed: the service agreement is governed by Italian law, while the data controller and the imprint are Swiss and the Data Policy reasons from a Swiss standpoint, allowing transfers outside Switzerland worldwide under the European Commission's standard contractual clauses with a prior risk assessment. The website itself is served behind Cloudflare on an anycast address, which indicates no hosting location. AWS, Microsoft Azure and Google Cloud appear as platform integrations rather than as a declared hosting commitment.
Things to keep in mind
Risks and trade-offs to weigh before adopting Spitch Agent Assist.
- The published gains of 20% lower handling time, 25% better compliance adherence, 15% higher satisfaction and 40% fewer knowledge searches are vendor claims with no methodology, scope or period attached, and should not be used as a business case on their own
- Real-time prompting and scripted compliance can erode agent judgement over time: staff who always follow the suggestion may stop building the product knowledge they need when the system is wrong or unavailable
- Live transcription, call summarisation and voice-based verification amount to continuous monitoring of employees, which carries works council, consultation and morale consequences well beyond the technical rollout
- The public API is claimed but undocumented, so its scope, stability and versioning cannot be assessed before committing; get them in writing
- The contract permits the vendor to exploit usage metadata to improve its services, and no mechanism is documented for excluding customer data from model training
- The list of around forty third-party services published in the Data Policy covers the website, not necessarily the subprocessors behind the product itself, which sit in the non-public Data Processing Agreement
- The Master Agreement is governed by Italian law and designates an Italian cloud instance even though the vendor, the data controller and the imprint are Swiss, so the service contract and the privacy policy do not sit in the same legal order
Setup & Integrations
Technical difficulty
High, and framed as a project rather than a sign-up. There is no self-service route: implementation means integrating with the contact centre stack and back-end systems, and Spitch assigns solution architects, deployment engineers and AI consultants to it. Environments can extend to high-availability Kubernetes clusters with complex third-party integrations. Two things reduce the load: ready-made adapters for Genesys, Microsoft Teams, Luware, Zendesk and Salesforce, and low-code visual tooling with pre-built components once the platform is in place. The vendor publishes no implementation timeline, so scheduling must be agreed directly.
Deployment
Integrations
Behind Spitch Agent Assist
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What exactly is Spitch Agent Assist?
What can it actually do during a call?
How much does it cost?
Is there a free plan or a free trial?
Can it be hosted on our own infrastructure?
Where is the data hosted?
Is there an API?
Which systems does it integrate with?
Who publishes the product, and what compliance guarantees exist?
What support is contractually guaranteed?
Should you pick Spitch Agent Assist?
Spitch Agent Assist is a focused, credible product rather than a broad promise. Its scope is clear: it sits with the human agent, and the five capabilities it advertises — next-best action, real-time knowledge, call summarisation, live coaching and compliance prompting — are each described concretely enough to be evaluated. It is one module of a nine-product platform, which matters at purchase time: much of what makes it attractive, from retrieval-augmented knowledge to voice biometrics, comes from neighbouring modules.
The strongest argument is governance. For a vendor established outside the European Union, the privacy posture is unusually complete: a named Data Protection Officer, a named Article 27 EU representative with published contact details, a Data Processing Agreement written into the contract rather than offered on request, ISO 27001:2022 certification with a dated and time-limited certificate, and a genuine choice between SaaS, private cloud and on-premises deployment. For a regulated contact centre, that combination is the point.
The weakness is commercial opacity. Nothing about the price is public, no free plan or trial exists, no list of processing languages is published for a product built on speech, and no API documentation is available despite the vendor claiming public APIs. The published performance gains carry no methodology. None of this makes the product worse, but it does mean nothing can be assessed without entering a sales conversation.
Treat this as an enterprise project purchase, not a tool to adopt on a whim. It suits organisations that already measure handling time, compliance adherence and agent ramp-up, and that have the procurement patience for a negotiated contract. Ask early for the price basis, the API scope, the processing languages, the product's own subprocessor list and the retention rules in the Data Processing Agreement, since none of the five is published.
- Choosing a selection results in a full page refresh.
- Opens in a new window.