Workleap
Also known as: GSoft, Officevibe, Workleap AI
Employee-experience platform whose agent works for the manager: six named capabilities that assemble a daily brief, draft performance reviews in the manager's own voice, schedule and prep one-on-ones, and flag engagement issues early — with HR setting the cadences, rubrics and thresholds it works to.
Workleap is a Montreal software company, founded in 2006 as GSoft and renamed Workleap in 2023, whose platform helps managers lead teams they do not have time to lead well. More than twenty thousand companies use it.
The product is organized around six agent capabilities, each doing a different job. Briefing assembles a daily read of what a manager needs to know, do and unblock. Performance Reviews gathers the evidence, drafts in the manager's own voice, refines the draft with them and preps the conversation. Team Health and Feedback watches engagement signals and raises issues before they become problems.
People and Org maintains a living directory that surfaces relationship patterns, roles, expertise and team health. Meetings and 1:1s schedules the meeting, preps the agenda, takes the notes and follows up. Goals drafts objectives, checks how they cascade through the organization, flags the gaps and tracks progress.
What makes those possible is a single layer of people data drawn from engagement, performance, onboarding, learning and compensation, kept current from the systems a company already runs (Workday, BambooHR and ADP among them), with more than a hundred integrations available natively or through custom MCP connections. Underneath, Anthropic's models generate, Microsoft's Azure AI Foundry hosts the inference, and a dedicated ranking model retrieves across the corpus.
HR sets the boundaries instead of leaving the agent to find them. The cadences, frameworks, rubrics, tone and thresholds that govern how it behaves are configured up front from a library of skills, and more work is handed over as trust builds. Performance drafts are refined with the manager before they go anywhere. Managers reach the agent in Slack or Microsoft Teams as well as in Workleap, and a customer's own AI assistant can connect to the platform over the Model Context Protocol.
Workleap publishes a detailed trust center with a data processing addendum covering international transfers, a thirty-two entry sub-processor register naming each processor's role and country, and a separate security portal. Its HR modules are priced publicly per user; the agent is quoted through a demo. It runs as cloud software with no self-hosted option.
Vendor details
Canonical URL
https://workleap.com
Category
Enterprise operations agent
Subcategory
Employee experience — manager-facing agent capabilities
Funding status
Independent, headquartered in Montreal, Quebec, founded in 2006 as GSoft by Simon De Baene (co-founder and CEO) and partners, and rebranded to Workleap in 2023. After seventeen years bootstrapped, the company took a C$125M investment from CDPQ (La Caisse) in 2023 to fund growth and acquisitions, having acquired the learning platform Didacte. Workleap serves more than twenty thousand companies across 100+ countries with a team of roughly four hundred people, and layered agentic AI onto its employee experience suite in 2025 and 2026.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
Workleap states 100+ vendors supported, natively or via custom MCP, with Microsoft Teams, Slack and HubSpot shown as representative connectors. HRIS coverage runs through Merge API, named in the sub-processor register as the unified API tool for provisioning multiple HR information systems, with Workday, BambooHR and ADP among the systems reached; Zoom is registered for virtual meetings and video recording on onboarding. The agent acts through these rather than beside them — the Meetings and 1:1s capability schedules meetings, preps agendas, takes notes and follows up, which writes into the customer's calendar. A Slack integration built in partnership with Slack rolled out to all customers through February 2026, delivering search and context retrieval inside Slack; Workleap describes its MCP-based architecture as laying the foundation for AI-assisted actions there, which is not yet shipped. Inbound, Workleap's terms of service define AI Integration Features that let a customer's own AI tools connect to, query, retrieve data from and transmit data to the products over the Model Context Protocol, reachable from Claude, Gemini and ChatGPT.
In practice
A manager is buried in admin and keeps missing one-on-ones. Workleap's agent spots who needs a conversation, books the meeting, and drafts talking points grounded in the week's engagement and performance data.
Engagement dips in a team and no one notices until it is too late. Workleap flags the risk early and is already planning the next move with the manager before it escalates.
HR wants consistent management without micromanaging every team. Workleap runs on the cadences, rubrics, and thresholds HR configures, delivering personalized manager support at scale.
Sources & related URLs
Agentic Index coverage score
8.5 / 14 capabilities · 61%
| Integrations & Tool Calling | Full |
|---|---|
|
By its own count Workleap has "100+ vendors supported, native or via custom MCP," with Teams, Slack and HubSpot shown as representative connectors. The custom MCP route lets a customer extend the set instead of waiting for Workleap to build a connector. HRIS connections cover Workday, BambooHR and ADP and run through Merge API, Inc., a processor described as a "unified API tool for provisioning multiple human resources information software." Collaboration runs through Slack and Microsoft Teams, productivity through Google Workspace and Microsoft 365, and Zoom handles virtual meetings and video recording for Workleap Onboarding. The agent acts through these systems. Meetings and 1:1s "schedules meetings, preps agendas, takes notes, follows-up," so scheduling writes into the customer's calendar and following up sends into their inbox or Slack. The Slack integration was built in partnership with Slack and has rolled out to all customers. Inside Slack it offers search and context retrieval today, with an MCP based architecture "laying the foundation for future AI-assisted actions within Slack workflows," so write actions there are still to come while calendar scheduling and HRIS provisioning work now. Workleap also runs its own inbound MCP server, and the people data the connectors deliver feeds the AI stack behind the agent (Anthropic, Cohere and Azure AI Foundry). Sourceworkleap.com integrations navigation entry, the sub-processor register and the Slack agent announcementread 2026-09-12 |
|
| Workflow Orchestration | Full |
|
Each of Workleap's six agent capabilities runs as a sequence of steps. Meetings and 1:1s "schedules meetings, preps agendas, takes notes, follows-up," four ordered steps that span a calendar, a document and a follow up action with the meeting itself in the middle, so the process holds state across a gap it does not control. Performance Reviews puts a person in the loop by design. It "gathers the evidence, drafts in your voice, refines with the manager, and preps the conversation," so the sequence pauses on a human turn and then resumes. Goals "drafts goals, checks the cascade, flags gaps, tracks progress," which means validating each goal against the org hierarchy above it and keeping a long running commitment open. The product acts on its own initiative instead of waiting to be invoked. Briefing delivers "your daily read on what to know, do, and unblock," composed across the other capabilities, and Team Health and Feedback "spots engagement issues early, before they become problems." Four kinds of participant take part, namely the manager the agent works for, the HR administrator who sets the cadences and rubrics that bound it, the employee on the other side of a one on one or a review, and the connected systems the agent reads and writes. All the sequences draw on one shared context layer, which is why they compose, so a review draft can draw on engagement signals and a one on one agenda on what the review surfaced. Sourceworkleap.com agent capability descriptionsread 2026-09-12 |
|
| Knowledge Grounding & RAG | Full |
|
The agent grounds in the customer's own people data, centralized and kept current. Workleap unifies engagement, performance, onboarding, learning, skills and compensation data into one layer that gives the agent the organization's context. People and Org is "a living directory that surfaces relationship patterns, roles, expertise, and team health," refreshed from connected systems instead of snapshotted for each run. Cohere Inc. serves as a "search result ranking tool" for Workleap AI and AI Features, processing all categories of personal information a customer may disclose, "including data provided through third-party integrations," so a dedicated ranking model sits over the corpus. Unstructured Technologies, Inc. parses documents uploaded through the services, giving an ingestion path, and Rockset runs as a real time analytics database with MongoDB and Snowflake underneath. Ranking, parsing and a real time index together make up a RAG stack. HRIS connectors via Merge bring worker records from Workday, BambooHR and ADP into the same layer, so the agent reasons over the customer's systems of record, not only what Workleap holds. The grounding shows in the output, since Performance Reviews "gathers the evidence" before drafting and Meetings and 1:1s preps agendas from what has been tracked. Anthropic's model, hosted for inference on Azure AI Foundry, works over this corpus. Sourceworkleap.com/trust-center/workleap-sub-processors and the agent capability descriptionsread 2026-09-12 |
|
| Human Oversight & Guardrails | Full |
|
Oversight works at two levels. Before the agent does anything, HR calibrates the cadences, frameworks, rubrics, tone and thresholds that define how it behaves, drawing on a growing library of skills or writing custom ones, and hands over more work as trust builds. The customer decides what the agent may do and how, and can widen that scope deliberately. Tone and rubric settings govern how the output reads as well as whether the agent acts, which is unusual. At run time the gate sits inside the capability. Performance Reviews "gathers the evidence, drafts in your voice, refines with the manager, and preps the conversation," so refining with the manager is a step in the sequence and a draft never leaves the agent without a human turn on it. A performance review is the most consequential document the product produces. Meetings and 1:1s has the same shape, with the agent preparing the agenda and taking the notes while the manager runs the conversation. The agent's outputs are briefs, drafts, agendas, flags and nudges directed at a manager, who then acts. None of the six capabilities completes an irreversible action against an employee on its own, so there is no automated review submission, compensation change or message to an employee without a manager in between. On the Slack path the MCP architecture would let Workleap AI "act on a user's behalf when explicitly authorized," but that action set is still to come. Workleap describes no reviewer queue, approval object, escalation path or audit of overrides. Sourceworkleap.com performance and meetings capability descriptions, with the Slack agent announcementread 2026-09-12 |
|
| Security, Identity & Governance | Partial |
|
Governance is well documented, but no SOC 2, ISO or other certification is claimed on Workleap's main site. A Security Trust Center at security.workleap.com, hosted on Drata and linked from the main trust center, covers "key organizational and technical measures to safeguard customer data," and nothing from Workleap backs a SOC 2 claim. The main trust center carries Terms of Service, a Privacy Policy, a Data Processing Addendum that explicitly covers international transfers, a full processor register, a modern slavery statement, separate Barley Inc. agreements following that acquisition, a status page and a security@ reporting address, and it names a Data Protection Officer. The register gives 32 processors with the nature of processing, categories of personal information, countries, transfer mechanism and product scope for each, and marks Standard Contractual Clauses and EU-US Data Privacy Framework self certification per entry. FusionAuth appears there as a "security access management tool" handling usernames, passwords and email, stored and processed in Canada, which points to the identity layer. Workleap describes no single sign on, SCIM or role based access control for customers. Sourceworkleap.com/trust-center and /trust-center/workleap-sub-processorsread 2026-09-12 |
|
| Observability & Auditability | Partial |
|
The product shows what the agent produced, not what it did. Briefing delivers a daily read a manager can inspect, Performance Reviews "gathers the evidence" so the evidence is attached to the draft, Team Health surfaces the engagement signals behind a flag, and Goals "flags gaps" and "tracks progress" against a visible cascade. Real time engagement and performance dashboards run underneath, with Rockset serving as a "real-time analytics database service." A manager can see the reasoning attached to an output and the data it came from. There is no run log, action history, execution trace, export or retention policy for agent activity. For an agent that schedules meetings, drafts reviews in a manager's voice and nudges on engagement signals, there is no way to reconstruct which actions it took, on whose behalf, and when. Workleap's own monitoring is a separate matter. Datadog handles "cloud infrastructure logging, analysis and monitoring," processing logs, traces, error reports and session replay data, and Sentry handles "application performance monitoring and error tracking," both for Workleap Compensation, so those tools watch Workleap's own systems. Sourceworkleap.com/trust-center, the sub-processor register and the agent capability descriptionsread 2026-09-12 |
|
| Memory & State Persistence | Not documented |
|
What Workleap offers is learning and a corpus, not agent memory. An insight layer continuously teaches the agent about business context, but there is no distinct persistent memory store, and context folded into how the agent behaves is not state it keeps and can retrieve. Nothing says whether what it learns is scoped, inspectable, editable or tied to a run. The insight layer itself is the centralized people data across engagement, performance and onboarding, which the agent queries. MongoDB and Snowflake hold the data, Rockset runs real time analytics, Cohere ranks search results and Unstructured parses files, which together make a retrieval stack. People and Org is "a living directory that surfaces relationship patterns, roles, expertise, and team health," kept current from the customer's own connected systems. Performance Reviews comes closest, since it "gathers the evidence, drafts in your voice, refines with the manager," and drafting in a specific manager's voice suggests something retained about that manager across reviews. Workleap does not say whether that voice is learned and stored, inferred on each run from earlier writing, or configured. There is no named memory component, no retained preference or correction applied to the next cycle, no decision trace between runs, no stated scope or lifetime, and no way to inspect, edit, export or purge agent state apart from the people records. Sourceworkleap.com/trust-center sub-processor register and the agent capability descriptionsread 2026-09-12 |
|
| Deployment & Data Residency | Not documented |
|
Where data is processed is disclosed in unusual detail, but the customer gets no choice over it. Microsoft Corporation provides general cloud services and AI inference through Azure AI Foundry for all products and Workleap AI, with processing in eighteen countries, which are Australia, Brazil, Canada, the United States, France, Germany, Italy, Japan, South Korea, Norway, Poland, South Africa, India, Spain, Sweden, Switzerland, the United Arab Emirates and the United Kingdom. Anthropic processes in the USA, and MongoDB, AWS, Snowflake, Merge, FusionAuth and Rockset in the USA and Canada. The list tells a customer where their data could end up without letting them choose. Each processor carries its transfer mechanism, with Standard Contractual Clauses throughout, an adequacy decision for Canada on Cohere, and EU-US Data Privacy Framework self certification on most entries. A customer with an EU localization obligation cannot get EU only processing, and there is no region selection, customer environment, single tenant option or self hosting. The disclosure itself is thorough, with 32 processors each given their nature of processing, data categories, countries, transfer mechanism and product scope, alongside a DPA that explicitly covers international transfers. Sourceworkleap.com/trust-center/workleap-sub-processors and /trust-centerread 2026-09-12 |
|
| Prebuilt Agents, Templates & Packs | Full |
|
Six named agent capabilities, each with its own page, now make up the product instead of a single Manager Agent, and each does a distinct job. Briefing is "your daily read on what to know, do, and unblock." Performance Reviews "gathers the evidence, drafts in your voice, refines with the manager, and preps the conversation." Team Health and Feedback "spots engagement issues early, before they become problems." People and Org is "a living directory that surfaces relationship patterns, roles, expertise, and team health." Meetings and 1:1s "schedules meetings, preps agendas, takes notes, follows-up." Goals "drafts goals, checks the cascade, flags gaps, tracks progress." Each stands on its own, serving one customer at different moments and on different objects. Two more libraries sit underneath. HR calibrates agent behavior "using a growing library of skills or custom ones," a set Workleap supplies for the customer to choose from, with authoring as the fallback. A separate template library at workleap.com/template-categories is organized by category and reachable from the main navigation. Prebuilt review cycles, question banks and engagement frameworks ship with the performance and Officevibe modules, so a customer can run a cycle in week one. There is no third party or community contribution path and no marketplace of outside agents, so everything in the catalog comes from Workleap. Sourceworkleap.com navigation agent capability set, template library entry and skills calibration descriptionread 2026-09-12 |
|
| Triggers & Channel Coverage | Full |
|
Workleap is proactive by design and builds a whole capability around a schedule. Briefing, "your daily read on what to know, do, and unblock," is a morning brief that assembles itself and arrives without being asked for, a scheduled run with a composed output instead of a report a manager opens. Condition based triggers fire on the data instead of the clock. Team Health and Feedback "spots engagement issues early, before they become problems," and Goals "checks the cascade, flags gaps, tracks progress," with the agent watching the people data and raising something when a threshold HR calibrates is crossed. Calendar and lifecycle events drive a third kind. Meetings and 1:1s schedules the meeting, preps the agenda ahead of it, takes notes during and follows up after, three actions hung off one dated event, and review cycles run on cadences HR sets. Slack and Microsoft Teams are both connectors, and the Slack integration, built in partnership with Slack and rolled out to all customers, delivers search and context retrieval inside Slack. Email notifications go through Courier and Mailchimp, and there is an in app surface plus the MCP path from Claude, Gemini or ChatGPT, so a manager meets the agent in the tool they already use. Workleap says the Slack MCP architecture "lays the foundation for future AI-assisted actions within Slack workflows", so Slack delivers and retrieves today but does not yet act. Events, schedules and multiple channels are all generally available. Sourceworkleap.com briefing and team health capability descriptions, with the Slack agent announcementread 2026-09-12 |
|
| Model Flexibility & Routing | Partial |
|
Several models run behind the agent and the stack is disclosed, but Workleap makes the choice, not the customer. Anthropic PBC provides the large language model for Workleap AI and AI Features, processing in the USA. Microsoft Corporation provides general cloud services (data hosting) and hosts and processes data for AI inference through Microsoft Azure AI Foundry, across all products and Workleap AI. Cohere Inc. ranks search results and Unstructured Technologies parses files, both for Workleap AI and AI Features. So Anthropic generates, Azure AI Foundry hosts the inference, and Cohere ranks. There is no model picker, admin setting, assignment per capability or bring your own key path, and nothing lets a customer substitute a provider. The MCP server works the other way round. Workleap's terms define AI Integration Features as letting a customer's third party AI tools "connect to, query, retrieve data from, transmit data to, or otherwise interact with the products" through the Model Context Protocol, which is the customer's assistant calling Workleap. Claude, Gemini and ChatGPT are named as clients that can connect, not as models that run the agent, and Workleap offers no way to connect a customer's preferred LLM. Sourceworkleap.com/trust-center/workleap-sub-processors and /trust-center/termsread 2026-09-12 |
|
| APIs, SDKs & MCP Extensibility | Full |
|
An MCP server is written into Workleap's terms of service, which define AI Integration Features as "the functionality within the products that enables customer's third-party AI tools to connect to, query, retrieve data from, transmit data to, or otherwise interact with the products through standardized integration protocols, including the Model Context Protocol (MCP) and any successor or alternative protocols supported by Workleap." It reads and writes, since "transmit data to" and "otherwise interact with" sit alongside query and retrieve, so an outside caller can put data into Workleap as well as read from it. The named clients are Claude, Gemini and ChatGPT. The Slack launch describes the MCP based architecture as laying the foundation for AI assisted actions in Slack workflows, "enabling Workleap AI to act on a user's behalf when explicitly authorized," which is still to come. MCP also works in the other direction. Workleap lists "100+ vendors supported, native or via custom MCP," so a customer can extend the connector set through MCP instead of waiting for a native build, an extension surface separate from the inbound server. Workleap describes APIs and webhooks, and Merge API, the "unified API tool for provisioning multiple human resources information software," powers the HRIS connectors. There is no developer portal, endpoint reference, SDK or MCP tool list for a builder. Sourceworkleap.com/trust-center/terms, with the integrations description and the Slack agent announcementread 2026-09-12 |
|
| Testing, Debugging & Optimization | Not documented |
|
Nothing puts a change to the agent under test. There is no test environment, sandbox, dry run, holdout, A/B comparison, versioning of agent configuration, evaluator or scored verdict on agent behavior. An HR admin can retune a rubric or threshold and it governs the next cycle, but nothing runs that change against known cases first, and no result compares before with after. Workleap's benchmarks, performance summaries and analytics measure the customer's workforce as assessed by its managers (engagement scores and trends, team health, review completion and individual performance), not the agent. The one check on agent output is the manager refining a draft before it goes out. Datadog for infrastructure logging and traces and Sentry for application performance and error tracking cover Workleap Compensation and watch Workleap's own systems. Sourceworkleap.com/trust-center, sub-processor register and agent capability descriptionsread 2026-09-12 |
|
| Browser & Computer Use | Not documented |
|
Every path in and out of Workleap is a protocol, not a screen. It reaches other systems through named connectors, with HRIS via Merge API, Slack and Microsoft Teams through their own platform integrations, calendars for meeting scheduling, and Zoom for onboarding video. It extends that set "native or via custom MCP," and its own inbound surface is an MCP server that works through "standardized integration protocols" under the terms of service. The agent reads people data from a central layer and writes calendar events, drafts and notifications through APIs, so a third party redesigning its interface does not touch it. There is no hosted or local browser, desktop session, remote or local computer control, RPA or recorder. A browser automation or virtual desktop component would have to appear among Workleap's processors, and none does, since the 32 entries cover cloud hosting, databases, analytics, model providers, notification and support tooling. Meetings and 1:1s "takes notes," which on some products means joining a call as a participant bot, but nothing says that here. Zoom is scoped to Workleap Onboarding for video recording, not to the meetings capability, and taking notes from a transcript or recording is document processing, not operating an interface. Sourceworkleap.com/trust-center, the sub-processor register, /trust-center/terms and the agent capability descriptionsread 2026-09-12 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Pricing
Standard starts at $4,999 a year and Pro at $11,999 a year, billed annually. Enterprise is custom for organizations with more than 250 employees.
Annual plans with published starting prices. Standard starts at $4,999 and Pro at $11,999, billed annually, and nothing published says how the price grows from that starting point. Enterprise is custom for organizations with more than 250 employees. The AI features are part of the plans rather than quoted separately.
What is public
Plan pricing is public. Standard starts at $4,999 and Pro at $11,999, and Enterprise is custom for organizations with more than 250 employees, through Workleap's team. The page reads "Plans are billed annually, pricing applicable for new customers." Standard includes end to end performance management, team health and engagement reports, AI analyses, summaries and chat, an org chart and people directory, goals, recognition, an integrations catalog and SSO. Pro adds Advanced Intelligence, MCP access, SAML SSO, advanced security, API access, custom properties and segments, and personalized onboarding. No free tier, trial or setup fee is published.
Billing mechanics
Plan subscriptions billed annually, with Standard and Pro at published starting prices and Enterprise priced on request. The AI features are part of the plans, and Pro adds Advanced Intelligence and MCP access.
Cost watchouts
Standard and Pro are starting prices, so the actual annual cost above them is quoted. Advanced Intelligence, MCP access, API access and SAML SSO sit in Pro, so needing any of them moves the bill from $4,999 to at least $11,999.
Variable cost rationale
Workleap sells annual plans, with Standard starting at $4,999 and Pro at $11,999 a year. No usage, credit or per action meter is published for the AI features, which come inside the plans. A starting price means the bill grows with something, most likely headcount, but no per user rate or employee range is published. Annual billing fixes the cost for the term. Organizations with more than 250 employees move to custom Enterprise pricing.
Additional watchouts
The published figures are starting prices, not list prices, and nothing says what moves a quote above them. Organizations with more than 250 employees get Enterprise pricing with no published figure.
Sales call required
Mixed (some tiers require a call)
Free / trial
None published
Lowest paid plan
Standard, starting at $4,999 a year, billed annually
Key ambiguities
Workleap sells three plans. Standard starts at $4,999 and Pro at $11,999, billed annually, and Enterprise is custom for organizations with more than 250 employees. What a starting price covers, and how it grows with headcount, is not published. The product leads with six named agent capabilities (Briefing, Performance Reviews, Team Health and Feedback, People and Org, Meetings and 1:1s, Goals), and the module names Officevibe and the LMS appear as use cases in the navigation. AI analyses, summaries and chat are part of Standard, and Pro adds Advanced Intelligence and MCP access, so the AI capability is priced inside the plans rather than behind a separate quote. The Barley acquisition added Compensation as a product with its own use case entry, service agreement, privacy policy and subprocessor list in the trust center. No price for it is published.
Missing data
What a starting price covers, and how the bill grows above it, whether by employee count or something else, is not published for Standard or Pro, and Enterprise pricing above 250 employees is custom. Also unpublished are which of the six agent capabilities each plan includes, the price of the Compensation product following the Barley acquisition, any usage or credit component attached to agent activity, and contract length and cancellation terms.
Related vendors
- 11th Estate — Agentic platform whose AI engine scans markets, matches an…
- Abacum — FP&A platform for growing companies whose AI Modeler builds, edits…
- Adopt AI — AI native accounting and bookkeeping firm whose agents run…
- Aera Technology — Decision intelligence platform where an always on agent executes…
- Akro AI — On premise operational intelligence that automates document heavy…
- alfred_ — AI executive assistant that triages the inbox overnight and scores…
Alternatives to Workleap
The closest documented capability profiles to Workleap among enterprise operations agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Ironclad9.5 / 14Adds documented Memory & State Persistence
- Zip9.5 / 14Fuller documented coverage on Security, Identity & Governance and Observability & Auditability
- Alloy.ai8.0 / 14Fuller documented coverage on Security, Identity & Governance
- Campfire9.0 / 14Fuller documented coverage on Security, Identity & Governance and Observability & Auditability
- Deel8.0 / 14Fuller documented coverage on Security, Identity & Governance
- GoComet7.0 / 14A lighter documented profile than Workleap
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded