Pinnacle
Also known as: Pinnacle Pascal, heypinnacle.com
AI coach for managers that works inside Slack, Teams and Zoom: attends meetings with permission, builds a knowledge graph of the team, intervenes proactively when it spots a pattern, and routes sensitive topics to HR under admin-set escalation triggers.
Pinnacle sells AI coaching for managers through Pascal, a coach that works inside Slack, Teams and Zoom rather than in a separate portal. Its argument is that standalone coaching platforms fail because they demand behavior change. Managers must remember to log in, book time and explain their situation from scratch. Pascal instead sits where the work already happens and starts conversations itself.
The coach is proactive rather than passive. It attends meetings with permission across Zoom, Teams and Google Meet, observes team dynamics, and follows up afterward about what it noticed. It watches for patterns over time and raises them unprompted, and it times guidance to what is coming up rather than waiting to be asked. It is available around the clock by text or voice.
What separates it from a chatbot is the context it holds. Pascal builds a knowledge graph of a team, mapping relationships between people and tracking communication patterns, and during onboarding it takes on a company's values, competencies and culture so its guidance reflects how that organization actually works. When a manager asks for advice about giving feedback, it draws on earlier conversations with that person, the company's own feedback framework and patterns it has observed. The coaching methods behind it were built by ICF-certified coaches as structured reflection prompts, goal-setting frameworks and skill development paths.
Guardrails are configurable and topic-aware. When a conversation touches mental health, harassment or legal matters, Pascal recognizes the subject and routes it to HR rather than offering AI guidance, with administrators setting the escalation triggers. Individual conversations stay confidential, with no HR visibility into what a manager discusses. Organizations receive anonymized aggregate insight instead.
The company publishes a SOC2 and GDPR compliance posture, with SOC2 Type II maintained through independently audited annual third-party reviews, alongside single sign-on, multi-factor authentication, encryption, configurable data retention, regular penetration testing and data residency controls for regulated industries. Customer data is never used to train the models and conversations are isolated between organizations.
Pinnacle is based in Palo Alto and names HubSpot, Ripple, Verkada, Springboard, Semgrep, FLYR, Overjet and BTS among the teams using it, with the product positioned at mid-market organizations whose managers work in Slack or Teams daily. It is sold through a demo-led motion, at a published $150 to $300 per manager per year.
Vendor details
Canonical URL
https://www.heypinnacle.com
Category
Enterprise operations agent
Funding status
Private, early stage. Palo Alto based, backed by Gaingels and the Alchemist Accelerator. No round figures are published.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Integrations
Pascal embeds directly into Slack, Microsoft Teams and Zoom as text and voice coaching surfaces, and with permission it joins Zoom, Teams and Google Meet sessions as a participant. It draws organizational context from HRIS systems and calendar context for review cycles and goal setting seasons. Pinnacle names no broader catalog of tool integrations and no way for a customer to add one.
In practice
Your managers leave a tense 1:1 with no chance to reflect on it. Pascal joins the meeting with permission and messages the manager afterward to offer a debrief.
A manager's 1:1s with one report keep running over, week after week. Pascal spots the pattern and messages the manager without waiting to be asked.
Your HR team wants sensitive topics kept away from AI advice. Pascal recognizes mental health, harassment and legal issues and routes them to HR, with administrators setting the escalation triggers.
Sources & related URLs
Related / legacy domains
Agentic Index coverage score
5.0 / 14 capabilities · 36%
| Integrations & Tool Calling | Partial |
|---|---|
|
Pascal works inside Slack, Microsoft Teams and Zoom and also joins Google Meet sessions. It draws organizational context from HRIS systems and calendar context for review cycles and goal setting seasons. With permission, Pascal joins Zoom, Teams or Google Meet sessions as a participant, and it monitors Slack and Teams conversations. Pascal reads meetings and messages and posts coaching guidance back into the same channel. Pinnacle names no write to a system of record, such as an HRIS update, a calendar entry or a ticket. The list of integrations is short and fixed. Pinnacle names no custom tool support, connector directory or way for a customer to add a system, and it names no specific HR system. Sourceheypinnacle.com homepage Plugged-in card, heypinnacle.com/blog how-pinnacle-handles-coaching-data data sourcesread 2026-09-14 |
|
| Workflow Orchestration | Not documented |
|
One agent, Pascal, does all the work. There is no second agent, so there are no specialized agents, roles or handoffs between them. Pinnacle names no orchestrator, runtime, task graph or BPMN, and no workflow builder or versioning. Each piece of work is a conversational exchange. Structured reflection prompts, goal setting frameworks and skill development paths, built by ICF certified coaches, are coaching content. Pascal delivers them in conversation, as the methods a human coach would follow. The one branching flow is a routing rule. On sensitive topics, Pascal recognizes the subject and routes it to HR instead of answering. Pinnacle positions the product around staying inside the channels managers already use, not around working across systems. Sourceheypinnacle.com homepage use-cases and Why teams choose Pascal, heypinnacle.com/blog best-ai-coaching-tools coaching frameworksread 2026-09-14 |
|
| Knowledge Grounding & RAG | Full |
|
A knowledge graph of the team sits behind Pascal's coaching. It maps relationships between team members, tracks communication patterns over time and holds the organization's context. When a manager asks for feedback advice, Pascal draws on the manager's previous conversations with that person, the company's feedback framework and patterns it has observed in the team's dynamics, joining three sources for one answer. During onboarding Pascal learns the company's values, competencies and culture. If a company prioritizes radical candor, the guidance reflects that, and in a compliance heavy industry Pascal adjusts its approach. Pinnacle calls the result personalized, built on the customer's culture and values instead of generic AI prompts. Inputs include meeting observation across Zoom, Teams and Google Meet, Slack and Teams conversation patterns, organizational context from HRIS and temporal context such as review cycles and goal setting seasons. Pinnacle contrasts Pascal with coaching tools that are chatbots with a new label, where a manager asks a question, gets generic advice and closes the tab. It names no retrieval mechanism, embedding approach or citation surface, and it does not say how a coached manager could see which observation produced a piece of advice. Sourceheypinnacle.com/blog best-ai-coaching-tools knowledge graph and adaptation sections, heypinnacle.com homepage Perceptive and Personalized cardsread 2026-09-14 |
|
| Human Oversight & Guardrails | Partial |
|
When coaching touches mental health, harassment or legal issues, Pascal recognizes the topic and routes it to HR instead of giving AI guidance. On those topics the agent holds back and hands over, and on everything else it responds on its own. Pinnacle names no step where a person sees, approves or checks a draft before it reaches the manager, and no review and approve surface, queue or pending state. Administrators set the escalation triggers that decide when Pascal stops and hands over, and they customize privacy settings. Customizable safeguards keep Pascal focused on coaching while it escalates sensitive topics to the customer's team. Pinnacle runs regular AI bias audits on its own model in general. Conversations carry individual level confidentiality, with zero HR visibility into personal conversations. Sourceheypinnacle.com/blog best-ai-coaching-tools sensitive topic escalation, heypinnacle.com homepage Safe and Protective cardsread 2026-09-14 |
|
| Security, Identity & Governance | Full |
|
By Pinnacle's account, the product meets the highest global standards of data protection, naming SOC2 and GDPR, to help customers meet their compliance obligations. Its security protocols include SSO, MFA, encryption, custom data retention and regular penetration testing. Customizable safeguards and escalation keep coaching safe. Conversations are confidential, enterprises own their data, and models are never trained on user data. Pascal holds SOC2 Type II certification, with independently audited SOC2 compliance and annual third party reviews. Pinnacle says this shows verified security controls, data handling procedures and incident response capabilities. People sign in to the product through SSO and MFA. Administrators set escalation triggers and customize privacy settings, and conversations are isolated between customers. Pinnacle publishes no trust center, RBAC model or security page, and it does not name the auditor or the report date. Sourceheypinnacle.com homepage Private Secure Trusted block, heypinnacle.com/blog best-ai-coaching-tools security sectionread 2026-09-14 |
|
| Observability & Auditability | Not documented |
|
There is no published trace per run, tool call log, decision record, audit trail of coaching interactions, admin activity view or export path. Pinnacle does not say how a customer could see what Pascal said to whom, on what basis, or when it chose to intervene. Pinnacle tracks adoption, behavioral outcomes and business impact. It reports monthly active managers, engagement frequency, manager Net Promoter Score movement and hours saved, plus anonymous aggregated insights for organizational learning that do not expose individual conversations. These figures measure the customer's managers, whether they used the product and whether they improved, and none of them records what the agent did. Conversations carry individual level confidentiality, with zero HR visibility into personal conversations. A status page at pinnacle.statuspage.io reports service uptime. Sourceheypinnacle.com/blog best-ai-coaching-tools results and security sections, heypinnacle.com homepage impact figuresread 2026-09-14 |
|
| Memory & State Persistence | Partial |
|
Real team interactions feed Pascal's learning, which Pinnacle says makes the coaching deeper and richer in context. Pascal also keeps a knowledge graph of the customer's organization, mapping relationships and communication patterns over time, and it refers back to a manager's previous conversations with a person. Pascal is a chat interface, available 24/7 by text or voice in Slack, Teams and Zoom, with threaded exchanges. Custom data retention and configurable retention settings govern stored data in general. Pinnacle names no memory feature, no memory object for each user or team and no way to purge retained coaching context. Sourceheypinnacle.com homepage Perceptive and Plugged-in cards, Secure card retention controlread 2026-09-14 |
|
| Deployment & Data Residency | Partial |
|
Data residency controls are offered for regulated industries. No region is published, whether a list, a single country or a cloud region. Pinnacle names no customer environment option, such as VPC, private cloud or on premises, and no way to select one. Conversations are isolated between customers inside Pinnacle's own estate. Custom data retention sets how long data is kept, not where it is kept. Sourceheypinnacle.com/blog best-ai-coaching-tools security section, heypinnacle.com homepage Secure cardread 2026-09-14 |
|
| Prebuilt Agents / Templates / Packs | Not documented |
|
Twelve named use cases are split between managers and people leaders. For managers they are real time coaching during meetings, conflict and team dynamics, leadership skill building, better 1:1 conversations, difficult conversations and role plays. For people leaders they are manager development, culture transformation, performance reviews, learning at scale, extending HRBP capacity, and goal setting and alignment. Each is a topic the one agent, Pascal, can be asked about. None has its own page, URL, enable action, entitlement or price. There is no catalog, template library, gallery, prebuilt agent set or bundled toolkit. The structured reflection prompts, goal setting frameworks and skill development paths built by ICF certified coaches are the method inside Pascal, not assets a customer picks up. Sourceheypinnacle.com homepage Explore the use-cases grid, heypinnacle.com/blog best-ai-coaching-tools coaching frameworksread 2026-09-14 |
|
| Triggers & Channel Coverage | Full |
|
Events in the customer's working day start the agent, without anyone asking. Pinnacle calls Pascal proactive, showing up before the manager asks with real time support in the moment. The end of a meeting is one such event. With permission, Pascal attends 1:1s, team meetings and performance conversations. After a tense exchange with a direct report, it messages the manager that it noticed tension when discussing the project timeline and asks whether they want to debrief. Patterns over time are another. When Sarah's 1:1s with Tom run over by 20 minutes for the third consecutive week, Pascal messages her without waiting for her to ask for help. The rule watches weeks of history and acts when the condition is met. The calendar is a third. Guidance arrives before a difficult conversation, during a team conflict or right after a meeting, and Pascal tracks temporal context such as performance review cycles and goal setting seasons. Administrators set escalation triggers and customize privacy settings. Pinnacle publishes no webhook, schedule, cron or event subscription model. Sourceheypinnacle.com/blog best-ai-coaching-tools proactive intervention and meeting sections, heypinnacle.com homepage Proactive cardread 2026-09-14 |
|
| Model Flexibility & Routing | Not documented |
|
No provider, family or version is named for what runs Pascal. Pinnacle names no model selector, admin control, model setting for each agent or conversation, disclosed routing or way to bring your own key or model. ICF certified coaches trained Pascal's models, by Pinnacle's account, and built its structured reflection prompts, goal setting frameworks and skill development paths so the coaching goes beyond plain LLM responses. Pinnacle publishes seven comparison articles, one of them Pascal vs Claude, where a frontier model appears as a competitor. Models are never trained on user data. Sourceheypinnacle.com homepage, heypinnacle.com/blog best-ai-coaching-tools coaching frameworks section, site footer alternatives listread 2026-09-14 |
|
| APIs / SDKs / MCP Extensibility | Not documented |
|
No developer surface is published. The site has no developers or documentation entry, API reference or docs subdomain. It offers no REST API, SDK, MCP server, A2A agent card, outbound webhook or embeddable component, and Pinnacle makes no API claim. The product is sold to a CHRO and used by managers inside Slack. Of the two subdomains, my.heypinnacle.com is a customer login and pinnacle.statuspage.io is a service status page. Pascal embeds in Slack, Teams and Zoom, joins Google Meet sessions and reads HRIS and calendar context, each time reaching into someone else's system. Being installed in Slack gives it a presence in a channel, not an interface for driving the platform. Sourceheypinnacle.com header and footer navigation read across two pagesread 2026-09-14 |
|
| Testing, Debugging & Optimization | Not documented |
|
Regular AI bias audits cover Pinnacle's own model in general, not the customer's configuration, and Pinnacle does not say a customer can run or obtain one for its own instance. Pinnacle reports that 83% of direct reports see improvement in their manager's effectiveness after six months of Pascal usage. The figure comes from internal data on 1,247 direct reports across 12 customer organizations, measured through surveys before and after. It measures whether managers improved, not the agent's output. Pinnacle measures every deployment and publishes aggregated results quarterly from its internal customer data. There is no published evaluation harness, set of scored test cases, quality gate in a release path or optimization loop that measures the deployed agent's own decisions. Pinnacle does not say how a customer could test whether a change made Pascal better, sample its coaching for quality or compare configurations, and it publishes no debugging or activity surface. Sourceheypinnacle.com/blog best-ai-coaching-tools results section, heypinnacle.com homepage Safe card and impact figuresread 2026-09-14 |
|
| Browser / Computer-use | Not documented |
|
Pascal reaches its surfaces through platform integrations and posts messages back, with a manager at the keyboard throughout. Pinnacle names no hosted or local browser, desktop session or remote computer control that the agent drives itself. With permission, Pascal joins 1:1s, team meetings and performance conversations across Zoom, Teams and Google Meet as a participant, observes the dynamics and messages the manager afterward. It joins through the meeting platform's participant integration. Pinnacle describes no screen capture, cursor control or reading of pixels to target UI elements, and the product makes no purchases or submissions. Monitoring Slack and Teams conversations builds context for coaching. Sourceheypinnacle.com/blog best-ai-coaching-tools meeting attendance section, heypinnacle.com homepage Plugged-in cardread 2026-09-14 |
|
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
$150 to $300 per manager per year.
Per manager per year, in a published range of $150 to $300 per manager annually. Managers are the unit, not employees.
Included quota
Not published. Scoped by contract.
What is public
The rate and nothing else. There is no pricing page in the navigation, and the figure appears only in a comparison article, alongside a table setting annual cost per manager against traditional coaching and learning management systems. There are no tiers, no unit breakdown beyond per manager per year, no minimum, no term, no discount structure and no statement of what is included at each end of the range. Everything else routes to Book a demo.
Cost watchouts
Adding managers increases cost. Nothing published bills voice interactions or active use separately.
Variable cost rationale
Pinnacle charges per manager per year, with no usage meter. Nothing published describes consumption billing, overage, token charges or any way a bill grows within a term. The unit is a manager and the period is a year. The bill grows with the number of managers enrolled, so growth in the management layer raises cost, as with any per seat subscription. Nothing published says what places a customer within the $150 to $300 range.
Additional watchouts
The published figure is a range with a 2x spread and no tier table behind it. Nothing states whether volume, term, module scope or organization size moves a customer from $150 to $300 per manager, so $150 is the floor rather than a planning number. The unit is per manager rather than per employee, and Pinnacle's own fit guidance sets a practical floor. Fewer than 50 managers will not justify the setup effort, which puts the smallest sensible commitment at roughly $7,500 a year. The price sits in a blog article rather than a pricing page, so it carries no contractual weight and can change without any visible page change. Data residency controls are described as being for regulated industries, which reads as a scoped rather than universal entitlement, and no price is published for them.
Overage / add-ons
Scoped by contract and seat count.
Sales call required
Yes, required for paid access
Free / trial
None published
Lowest paid plan
$150 per manager per year, the bottom of the published range
Key ambiguities
A price is published, but not on a pricing page. "Pascal costs $150-300 per manager annually" appears three times in Pinnacle's own comparison article (in the body, in the key takeaways and in the page description), with a comparison table giving annual cost per manager as $150 to $300 against $15,000 for traditional coaching and $100 to $200 for an LMS. There is no pricing page in the navigation, and every call to action is Book a demo, with a customer log in at my.heypinnacle.com and no self serve purchase path, free tier or trial. The figure is a range with a 2x spread and no tier structure explaining what moves a customer from $150 to $300. Nothing published states whether the driver is volume, contract length, module scope or organization size. The unit is per manager, not per employee, which changes the arithmetic on a people development purchase. Pinnacle's own fit criteria say the product suits organizations of 200 to 4,000 employees and that fewer than 50 managers will not justify the setup effort. At 50 managers the floor is roughly $7,500 a year.
Related vendors
- 11th Estate — Agentic platform whose AI engine scans markets, matches an…
- Adopt AI — AI CPA firm whose agents run reconciliations, close, AP and AR, and…
- 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…
- Alloy.ai — Commerce intelligence system for consumer brands that unifies four…
Alternatives to Pinnacle
The closest documented capability profiles to Pinnacle 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.
- Casca5.0 / 14Adds documented Workflow Orchestration and Observability & Auditability, among others
- ConverzAI5.0 / 14Adds documented Workflow Orchestration and Observability & Auditability
- Friendly5.0 / 14Adds documented Workflow Orchestration and Observability & Auditability
- HROne7.0 / 14Adds documented Workflow Orchestration and Observability & Auditability, among others
- Parachute6.0 / 14Adds documented Workflow Orchestration and Observability & Auditability, among others
- Akro AI6.5 / 14Adds documented Workflow Orchestration and Observability & Auditability
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded